source: draft-ietf-httpbis/latest/p1-messaging.xml @ 2614

Last change on this file since 2614 was 2614, checked in by julian.reschke@…, 9 years ago

note change for [2613] (see #552)

  • Property svn:eol-style set to native
  • Property svn:mime-type set to text/xml
File size: 246.9 KB
Line 
1<?xml version="1.0" encoding="utf-8"?>
2<?xml-stylesheet type='text/xsl' href='../myxml2rfc.xslt'?>
3<!DOCTYPE rfc [
4  <!ENTITY MAY "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>MAY</bcp14>">
5  <!ENTITY MUST "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>MUST</bcp14>">
6  <!ENTITY MUST-NOT "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>MUST NOT</bcp14>">
7  <!ENTITY OPTIONAL "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>OPTIONAL</bcp14>">
8  <!ENTITY RECOMMENDED "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>RECOMMENDED</bcp14>">
9  <!ENTITY REQUIRED "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>REQUIRED</bcp14>">
10  <!ENTITY SHALL "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>SHALL</bcp14>">
11  <!ENTITY SHALL-NOT "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>SHALL NOT</bcp14>">
12  <!ENTITY SHOULD "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>SHOULD</bcp14>">
13  <!ENTITY SHOULD-NOT "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>SHOULD NOT</bcp14>">
14  <!ENTITY ID-VERSION "latest">
15  <!ENTITY ID-MONTH "February">
16  <!ENTITY ID-YEAR "2014">
17  <!ENTITY mdash "&#8212;">
18  <!ENTITY Note "<x:h xmlns:x='http://purl.org/net/xml2rfc/ext'>Note:</x:h>">
19  <!ENTITY caching-overview       "<xref target='Part6' x:rel='#caching.overview' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
20  <!ENTITY cache-incomplete       "<xref target='Part6' x:rel='#response.cacheability' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
21  <!ENTITY cache-poisoning        "<xref target='Part6' x:rel='#security.considerations' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
22  <!ENTITY payload                "<xref target='Part2' x:rel='#payload' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
23  <!ENTITY media-type             "<xref target='Part2' x:rel='#media.type' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
24  <!ENTITY content-codings        "<xref target='Part2' x:rel='#content.codings' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
25  <!ENTITY CONNECT                "<xref target='Part2' x:rel='#CONNECT' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
26  <!ENTITY content.negotiation    "<xref target='Part2' x:rel='#content.negotiation' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
27  <!ENTITY diff-mime              "<xref target='Part2' x:rel='#differences.between.http.and.mime' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
28  <!ENTITY representation         "<xref target='Part2' x:rel='#representations' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
29  <!ENTITY GET                    "<xref target='Part2' x:rel='#GET' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
30  <!ENTITY HEAD                   "<xref target='Part2' x:rel='#HEAD' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
31  <!ENTITY header-allow           "<xref target='Part2' x:rel='#header.allow' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
32  <!ENTITY header-cache-control   "<xref target='Part6' x:rel='#header.cache-control' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
33  <!ENTITY header-content-encoding    "<xref target='Part2' x:rel='#header.content-encoding' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
34  <!ENTITY header-content-location    "<xref target='Part2' x:rel='#header.content-location' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
35  <!ENTITY header-content-range   "<xref target='Part5' x:rel='#header.content-range' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
36  <!ENTITY header-content-type    "<xref target='Part2' x:rel='#header.content-type' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
37  <!ENTITY header-date            "<xref target='Part2' x:rel='#header.date' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
38  <!ENTITY header-etag            "<xref target='Part4' x:rel='#header.etag' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
39  <!ENTITY header-expect          "<xref target='Part2' x:rel='#header.expect' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
40  <!ENTITY header-expires         "<xref target='Part6' x:rel='#header.expires' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
41  <!ENTITY header-last-modified   "<xref target='Part4' x:rel='#header.last-modified' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
42  <!ENTITY header-mime-version    "<xref target='Part2' x:rel='#mime-version' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
43  <!ENTITY header-pragma          "<xref target='Part6' x:rel='#header.pragma' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
44  <!ENTITY header-proxy-authenticate  "<xref target='Part7' x:rel='#header.proxy-authenticate' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
45  <!ENTITY header-proxy-authorization "<xref target='Part7' x:rel='#header.proxy-authorization' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
46  <!ENTITY header-server          "<xref target='Part2' x:rel='#header.server' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
47  <!ENTITY header-warning         "<xref target='Part6' x:rel='#header.warning' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
48  <!ENTITY idempotent-methods     "<xref target='Part2' x:rel='#idempotent.methods' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
49  <!ENTITY safe-methods           "<xref target='Part2' x:rel='#safe.methods' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
50  <!ENTITY methods                "<xref target='Part2' x:rel='#methods' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
51  <!ENTITY OPTIONS                "<xref target='Part2' x:rel='#OPTIONS' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
52  <!ENTITY qvalue                 "<xref target='Part2' x:rel='#quality.values' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
53  <!ENTITY request-header-fields  "<xref target='Part2' x:rel='#request.header.fields' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
54  <!ENTITY response-control-data  "<xref target='Part2' x:rel='#response.control.data' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
55  <!ENTITY resource               "<xref target='Part2' x:rel='#resources' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
56  <!ENTITY semantics              "<xref target='Part2' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
57  <!ENTITY status-codes           "<xref target='Part2' x:rel='#status.codes' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
58  <!ENTITY status-1xx             "<xref target='Part2' x:rel='#status.1xx' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
59  <!ENTITY status-203             "<xref target='Part2' x:rel='#status.203' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
60  <!ENTITY status-3xx             "<xref target='Part2' x:rel='#status.3xx' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
61  <!ENTITY status-304             "<xref target='Part4' x:rel='#status.304' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
62  <!ENTITY status-4xx             "<xref target='Part2' x:rel='#status.4xx' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
63  <!ENTITY status-414             "<xref target='Part2' x:rel='#status.414' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
64  <!ENTITY iana-header-registry   "<xref target='Part2' x:rel='#header.field.registry' xmlns:x='http://purl.org/net/xml2rfc/ext'/>">
65]>
66<?rfc toc="yes" ?>
67<?rfc symrefs="yes" ?>
68<?rfc sortrefs="yes" ?>
69<?rfc compact="yes"?>
70<?rfc subcompact="no" ?>
71<?rfc linkmailto="no" ?>
72<?rfc editing="no" ?>
73<?rfc comments="yes"?>
74<?rfc inline="yes"?>
75<?rfc rfcedstyle="yes"?>
76<?rfc-ext allow-markup-in-artwork="yes" ?>
77<?rfc-ext include-references-in-index="yes" ?>
78<rfc obsoletes="2145,2616" updates="2817,2818" category="std" x:maturity-level="proposed"
79     ipr="pre5378Trust200902" docName="draft-ietf-httpbis-p1-messaging-&ID-VERSION;"
80     xmlns:x='http://purl.org/net/xml2rfc/ext'>
81<x:link rel="next" basename="p2-semantics"/>
82<x:feedback template="mailto:ietf-http-wg@w3.org?subject={docname},%20%22{section}%22&amp;body=&lt;{ref}&gt;:"/>
83<front>
84
85  <title abbrev="HTTP/1.1 Message Syntax and Routing">Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing</title>
86
87  <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
88    <organization abbrev="Adobe">Adobe Systems Incorporated</organization>
89    <address>
90      <postal>
91        <street>345 Park Ave</street>
92        <city>San Jose</city>
93        <region>CA</region>
94        <code>95110</code>
95        <country>USA</country>
96      </postal>
97      <email>fielding@gbiv.com</email>
98      <uri>http://roy.gbiv.com/</uri>
99    </address>
100  </author>
101
102  <author initials="J. F." surname="Reschke" fullname="Julian F. Reschke" role="editor">
103    <organization abbrev="greenbytes">greenbytes GmbH</organization>
104    <address>
105      <postal>
106        <street>Hafenweg 16</street>
107        <city>Muenster</city><region>NW</region><code>48155</code>
108        <country>Germany</country>
109      </postal>
110      <email>julian.reschke@greenbytes.de</email>
111      <uri>http://greenbytes.de/tech/webdav/</uri>
112    </address>
113  </author>
114
115  <date month="&ID-MONTH;" year="&ID-YEAR;"/>
116
117  <area>Applications</area>
118  <workgroup>HTTPbis</workgroup>
119
120<abstract>
121<t>
122   The Hypertext Transfer Protocol (HTTP) is a stateless application-level
123   protocol for distributed, collaborative, hypertext information systems.
124   This document provides an overview of HTTP architecture and its associated
125   terminology, defines the "http" and "https" Uniform Resource Identifier
126   (URI) schemes, defines the HTTP/1.1 message syntax and parsing
127   requirements, and describes related security concerns for implementations.
128</t>   
129</abstract>
130
131<note title="Editorial Note (To be removed by RFC Editor)">
132  <t>
133    Discussion of this draft takes place on the HTTPBIS working group
134    mailing list (ietf-http-wg@w3.org), which is archived at
135    <eref target="http://lists.w3.org/Archives/Public/ietf-http-wg/"/>.
136  </t>
137  <t>
138    The current issues list is at
139    <eref target="http://tools.ietf.org/wg/httpbis/trac/report/3"/> and related
140    documents (including fancy diffs) can be found at
141    <eref target="http://tools.ietf.org/wg/httpbis/"/>.
142  </t>
143  <t>
144    The changes in this draft are summarized in <xref target="changes.since.25"/>.
145  </t>
146</note>
147</front>
148<middle>
149<section title="Introduction" anchor="introduction">
150<t>
151   The Hypertext Transfer Protocol (HTTP) is a stateless application-level
152   request/response protocol that uses extensible semantics and
153   self-descriptive message payloads for flexible interaction with
154   network-based hypertext information systems. This document is the first in
155   a series of documents that collectively form the HTTP/1.1 specification:
156   <list style="empty">
157    <t>RFC xxx1: Message Syntax and Routing</t>
158    <t><xref target="Part2" x:fmt="none">RFC xxx2</xref>: Semantics and Content</t>
159    <t><xref target="Part4" x:fmt="none">RFC xxx3</xref>: Conditional Requests</t>
160    <t><xref target="Part5" x:fmt="none">RFC xxx4</xref>: Range Requests</t>
161    <t><xref target="Part6" x:fmt="none">RFC xxx5</xref>: Caching</t>
162    <t><xref target="Part7" x:fmt="none">RFC xxx6</xref>: Authentication</t>
163   </list>
164</t>
165<t>
166   This HTTP/1.1 specification obsoletes
167   <xref target="RFC2616" x:fmt="none">RFC 2616</xref> and
168   <xref target="RFC2145" x:fmt="none">RFC 2145</xref> (on HTTP versioning).
169   This specification also updates the use of CONNECT to establish a tunnel,
170   previously defined in <xref target="RFC2817" x:fmt="none">RFC 2817</xref>,
171   and defines the "https" URI scheme that was described informally in
172   <xref target="RFC2818" x:fmt="none">RFC 2818</xref>.
173</t>
174<t>
175   HTTP is a generic interface protocol for information systems. It is
176   designed to hide the details of how a service is implemented by presenting
177   a uniform interface to clients that is independent of the types of
178   resources provided. Likewise, servers do not need to be aware of each
179   client's purpose: an HTTP request can be considered in isolation rather
180   than being associated with a specific type of client or a predetermined
181   sequence of application steps. The result is a protocol that can be used
182   effectively in many different contexts and for which implementations can
183   evolve independently over time.
184</t>
185<t>
186   HTTP is also designed for use as an intermediation protocol for translating
187   communication to and from non-HTTP information systems.
188   HTTP proxies and gateways can provide access to alternative information
189   services by translating their diverse protocols into a hypertext
190   format that can be viewed and manipulated by clients in the same way
191   as HTTP services.
192</t>
193<t>
194   One consequence of this flexibility is that the protocol cannot be
195   defined in terms of what occurs behind the interface. Instead, we
196   are limited to defining the syntax of communication, the intent
197   of received communication, and the expected behavior of recipients.
198   If the communication is considered in isolation, then successful
199   actions ought to be reflected in corresponding changes to the
200   observable interface provided by servers. However, since multiple
201   clients might act in parallel and perhaps at cross-purposes, we
202   cannot require that such changes be observable beyond the scope
203   of a single response.
204</t>
205<t>
206   This document describes the architectural elements that are used or
207   referred to in HTTP, defines the "http" and "https" URI schemes,
208   describes overall network operation and connection management,
209   and defines HTTP message framing and forwarding requirements.
210   Our goal is to define all of the mechanisms necessary for HTTP message
211   handling that are independent of message semantics, thereby defining the
212   complete set of requirements for message parsers and
213   message-forwarding intermediaries.
214</t>
215
216
217<section title="Requirement Notation" anchor="intro.requirements">
218<t>
219   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
220   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
221   document are to be interpreted as described in <xref target="RFC2119"/>.
222</t>
223<t>
224   Conformance criteria and considerations regarding error handling
225   are defined in <xref target="conformance"/>.
226</t>
227</section>
228
229<section title="Syntax Notation" anchor="notation">
230<iref primary="true" item="Grammar" subitem="ALPHA"/>
231<iref primary="true" item="Grammar" subitem="CR"/>
232<iref primary="true" item="Grammar" subitem="CRLF"/>
233<iref primary="true" item="Grammar" subitem="CTL"/>
234<iref primary="true" item="Grammar" subitem="DIGIT"/>
235<iref primary="true" item="Grammar" subitem="DQUOTE"/>
236<iref primary="true" item="Grammar" subitem="HEXDIG"/>
237<iref primary="true" item="Grammar" subitem="HTAB"/>
238<iref primary="true" item="Grammar" subitem="LF"/>
239<iref primary="true" item="Grammar" subitem="OCTET"/>
240<iref primary="true" item="Grammar" subitem="SP"/>
241<iref primary="true" item="Grammar" subitem="VCHAR"/>
242<t>
243   This specification uses the Augmented Backus-Naur Form (ABNF) notation of
244   <xref target="RFC5234"/> with a list extension, defined in
245   <xref target="abnf.extension"/>, that allows for compact definition of
246   comma-separated lists using a '#' operator (similar to how the '*' operator
247   indicates repetition).
248   <xref target="collected.abnf"/> shows the collected grammar with all list
249   operators expanded to standard ABNF notation.
250</t>
251<t anchor="core.rules">
252  <x:anchor-alias value="ALPHA"/>
253  <x:anchor-alias value="CTL"/>
254  <x:anchor-alias value="CR"/>
255  <x:anchor-alias value="CRLF"/>
256  <x:anchor-alias value="DIGIT"/>
257  <x:anchor-alias value="DQUOTE"/>
258  <x:anchor-alias value="HEXDIG"/>
259  <x:anchor-alias value="HTAB"/>
260  <x:anchor-alias value="LF"/>
261  <x:anchor-alias value="OCTET"/>
262  <x:anchor-alias value="SP"/>
263  <x:anchor-alias value="VCHAR"/>
264   The following core rules are included by
265   reference, as defined in <xref target="RFC5234" x:fmt="," x:sec="B.1"/>:
266   ALPHA (letters), CR (carriage return), CRLF (CR LF), CTL (controls),
267   DIGIT (decimal 0-9), DQUOTE (double quote),
268   HEXDIG (hexadecimal 0-9/A-F/a-f), HTAB (horizontal tab), LF (line feed),
269   OCTET (any 8-bit sequence of data), SP (space), and
270   VCHAR (any visible <xref target="USASCII"/> character).
271</t>
272<t>
273   As a convention, ABNF rule names prefixed with "obs-" denote
274   "obsolete" grammar rules that appear for historical reasons.
275</t>
276</section>
277</section>
278
279<section title="Architecture" anchor="architecture">
280<t>
281   HTTP was created for the World Wide Web (WWW) architecture
282   and has evolved over time to support the scalability needs of a worldwide
283   hypertext system. Much of that architecture is reflected in the terminology
284   and syntax productions used to define HTTP.
285</t>
286
287<section title="Client/Server Messaging" anchor="operation">
288<iref primary="true" item="client"/>
289<iref primary="true" item="server"/>
290<iref primary="true" item="connection"/>
291<t>
292   HTTP is a stateless request/response protocol that operates by exchanging
293   <x:dfn>messages</x:dfn> (<xref target="http.message"/>) across a reliable
294   transport or session-layer
295   "<x:dfn>connection</x:dfn>" (<xref target="connection.management"/>).
296   An HTTP "<x:dfn>client</x:dfn>" is a program that establishes a connection
297   to a server for the purpose of sending one or more HTTP requests.
298   An HTTP "<x:dfn>server</x:dfn>" is a program that accepts connections
299   in order to service HTTP requests by sending HTTP responses.
300</t>
301<iref primary="true" item="user agent"/>
302<iref primary="true" item="origin server"/>
303<iref primary="true" item="browser"/>
304<iref primary="true" item="spider"/>
305<iref primary="true" item="sender"/>
306<iref primary="true" item="recipient"/>
307<t>
308   The terms client and server refer only to the roles that
309   these programs perform for a particular connection.  The same program
310   might act as a client on some connections and a server on others.
311   The term "<x:dfn>user agent</x:dfn>" refers to any of the various
312   client programs that initiate a request, including (but not limited to)
313   browsers, spiders (web-based robots), command-line tools, custom
314   applications, and mobile apps.
315   The term "<x:dfn>origin server</x:dfn>" refers to the program that can
316   originate authoritative responses for a given target resource.
317   The terms "<x:dfn>sender</x:dfn>" and "<x:dfn>recipient</x:dfn>" refer to
318   any implementation that sends or receives a given message, respectively.
319</t>
320<t>
321   HTTP relies upon the Uniform Resource Identifier (URI)
322   standard <xref target="RFC3986"/> to indicate the target resource
323   (<xref target="target-resource"/>) and relationships between resources.
324   Messages are passed in a format similar to that used by Internet mail
325   <xref target="RFC5322"/> and the Multipurpose Internet Mail Extensions
326   (MIME) <xref target="RFC2045"/> (see &diff-mime; for the differences
327   between HTTP and MIME messages).
328</t>
329<t>
330   Most HTTP communication consists of a retrieval request (GET) for
331   a representation of some resource identified by a URI.  In the
332   simplest case, this might be accomplished via a single bidirectional
333   connection (===) between the user agent (UA) and the origin server (O).
334</t>
335<figure><artwork type="drawing">
336         request   &gt;
337    <x:highlight>UA</x:highlight> ======================================= <x:highlight>O</x:highlight>
338                                &lt;   response
339</artwork></figure>
340<iref primary="true" item="message"/>
341<iref primary="true" item="request"/>
342<iref primary="true" item="response"/>
343<t>
344   A client sends an HTTP request to a server in the form of a <x:dfn>request</x:dfn>
345   message, beginning with a request-line that includes a method, URI, and
346   protocol version (<xref target="request.line"/>),
347   followed by header fields containing
348   request modifiers, client information, and representation metadata
349   (<xref target="header.fields"/>),
350   an empty line to indicate the end of the header section, and finally
351   a message body containing the payload body (if any,
352   <xref target="message.body"/>).
353</t>
354<t>
355   A server responds to a client's request by sending one or more HTTP
356   <x:dfn>response</x:dfn>
357   messages, each beginning with a status line that
358   includes the protocol version, a success or error code, and textual
359   reason phrase (<xref target="status.line"/>),
360   possibly followed by header fields containing server
361   information, resource metadata, and representation metadata
362   (<xref target="header.fields"/>),
363   an empty line to indicate the end of the header section, and finally
364   a message body containing the payload body (if any,
365   <xref target="message.body"/>).
366</t>
367<t>
368   A connection might be used for multiple request/response exchanges,
369   as defined in <xref target="persistent.connections"/>.
370</t>
371<t>
372   The following example illustrates a typical message exchange for a
373   GET request (&GET;) on the URI "http://www.example.com/hello.txt":
374</t>
375<figure><preamble>
376Client request:
377</preamble><artwork type="message/http; msgtype=&#34;request&#34;" x:indent-with="  ">
378GET /hello.txt HTTP/1.1
379User-Agent: curl/7.16.3 libcurl/7.16.3 OpenSSL/0.9.7l zlib/1.2.3
380Host: www.example.com
381Accept-Language: en, mi
382
383</artwork></figure>
384<figure><preamble>
385Server response:
386</preamble><artwork type="message/http; msgtype=&#34;response&#34;" x:indent-with="  ">
387HTTP/1.1 200 OK
388Date: Mon, 27 Jul 2009 12:28:53 GMT
389Server: Apache
390Last-Modified: Wed, 22 Jul 2009 19:15:56 GMT
391ETag: "34aa387-d-1568eb00"
392Accept-Ranges: bytes
393Content-Length: <x:length-of target="exbody"/>
394Vary: Accept-Encoding
395Content-Type: text/plain
396
397<x:span anchor="exbody">Hello World! My payload includes a trailing CRLF.
398</x:span></artwork>
399</figure>
400</section>
401
402<section title="Implementation Diversity" anchor="implementation-diversity">
403<t>
404   When considering the design of HTTP, it is easy to fall into a trap of
405   thinking that all user agents are general-purpose browsers and all origin
406   servers are large public websites. That is not the case in practice.
407   Common HTTP user agents include household appliances, stereos, scales,
408   firmware update scripts, command-line programs, mobile apps,
409   and communication devices in a multitude of shapes and sizes.  Likewise,
410   common HTTP origin servers include home automation units, configurable
411   networking components, office machines, autonomous robots, news feeds,
412   traffic cameras, ad selectors, and video delivery platforms.
413</t>
414<t>
415   The term "user agent" does not imply that there is a human user directly
416   interacting with the software agent at the time of a request. In many
417   cases, a user agent is installed or configured to run in the background
418   and save its results for later inspection (or save only a subset of those
419   results that might be interesting or erroneous). Spiders, for example, are
420   typically given a start URI and configured to follow certain behavior while
421   crawling the Web as a hypertext graph.
422</t>
423<t>
424   The implementation diversity of HTTP means that not all user agents can
425   make interactive suggestions to their user or provide adequate warning for
426   security or privacy concerns. In the few cases where this
427   specification requires reporting of errors to the user, it is acceptable
428   for such reporting to only be observable in an error console or log file.
429   Likewise, requirements that an automated action be confirmed by the user
430   before proceeding might be met via advance configuration choices,
431   run-time options, or simple avoidance of the unsafe action; confirmation
432   does not imply any specific user interface or interruption of normal
433   processing if the user has already made that choice.
434</t>
435</section>
436
437<section title="Intermediaries" anchor="intermediaries">
438<iref primary="true" item="intermediary"/>
439<t>
440   HTTP enables the use of intermediaries to satisfy requests through
441   a chain of connections.  There are three common forms of HTTP
442   <x:dfn>intermediary</x:dfn>: proxy, gateway, and tunnel.  In some cases,
443   a single intermediary might act as an origin server, proxy, gateway,
444   or tunnel, switching behavior based on the nature of each request.
445</t>
446<figure><artwork type="drawing">
447         &gt;             &gt;             &gt;             &gt;
448    <x:highlight>UA</x:highlight> =========== <x:highlight>A</x:highlight> =========== <x:highlight>B</x:highlight> =========== <x:highlight>C</x:highlight> =========== <x:highlight>O</x:highlight>
449               &lt;             &lt;             &lt;             &lt;
450</artwork></figure>
451<t>
452   The figure above shows three intermediaries (A, B, and C) between the
453   user agent and origin server. A request or response message that
454   travels the whole chain will pass through four separate connections.
455   Some HTTP communication options
456   might apply only to the connection with the nearest, non-tunnel
457   neighbor, only to the end-points of the chain, or to all connections
458   along the chain. Although the diagram is linear, each participant might
459   be engaged in multiple, simultaneous communications. For example, B
460   might be receiving requests from many clients other than A, and/or
461   forwarding requests to servers other than C, at the same time that it
462   is handling A's request. Likewise, later requests might be sent through a
463   different path of connections, often based on dynamic configuration for
464   load balancing.   
465</t>
466<t>
467<iref primary="true" item="upstream"/><iref primary="true" item="downstream"/>
468<iref primary="true" item="inbound"/><iref primary="true" item="outbound"/>
469   The terms "<x:dfn>upstream</x:dfn>" and "<x:dfn>downstream</x:dfn>" are
470   used to describe directional requirements in relation to the message flow:
471   all messages flow from upstream to downstream.
472   The terms inbound and outbound are used to describe directional
473   requirements in relation to the request route:
474   "<x:dfn>inbound</x:dfn>" means toward the origin server and
475   "<x:dfn>outbound</x:dfn>" means toward the user agent.
476</t>
477<t><iref primary="true" item="proxy"/>
478   A "<x:dfn>proxy</x:dfn>" is a message forwarding agent that is selected by the
479   client, usually via local configuration rules, to receive requests
480   for some type(s) of absolute URI and attempt to satisfy those
481   requests via translation through the HTTP interface.  Some translations
482   are minimal, such as for proxy requests for "http" URIs, whereas
483   other requests might require translation to and from entirely different
484   application-level protocols. Proxies are often used to group an
485   organization's HTTP requests through a common intermediary for the
486   sake of security, annotation services, or shared caching. Some proxies
487   are designed to apply transformations to selected messages or payloads
488   while they are being forwarded, as described in
489   <xref target="message.transformations"/>.
490</t>
491<t><iref primary="true" item="gateway"/><iref primary="true" item="reverse proxy"/>
492<iref primary="true" item="accelerator"/>
493   A "<x:dfn>gateway</x:dfn>" (a.k.a., "<x:dfn>reverse proxy</x:dfn>") is an
494   intermediary that acts as an origin server for the outbound connection, but
495   translates received requests and forwards them inbound to another server or
496   servers. Gateways are often used to encapsulate legacy or untrusted
497   information services, to improve server performance through
498   "<x:dfn>accelerator</x:dfn>" caching, and to enable partitioning or load
499   balancing of HTTP services across multiple machines.
500</t>
501<t>
502   All HTTP requirements applicable to an origin server
503   also apply to the outbound communication of a gateway.
504   A gateway communicates with inbound servers using any protocol that
505   it desires, including private extensions to HTTP that are outside
506   the scope of this specification.  However, an HTTP-to-HTTP gateway
507   that wishes to interoperate with third-party HTTP servers ought to conform
508   to user agent requirements on the gateway's inbound connection.
509</t>
510<t><iref primary="true" item="tunnel"/>
511   A "<x:dfn>tunnel</x:dfn>" acts as a blind relay between two connections
512   without changing the messages. Once active, a tunnel is not
513   considered a party to the HTTP communication, though the tunnel might
514   have been initiated by an HTTP request. A tunnel ceases to exist when
515   both ends of the relayed connection are closed. Tunnels are used to
516   extend a virtual connection through an intermediary, such as when
517   Transport Layer Security (TLS, <xref target="RFC5246"/>) is used to
518   establish confidential communication through a shared firewall proxy.
519</t>
520<t>
521   The above categories for intermediary only consider those acting as
522   participants in the HTTP communication.  There are also intermediaries
523   that can act on lower layers of the network protocol stack, filtering or
524   redirecting HTTP traffic without the knowledge or permission of message
525   senders. Network intermediaries are indistinguishable (at a protocol level)
526   from a man-in-the-middle attack, often introducing security flaws or
527   interoperability problems due to mistakenly violating HTTP semantics.
528</t>
529<t><iref primary="true" item="interception proxy"/>
530<iref primary="true" item="transparent proxy"/>
531<iref primary="true" item="captive portal"/>
532   For example, an
533   "<x:dfn>interception proxy</x:dfn>" <xref target="RFC3040"/> (also commonly
534   known as a "<x:dfn>transparent proxy</x:dfn>" <xref target="RFC1919"/> or
535   "<x:dfn>captive portal</x:dfn>")
536   differs from an HTTP proxy because it is not selected by the client.
537   Instead, an interception proxy filters or redirects outgoing TCP port 80
538   packets (and occasionally other common port traffic).
539   Interception proxies are commonly found on public network access points,
540   as a means of enforcing account subscription prior to allowing use of
541   non-local Internet services, and within corporate firewalls to enforce
542   network usage policies.
543</t>
544<t>
545   HTTP is defined as a stateless protocol, meaning that each request message
546   can be understood in isolation.  Many implementations depend on HTTP's
547   stateless design in order to reuse proxied connections or dynamically
548   load-balance requests across multiple servers.  Hence, a server &MUST-NOT;
549   assume that two requests on the same connection are from the same user
550   agent unless the connection is secured and specific to that agent.
551   Some non-standard HTTP extensions (e.g., <xref target="RFC4559"/>) have
552   been known to violate this requirement, resulting in security and
553   interoperability problems.
554</t>
555</section>
556
557<section title="Caches" anchor="caches">
558<iref primary="true" item="cache"/>
559<t>
560   A "<x:dfn>cache</x:dfn>" is a local store of previous response messages and the
561   subsystem that controls its message storage, retrieval, and deletion.
562   A cache stores cacheable responses in order to reduce the response
563   time and network bandwidth consumption on future, equivalent
564   requests. Any client or server &MAY; employ a cache, though a cache
565   cannot be used by a server while it is acting as a tunnel.
566</t>
567<t>
568   The effect of a cache is that the request/response chain is shortened
569   if one of the participants along the chain has a cached response
570   applicable to that request. The following illustrates the resulting
571   chain if B has a cached copy of an earlier response from O (via C)
572   for a request that has not been cached by UA or A.
573</t>
574<figure><artwork type="drawing">
575            &gt;             &gt;
576       <x:highlight>UA</x:highlight> =========== <x:highlight>A</x:highlight> =========== <x:highlight>B</x:highlight> - - - - - - <x:highlight>C</x:highlight> - - - - - - <x:highlight>O</x:highlight>
577                  &lt;             &lt;
578</artwork></figure>
579<t><iref primary="true" item="cacheable"/>
580   A response is "<x:dfn>cacheable</x:dfn>" if a cache is allowed to store a copy of
581   the response message for use in answering subsequent requests.
582   Even when a response is cacheable, there might be additional
583   constraints placed by the client or by the origin server on when
584   that cached response can be used for a particular request. HTTP
585   requirements for cache behavior and cacheable responses are
586   defined in &caching-overview;. 
587</t>
588<t>
589   There are a wide variety of architectures and configurations
590   of caches deployed across the World Wide Web and
591   inside large organizations. These include national hierarchies
592   of proxy caches to save transoceanic bandwidth, collaborative systems that
593   broadcast or multicast cache entries, archives of pre-fetched cache
594   entries for use in off-line or high-latency environments, and so on.
595</t>
596</section>
597
598<section title="Conformance and Error Handling" anchor="conformance">
599<t>
600   This specification targets conformance criteria according to the role of
601   a participant in HTTP communication.  Hence, HTTP requirements are placed
602   on senders, recipients, clients, servers, user agents, intermediaries,
603   origin servers, proxies, gateways, or caches, depending on what behavior
604   is being constrained by the requirement. Additional (social) requirements
605   are placed on implementations, resource owners, and protocol element
606   registrations when they apply beyond the scope of a single communication.
607</t>
608<t>
609   The verb "generate" is used instead of "send" where a requirement
610   differentiates between creating a protocol element and merely forwarding a
611   received element downstream.
612</t>
613<t>
614   An implementation is considered conformant if it complies with all of the
615   requirements associated with the roles it partakes in HTTP.
616</t>
617<t>
618   Conformance includes both the syntax and semantics of protocol
619   elements. A sender &MUST-NOT; generate protocol elements that convey a
620   meaning that is known by that sender to be false. A sender &MUST-NOT;
621   generate protocol elements that do not match the grammar defined by the
622   corresponding ABNF rules. Within a given message, a sender &MUST-NOT;
623   generate protocol elements or syntax alternatives that are only allowed to
624   be generated by participants in other roles (i.e., a role that the sender
625   does not have for that message).
626</t>
627<t>
628   When a received protocol element is parsed, the recipient &MUST; be able to
629   parse any value of reasonable length that is applicable to the recipient's
630   role and matches the grammar defined by the corresponding ABNF rules.
631   Note, however, that some received protocol elements might not be parsed.
632   For example, an intermediary forwarding a message might parse a
633   header-field into generic field-name and field-value components, but then
634   forward the header field without further parsing inside the field-value.
635</t>
636<t>
637   HTTP does not have specific length limitations for many of its protocol
638   elements because the lengths that might be appropriate will vary widely,
639   depending on the deployment context and purpose of the implementation.
640   Hence, interoperability between senders and recipients depends on shared
641   expectations regarding what is a reasonable length for each protocol
642   element. Furthermore, what is commonly understood to be a reasonable length
643   for some protocol elements has changed over the course of the past two
644   decades of HTTP use, and is expected to continue changing in the future.
645</t>
646<t>
647   At a minimum, a recipient &MUST; be able to parse and process protocol
648   element lengths that are at least as long as the values that it generates
649   for those same protocol elements in other messages. For example, an origin
650   server that publishes very long URI references to its own resources needs
651   to be able to parse and process those same references when received as a
652   request target.
653</t>
654<t>
655   A recipient &MUST; interpret a received protocol element according to the
656   semantics defined for it by this specification, including extensions to
657   this specification, unless the recipient has determined (through experience
658   or configuration) that the sender incorrectly implements what is implied by
659   those semantics.
660   For example, an origin server might disregard the contents of a received
661   <x:ref>Accept-Encoding</x:ref> header field if inspection of the
662   <x:ref>User-Agent</x:ref> header field indicates a specific implementation
663   version that is known to fail on receipt of certain content codings.
664</t>
665<t>
666   Unless noted otherwise, a recipient &MAY; attempt to recover a usable
667   protocol element from an invalid construct.  HTTP does not define
668   specific error handling mechanisms except when they have a direct impact
669   on security, since different applications of the protocol require
670   different error handling strategies.  For example, a Web browser might
671   wish to transparently recover from a response where the
672   <x:ref>Location</x:ref> header field doesn't parse according to the ABNF,
673   whereas a systems control client might consider any form of error recovery
674   to be dangerous.
675</t>
676</section>
677
678<section title="Protocol Versioning" anchor="http.version">
679  <x:anchor-alias value="HTTP-version"/>
680  <x:anchor-alias value="HTTP-name"/>
681<t>
682   HTTP uses a "&lt;major&gt;.&lt;minor&gt;" numbering scheme to indicate
683   versions of the protocol. This specification defines version "1.1".
684   The protocol version as a whole indicates the sender's conformance
685   with the set of requirements laid out in that version's corresponding
686   specification of HTTP.
687</t>
688<t>
689   The version of an HTTP message is indicated by an HTTP-version field
690   in the first line of the message. HTTP-version is case-sensitive.
691</t>
692<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="HTTP-version"/><iref primary="true" item="Grammar" subitem="HTTP-name"/>
693  <x:ref>HTTP-version</x:ref>  = <x:ref>HTTP-name</x:ref> "/" <x:ref>DIGIT</x:ref> "." <x:ref>DIGIT</x:ref>
694  <x:ref>HTTP-name</x:ref>     = <x:abnf-char-sequence>"HTTP"</x:abnf-char-sequence> ; "HTTP", case-sensitive
695</artwork></figure>
696<t>
697   The HTTP version number consists of two decimal digits separated by a "."
698   (period or decimal point).  The first digit ("major version") indicates the
699   HTTP messaging syntax, whereas the second digit ("minor version") indicates
700   the highest minor version within that major version to which the sender is
701   conformant and able to understand for future communication.  The minor
702   version advertises the sender's communication capabilities even when the
703   sender is only using a backwards-compatible subset of the protocol,
704   thereby letting the recipient know that more advanced features can
705   be used in response (by servers) or in future requests (by clients).
706</t>
707<t>
708   When an HTTP/1.1 message is sent to an HTTP/1.0 recipient
709   <xref target="RFC1945"/> or a recipient whose version is unknown,
710   the HTTP/1.1 message is constructed such that it can be interpreted
711   as a valid HTTP/1.0 message if all of the newer features are ignored.
712   This specification places recipient-version requirements on some
713   new features so that a conformant sender will only use compatible
714   features until it has determined, through configuration or the
715   receipt of a message, that the recipient supports HTTP/1.1.
716</t>
717<t>
718   The interpretation of a header field does not change between minor
719   versions of the same major HTTP version, though the default
720   behavior of a recipient in the absence of such a field can change.
721   Unless specified otherwise, header fields defined in HTTP/1.1 are
722   defined for all versions of HTTP/1.x.  In particular, the <x:ref>Host</x:ref>
723   and <x:ref>Connection</x:ref> header fields ought to be implemented by all
724   HTTP/1.x implementations whether or not they advertise conformance with
725   HTTP/1.1.
726</t>
727<t>
728   New header fields can be introduced without changing the protocol version
729   if their defined semantics allow them to be safely ignored by recipients
730   that do not recognize them. Header field extensibility is discussed in
731   <xref target="field.extensibility"/>.
732</t>
733<t>
734   Intermediaries that process HTTP messages (i.e., all intermediaries
735   other than those acting as tunnels) &MUST; send their own HTTP-version
736   in forwarded messages.  In other words, they are not allowed to blindly
737   forward the first line of an HTTP message without ensuring that the
738   protocol version in that message matches a version to which that
739   intermediary is conformant for both the receiving and
740   sending of messages.  Forwarding an HTTP message without rewriting
741   the HTTP-version might result in communication errors when downstream
742   recipients use the message sender's version to determine what features
743   are safe to use for later communication with that sender.
744</t>
745<t>
746   A client &SHOULD; send a request version equal to the highest
747   version to which the client is conformant and
748   whose major version is no higher than the highest version supported
749   by the server, if this is known.  A client &MUST-NOT; send a
750   version to which it is not conformant.
751</t>
752<t>
753   A client &MAY; send a lower request version if it is known that
754   the server incorrectly implements the HTTP specification, but only
755   after the client has attempted at least one normal request and determined
756   from the response status code or header fields (e.g., <x:ref>Server</x:ref>) that
757   the server improperly handles higher request versions.
758</t>
759<t>
760   A server &SHOULD; send a response version equal to the highest version to
761   which the server is conformant that has a major version less than or equal
762   to the one received in the request.
763   A server &MUST-NOT; send a version to which it is not conformant.
764   A server can send a <x:ref>505 (HTTP Version Not Supported)</x:ref>
765   response if it wishes, for any reason, to refuse service of the client's
766   major protocol version.
767</t>
768<t>
769   A server &MAY; send an HTTP/1.0 response to a request
770   if it is known or suspected that the client incorrectly implements the
771   HTTP specification and is incapable of correctly processing later
772   version responses, such as when a client fails to parse the version
773   number correctly or when an intermediary is known to blindly forward
774   the HTTP-version even when it doesn't conform to the given minor
775   version of the protocol. Such protocol downgrades &SHOULD-NOT; be
776   performed unless triggered by specific client attributes, such as when
777   one or more of the request header fields (e.g., <x:ref>User-Agent</x:ref>)
778   uniquely match the values sent by a client known to be in error.
779</t>
780<t>
781   The intention of HTTP's versioning design is that the major number
782   will only be incremented if an incompatible message syntax is
783   introduced, and that the minor number will only be incremented when
784   changes made to the protocol have the effect of adding to the message
785   semantics or implying additional capabilities of the sender.  However,
786   the minor version was not incremented for the changes introduced between
787   <xref target="RFC2068"/> and <xref target="RFC2616"/>, and this revision
788   has specifically avoided any such changes to the protocol.
789</t>
790<t>
791   When an HTTP message is received with a major version number that the
792   recipient implements, but a higher minor version number than what the
793   recipient implements, the recipient &SHOULD; process the message as if it
794   were in the highest minor version within that major version to which the
795   recipient is conformant. A recipient can assume that a message with a
796   higher minor version, when sent to a recipient that has not yet indicated
797   support for that higher version, is sufficiently backwards-compatible to be
798   safely processed by any implementation of the same major version.
799</t>
800</section>
801
802<section title="Uniform Resource Identifiers" anchor="uri">
803<iref primary="true" item="resource"/>
804<t>
805   Uniform Resource Identifiers (URIs) <xref target="RFC3986"/> are used
806   throughout HTTP as the means for identifying resources (&resource;).
807   URI references are used to target requests, indicate redirects, and define
808   relationships.
809</t>
810  <x:anchor-alias value="URI-reference"/>
811  <x:anchor-alias value="absolute-URI"/>
812  <x:anchor-alias value="relative-part"/>
813  <x:anchor-alias value="scheme"/>
814  <x:anchor-alias value="authority"/>
815  <x:anchor-alias value="uri-host"/>
816  <x:anchor-alias value="port"/>
817  <x:anchor-alias value="path"/>
818  <x:anchor-alias value="path-abempty"/>
819  <x:anchor-alias value="segment"/>
820  <x:anchor-alias value="query"/>
821  <x:anchor-alias value="fragment"/>
822  <x:anchor-alias value="absolute-path"/>
823  <x:anchor-alias value="partial-URI"/>
824<t>
825   The definitions of "URI-reference",
826   "absolute-URI", "relative-part", "scheme", "authority", "port", "host",
827   "path-abempty", "segment", "query", and "fragment" are adopted from the
828   URI generic syntax.
829   An "absolute-path" rule is defined for protocol elements that can contain a
830   non-empty path component. (This rule differs slightly from RFC 3986's
831   path-abempty rule, which allows for an empty path to be used in references,
832   and path-absolute rule, which does not allow paths that begin with "//".)
833   A "partial-URI" rule is defined for protocol elements
834   that can contain a relative URI but not a fragment component.
835</t>
836<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="URI-reference"><!--exported production--></iref><iref primary="true" item="Grammar" subitem="absolute-URI"/><iref primary="true" item="Grammar" subitem="scheme"/><iref primary="true" item="Grammar" subitem="authority"/><iref primary="true" item="Grammar" subitem="absolute-path"/><iref primary="true" item="Grammar" subitem="port"/><iref primary="true" item="Grammar" subitem="query"/><iref primary="true" item="Grammar" subitem="fragment"/><iref primary="true" item="Grammar" subitem="segment"/><iref primary="true" item="Grammar" subitem="uri-host"/><iref primary="true" item="Grammar" subitem="partial-URI"><!--exported production--></iref>
837  <x:ref>URI-reference</x:ref> = &lt;URI-reference, defined in <xref target="RFC3986" x:fmt="," x:sec="4.1"/>&gt;
838  <x:ref>absolute-URI</x:ref>  = &lt;absolute-URI, defined in <xref target="RFC3986" x:fmt="," x:sec="4.3"/>&gt;
839  <x:ref>relative-part</x:ref> = &lt;relative-part, defined in <xref target="RFC3986" x:fmt="," x:sec="4.2"/>&gt;
840  <x:ref>scheme</x:ref>        = &lt;scheme, defined in <xref target="RFC3986" x:fmt="," x:sec="3.1"/>&gt;
841  <x:ref>authority</x:ref>     = &lt;authority, defined in <xref target="RFC3986" x:fmt="," x:sec="3.2"/>&gt;
842  <x:ref>uri-host</x:ref>      = &lt;host, defined in <xref target="RFC3986" x:fmt="," x:sec="3.2.2"/>&gt;
843  <x:ref>port</x:ref>          = &lt;port, defined in <xref target="RFC3986" x:fmt="," x:sec="3.2.3"/>&gt;
844  <x:ref>path-abempty</x:ref>  = &lt;path-abempty, defined in <xref target="RFC3986" x:fmt="," x:sec="3.3"/>&gt;
845  <x:ref>segment</x:ref>       = &lt;segment, defined in <xref target="RFC3986" x:fmt="," x:sec="3.3"/>&gt;
846  <x:ref>query</x:ref>         = &lt;query, defined in <xref target="RFC3986" x:fmt="," x:sec="3.4"/>&gt;
847  <x:ref>fragment</x:ref>      = &lt;fragment, defined in <xref target="RFC3986" x:fmt="," x:sec="3.5"/>&gt;
848 
849  <x:ref>absolute-path</x:ref> = 1*( "/" segment )
850  <x:ref>partial-URI</x:ref>   = relative-part [ "?" query ]
851</artwork></figure>
852<t>
853   Each protocol element in HTTP that allows a URI reference will indicate
854   in its ABNF production whether the element allows any form of reference
855   (URI-reference), only a URI in absolute form (absolute-URI), only the
856   path and optional query components, or some combination of the above.
857   Unless otherwise indicated, URI references are parsed
858   relative to the effective request URI
859   (<xref target="effective.request.uri"/>).
860</t>
861
862<section title="http URI scheme" anchor="http.uri">
863  <x:anchor-alias value="http-URI"/>
864  <iref item="http URI scheme" primary="true"/>
865  <iref item="URI scheme" subitem="http" primary="true"/>
866<t>
867   The "http" URI scheme is hereby defined for the purpose of minting
868   identifiers according to their association with the hierarchical
869   namespace governed by a potential HTTP origin server listening for
870   TCP (<xref target="RFC0793"/>) connections on a given port.
871</t>
872<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="http-URI"><!--terminal production--></iref>
873  <x:ref>http-URI</x:ref> = "http:" "//" <x:ref>authority</x:ref> <x:ref>path-abempty</x:ref> [ "?" <x:ref>query</x:ref> ]
874             [ "#" <x:ref>fragment</x:ref> ]
875</artwork></figure>
876<t>
877   The origin server for an "http" URI is identified by the
878   <x:ref>authority</x:ref> component, which includes a host identifier
879   and optional TCP port (<xref target="RFC3986" x:fmt="," x:sec="3.2.2"/>).
880   The hierarchical path component and optional query component serve as an
881   identifier for a potential target resource within that origin server's name
882   space. The optional fragment component allows for indirect identification
883   of a secondary resource, independent of the URI scheme, as defined in
884   <xref target="RFC3986" x:fmt="of" x:sec="3.5"/>.
885</t>
886<t>
887   A sender &MUST-NOT; generate an "http" URI with an empty host identifier.
888   A recipient that processes such a URI reference &MUST; reject it as invalid.
889</t>
890<t>
891   If the host identifier is provided as an IP address, the origin server is
892   the listener (if any) on the indicated TCP port at that IP address.
893   If host is a registered name, the registered name is an indirect identifier
894   for use with a name resolution service, such as DNS, to find an address for
895   that origin server.
896   If the port subcomponent is empty or not given, TCP port 80 (the
897   reserved port for WWW services) is the default.
898</t>
899<t>
900   Note that the presence of a URI with a given authority component does not
901   imply that there is always an HTTP server listening for connections on
902   that host and port. Anyone can mint a URI. What the authority component
903   determines is who has the right to respond authoritatively to requests that
904   target the identified resource. The delegated nature of registered names
905   and IP addresses creates a federated namespace, based on control over the
906   indicated host and port, whether or not an HTTP server is present.
907   See <xref target="establishing.authority"/> for security considerations
908   related to establishing authority.
909</t>
910<t>
911   When an "http" URI is used within a context that calls for access to the
912   indicated resource, a client &MAY; attempt access by resolving
913   the host to an IP address, establishing a TCP connection to that address
914   on the indicated port, and sending an HTTP request message
915   (<xref target="http.message"/>) containing the URI's identifying data
916   (<xref target="message.routing"/>) to the server.
917   If the server responds to that request with a non-interim HTTP response
918   message, as described in &status-codes;, then that response
919   is considered an authoritative answer to the client's request.
920</t>
921<t>
922   Although HTTP is independent of the transport protocol, the "http"
923   scheme is specific to TCP-based services because the name delegation
924   process depends on TCP for establishing authority.
925   An HTTP service based on some other underlying connection protocol
926   would presumably be identified using a different URI scheme, just as
927   the "https" scheme (below) is used for resources that require an
928   end-to-end secured connection. Other protocols might also be used to
929   provide access to "http" identified resources &mdash; it is only the
930   authoritative interface that is specific to TCP.
931</t>
932<t>
933   The URI generic syntax for authority also includes a deprecated
934   userinfo subcomponent (<xref target="RFC3986" x:fmt="," x:sec="3.2.1"/>)
935   for including user authentication information in the URI.  Some
936   implementations make use of the userinfo component for internal
937   configuration of authentication information, such as within command
938   invocation options, configuration files, or bookmark lists, even
939   though such usage might expose a user identifier or password.
940   A sender &MUST-NOT; generate the userinfo subcomponent (and its "@"
941   delimiter) when an "http" URI reference is generated within a message as a
942   request target or header field value.
943   Before making use of an "http" URI reference received from an untrusted
944   source, a recipient &SHOULD; parse for userinfo and treat its presence as
945   an error; it is likely being used to obscure the authority for the sake of
946   phishing attacks.
947</t>
948</section>
949
950<section title="https URI scheme" anchor="https.uri">
951   <x:anchor-alias value="https-URI"/>
952   <iref item="https URI scheme"/>
953   <iref item="URI scheme" subitem="https"/>
954<t>
955   The "https" URI scheme is hereby defined for the purpose of minting
956   identifiers according to their association with the hierarchical
957   namespace governed by a potential HTTP origin server listening to a
958   given TCP port for TLS-secured connections (<xref target="RFC5246"/>).
959</t>
960<t>
961   All of the requirements listed above for the "http" scheme are also
962   requirements for the "https" scheme, except that TCP port 443 is the
963   default if the port subcomponent is empty or not given,
964   and the user agent &MUST; ensure that its connection to the origin
965   server is secured through the use of strong encryption, end-to-end,
966   prior to sending the first HTTP request.
967</t>
968<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="https-URI"><!--terminal production--></iref>
969  <x:ref>https-URI</x:ref> = "https:" "//" <x:ref>authority</x:ref> <x:ref>path-abempty</x:ref> [ "?" <x:ref>query</x:ref> ]
970              [ "#" <x:ref>fragment</x:ref> ]
971</artwork></figure>
972<t>
973   Note that the "https" URI scheme depends on both TLS and TCP for
974   establishing authority.
975   Resources made available via the "https" scheme have no shared
976   identity with the "http" scheme even if their resource identifiers
977   indicate the same authority (the same host listening to the same
978   TCP port).  They are distinct name spaces and are considered to be
979   distinct origin servers.  However, an extension to HTTP that is
980   defined to apply to entire host domains, such as the Cookie protocol
981   <xref target="RFC6265"/>, can allow information
982   set by one service to impact communication with other services
983   within a matching group of host domains.
984</t>
985<t>
986   The process for authoritative access to an "https" identified
987   resource is defined in <xref target="RFC2818"/>.
988</t>
989</section>
990
991<section title="http and https URI Normalization and Comparison" anchor="uri.comparison">
992<t>
993   Since the "http" and "https" schemes conform to the URI generic syntax,
994   such URIs are normalized and compared according to the algorithm defined
995   in <xref target="RFC3986" x:fmt="of" x:sec="6"/>, using the defaults
996   described above for each scheme.
997</t>
998<t>
999   If the port is equal to the default port for a scheme, the normal form is
1000   to omit the port subcomponent. When not being used in absolute form as the
1001   request target of an OPTIONS request, an empty path component is equivalent
1002   to an absolute path of "/", so the normal form is to provide a path of "/"
1003   instead. The scheme and host are case-insensitive and normally provided in
1004   lowercase; all other components are compared in a case-sensitive manner.
1005   Characters other than those in the "reserved" set are equivalent to their
1006   percent-encoded octets: the normal form is to not encode them
1007   (see Sections <xref target="RFC3986" x:fmt="number" x:sec="2.1"/> and
1008   <xref target="RFC3986" x:fmt="number" x:sec="2.2"/> of
1009   <xref target="RFC3986"/>).
1010</t>
1011<t>
1012   For example, the following three URIs are equivalent:
1013</t>
1014<figure><artwork type="example">
1015   http://example.com:80/~smith/home.html
1016   http://EXAMPLE.com/%7Esmith/home.html
1017   http://EXAMPLE.com:/%7esmith/home.html
1018</artwork></figure>
1019</section>
1020</section>
1021</section>
1022
1023<section title="Message Format" anchor="http.message">
1024<x:anchor-alias value="generic-message"/>
1025<x:anchor-alias value="message.types"/>
1026<x:anchor-alias value="HTTP-message"/>
1027<x:anchor-alias value="start-line"/>
1028<iref item="header section"/>
1029<iref item="headers"/>
1030<iref item="header field"/>
1031<t>
1032   All HTTP/1.1 messages consist of a start-line followed by a sequence of
1033   octets in a format similar to the Internet Message Format
1034   <xref target="RFC5322"/>: zero or more header fields (collectively
1035   referred to as the "headers" or the "header section"), an empty line
1036   indicating the end of the header section, and an optional message body.
1037</t>
1038<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="HTTP-message"><!--terminal production--></iref>
1039  <x:ref>HTTP-message</x:ref>   = <x:ref>start-line</x:ref>
1040                   *( <x:ref>header-field</x:ref> <x:ref>CRLF</x:ref> )
1041                   <x:ref>CRLF</x:ref>
1042                   [ <x:ref>message-body</x:ref> ]
1043</artwork></figure>
1044<t>
1045   The normal procedure for parsing an HTTP message is to read the
1046   start-line into a structure, read each header field into a hash
1047   table by field name until the empty line, and then use the parsed
1048   data to determine if a message body is expected.  If a message body
1049   has been indicated, then it is read as a stream until an amount
1050   of octets equal to the message body length is read or the connection
1051   is closed.
1052</t>
1053<t>
1054   A recipient &MUST; parse an HTTP message as a sequence of octets in an
1055   encoding that is a superset of US-ASCII <xref target="USASCII"/>.
1056   Parsing an HTTP message as a stream of Unicode characters, without regard
1057   for the specific encoding, creates security vulnerabilities due to the
1058   varying ways that string processing libraries handle invalid multibyte
1059   character sequences that contain the octet LF (%x0A).  String-based
1060   parsers can only be safely used within protocol elements after the element
1061   has been extracted from the message, such as within a header field-value
1062   after message parsing has delineated the individual fields.
1063</t>
1064<t>
1065   An HTTP message can be parsed as a stream for incremental processing or
1066   forwarding downstream.  However, recipients cannot rely on incremental
1067   delivery of partial messages, since some implementations will buffer or
1068   delay message forwarding for the sake of network efficiency, security
1069   checks, or payload transformations.
1070</t>
1071<t>
1072   A sender &MUST-NOT; send whitespace between the start-line and
1073   the first header field.
1074   A recipient that receives whitespace between the start-line and
1075   the first header field &MUST; either reject the message as invalid or
1076   consume each whitespace-preceded line without further processing of it
1077   (i.e., ignore the entire line, along with any subsequent lines preceded
1078   by whitespace, until a properly formed header field is received or the
1079   header section is terminated).
1080</t>
1081<t>
1082   The presence of such whitespace in a request
1083   might be an attempt to trick a server into ignoring that field or
1084   processing the line after it as a new request, either of which might
1085   result in a security vulnerability if other implementations within
1086   the request chain interpret the same message differently.
1087   Likewise, the presence of such whitespace in a response might be
1088   ignored by some clients or cause others to cease parsing.
1089</t>
1090
1091<section title="Start Line" anchor="start.line">
1092  <x:anchor-alias value="Start-Line"/>
1093<t>
1094   An HTTP message can either be a request from client to server or a
1095   response from server to client.  Syntactically, the two types of message
1096   differ only in the start-line, which is either a request-line (for requests)
1097   or a status-line (for responses), and in the algorithm for determining
1098   the length of the message body (<xref target="message.body"/>).
1099</t>
1100<t>
1101   In theory, a client could receive requests and a server could receive
1102   responses, distinguishing them by their different start-line formats,
1103   but in practice servers are implemented to only expect a request
1104   (a response is interpreted as an unknown or invalid request method)
1105   and clients are implemented to only expect a response.
1106</t>
1107<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="start-line"/>
1108  <x:ref>start-line</x:ref>     = <x:ref>request-line</x:ref> / <x:ref>status-line</x:ref>
1109</artwork></figure>
1110
1111<section title="Request Line" anchor="request.line">
1112  <x:anchor-alias value="Request"/>
1113  <x:anchor-alias value="request-line"/>
1114<t>
1115   A request-line begins with a method token, followed by a single
1116   space (SP), the request-target, another single space (SP), the
1117   protocol version, and ending with CRLF.
1118</t>
1119<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="request-line"/>
1120  <x:ref>request-line</x:ref>   = <x:ref>method</x:ref> <x:ref>SP</x:ref> <x:ref>request-target</x:ref> <x:ref>SP</x:ref> <x:ref>HTTP-version</x:ref> <x:ref>CRLF</x:ref>
1121</artwork></figure>
1122<iref primary="true" item="method"/>
1123<t anchor="method">
1124   The method token indicates the request method to be performed on the
1125   target resource. The request method is case-sensitive.
1126</t>
1127<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="method"/>
1128  <x:ref>method</x:ref>         = <x:ref>token</x:ref>
1129</artwork></figure>
1130<t>
1131   The request methods defined by this specification can be found in
1132   &methods;, along with information regarding the HTTP method registry
1133   and considerations for defining new methods.
1134</t>
1135<iref item="request-target"/>
1136<t>
1137   The request-target identifies the target resource upon which to apply
1138   the request, as defined in <xref target="request-target"/>.
1139</t>
1140<t>
1141   Recipients typically parse the request-line into its component parts by
1142   splitting on whitespace (see <xref target="message.robustness"/>), since
1143   no whitespace is allowed in the three components.
1144   Unfortunately, some user agents fail to properly encode or exclude
1145   whitespace found in hypertext references, resulting in those disallowed
1146   characters being sent in a request-target.
1147</t>
1148<t>
1149   Recipients of an invalid request-line &SHOULD; respond with either a
1150   <x:ref>400 (Bad Request)</x:ref> error or a <x:ref>301 (Moved Permanently)</x:ref>
1151   redirect with the request-target properly encoded.  A recipient &SHOULD-NOT;
1152   attempt to autocorrect and then process the request without a redirect,
1153   since the invalid request-line might be deliberately crafted to bypass
1154   security filters along the request chain.
1155</t>
1156<t>
1157   HTTP does not place a pre-defined limit on the length of a request-line,
1158   as described in <xref target="conformance"/>.
1159   A server that receives a method longer than any that it implements
1160   &SHOULD; respond with a <x:ref>501 (Not Implemented)</x:ref> status code.
1161   A server that receives a request-target longer than any URI it wishes to
1162   parse &MUST; respond with a
1163   <x:ref>414 (URI Too Long)</x:ref> status code (see &status-414;).
1164</t>
1165<t>
1166   Various ad-hoc limitations on request-line length are found in practice.
1167   It is &RECOMMENDED; that all HTTP senders and recipients support, at a
1168   minimum, request-line lengths of 8000 octets.
1169</t>
1170</section>
1171
1172<section title="Status Line" anchor="status.line">
1173  <x:anchor-alias value="response"/>
1174  <x:anchor-alias value="status-line"/>
1175  <x:anchor-alias value="status-code"/>
1176  <x:anchor-alias value="reason-phrase"/>
1177<t>
1178   The first line of a response message is the status-line, consisting
1179   of the protocol version, a space (SP), the status code, another space,
1180   a possibly-empty textual phrase describing the status code, and
1181   ending with CRLF.
1182</t>
1183<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="status-line"/>
1184  <x:ref>status-line</x:ref> = <x:ref>HTTP-version</x:ref> <x:ref>SP</x:ref> <x:ref>status-code</x:ref> <x:ref>SP</x:ref> <x:ref>reason-phrase</x:ref> <x:ref>CRLF</x:ref>
1185</artwork></figure>
1186<t>
1187   The status-code element is a 3-digit integer code describing the
1188   result of the server's attempt to understand and satisfy the client's
1189   corresponding request. The rest of the response message is to be
1190   interpreted in light of the semantics defined for that status code.
1191   See &status-codes; for information about the semantics of status codes,
1192   including the classes of status code (indicated by the first digit),
1193   the status codes defined by this specification, considerations for the
1194   definition of new status codes, and the IANA registry.
1195</t>
1196<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="status-code"/>
1197  <x:ref>status-code</x:ref>    = 3<x:ref>DIGIT</x:ref>
1198</artwork></figure>
1199<t>  
1200   The reason-phrase element exists for the sole purpose of providing a
1201   textual description associated with the numeric status code, mostly
1202   out of deference to earlier Internet application protocols that were more
1203   frequently used with interactive text clients. A client &SHOULD; ignore
1204   the reason-phrase content.
1205</t>
1206<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="reason-phrase"/>
1207  <x:ref>reason-phrase</x:ref>  = *( <x:ref>HTAB</x:ref> / <x:ref>SP</x:ref> / <x:ref>VCHAR</x:ref> / <x:ref>obs-text</x:ref> )
1208</artwork></figure>
1209</section>
1210</section>
1211
1212<section title="Header Fields" anchor="header.fields">
1213  <x:anchor-alias value="header-field"/>
1214  <x:anchor-alias value="field-content"/>
1215  <x:anchor-alias value="field-name"/>
1216  <x:anchor-alias value="field-value"/>
1217  <x:anchor-alias value="field-vchar"/>
1218  <x:anchor-alias value="obs-fold"/>
1219<t>
1220   Each header field consists of a case-insensitive field name
1221   followed by a colon (":"), optional leading whitespace, the field value,
1222   and optional trailing whitespace.
1223</t>
1224<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="header-field"/><iref primary="true" item="Grammar" subitem="field-name"/><iref primary="true" item="Grammar" subitem="field-value"/><iref primary="true" item="Grammar" subitem="field-vchar"/><iref primary="true" item="Grammar" subitem="field-content"/><iref primary="true" item="Grammar" subitem="obs-fold"/>
1225  <x:ref>header-field</x:ref>   = <x:ref>field-name</x:ref> ":" <x:ref>OWS</x:ref> <x:ref>field-value</x:ref> <x:ref>OWS</x:ref>
1226
1227  <x:ref>field-name</x:ref>     = <x:ref>token</x:ref>
1228  <x:ref>field-value</x:ref>    = *( <x:ref>field-content</x:ref> / <x:ref>obs-fold</x:ref> )
1229  <x:ref>field-content</x:ref>  = <x:ref>field-vchar</x:ref> [ 1*( <x:ref>SP</x:ref> / <x:ref>HTAB</x:ref> ) <x:ref>field-vchar</x:ref> ]
1230  <x:ref>field-vchar</x:ref>    = <x:ref>VCHAR</x:ref> / <x:ref>obs-text</x:ref>
1231
1232  <x:ref>obs-fold</x:ref>       = <x:ref>CRLF</x:ref> 1*( <x:ref>SP</x:ref> / <x:ref>HTAB</x:ref> )
1233                 ; obsolete line folding
1234                 ; see <xref target="field.parsing"/>
1235</artwork></figure>
1236<t>
1237   The field-name token labels the corresponding field-value as having the
1238   semantics defined by that header field.  For example, the <x:ref>Date</x:ref>
1239   header field is defined in &header-date; as containing the origination
1240   timestamp for the message in which it appears.
1241</t>
1242
1243<section title="Field Extensibility" anchor="field.extensibility">
1244<t>
1245   Header fields are fully extensible: there is no limit on the
1246   introduction of new field names, each presumably defining new semantics,
1247   nor on the number of header fields used in a given message.  Existing
1248   fields are defined in each part of this specification and in many other
1249   specifications outside this document set.
1250</t>
1251<t>
1252   New header fields can be defined such that, when they are understood by a
1253   recipient, they might override or enhance the interpretation of previously
1254   defined header fields, define preconditions on request evaluation, or
1255   refine the meaning of responses.
1256</t>
1257<t>
1258   A proxy &MUST; forward unrecognized header fields unless the
1259   field-name is listed in the <x:ref>Connection</x:ref> header field
1260   (<xref target="header.connection"/>) or the proxy is specifically
1261   configured to block, or otherwise transform, such fields.
1262   Other recipients &SHOULD; ignore unrecognized header fields.
1263   These requirements allow HTTP's functionality to be enhanced without
1264   requiring prior update of deployed intermediaries.
1265</t>
1266<t>
1267   All defined header fields ought to be registered with IANA in the
1268   Message Header Field Registry, as described in &iana-header-registry;.
1269</t>
1270</section>
1271
1272<section title="Field Order" anchor="field.order">
1273<t>
1274   The order in which header fields with differing field names are
1275   received is not significant. However, it is good practice to send
1276   header fields that contain control data first, such as <x:ref>Host</x:ref>
1277   on requests and <x:ref>Date</x:ref> on responses, so that implementations
1278   can decide when not to handle a message as early as possible.
1279   A server &MUST-NOT; apply a request to the target resource until the entire
1280   request header section is received, since later header fields might include
1281   conditionals, authentication credentials, or deliberately misleading
1282   duplicate header fields that would impact request processing.
1283</t>
1284<t>
1285   A sender &MUST-NOT; generate multiple header fields with the same field
1286   name in a message unless either the entire field value for that
1287   header field is defined as a comma-separated list [i.e., #(values)]
1288   or the header field is a well-known exception (as noted below).
1289</t>
1290<t>
1291   A recipient &MAY; combine multiple header fields with the same field name
1292   into one "field-name: field-value" pair, without changing the semantics of
1293   the message, by appending each subsequent field value to the combined
1294   field value in order, separated by a comma. The order in which
1295   header fields with the same field name are received is therefore
1296   significant to the interpretation of the combined field value;
1297   a proxy &MUST-NOT; change the order of these field values when
1298   forwarding a message.
1299</t>
1300<x:note>
1301  <t>
1302   &Note; In practice, the "Set-Cookie" header field (<xref target="RFC6265"/>)
1303   often appears multiple times in a response message and does not use the
1304   list syntax, violating the above requirements on multiple header fields
1305   with the same name. Since it cannot be combined into a single field-value,
1306   recipients ought to handle "Set-Cookie" as a special case while processing
1307   header fields. (See Appendix A.2.3 of <xref target="Kri2001"/> for details.)
1308  </t>
1309</x:note>
1310</section>
1311
1312<section title="Whitespace" anchor="whitespace">
1313<t anchor="rule.LWS">
1314   This specification uses three rules to denote the use of linear
1315   whitespace: OWS (optional whitespace), RWS (required whitespace), and
1316   BWS ("bad" whitespace).
1317</t>
1318<t anchor="rule.OWS">
1319   The OWS rule is used where zero or more linear whitespace octets might
1320   appear. For protocol elements where optional whitespace is preferred to
1321   improve readability, a sender &SHOULD; generate the optional whitespace
1322   as a single SP; otherwise, a sender &SHOULD-NOT; generate optional
1323   whitespace except as needed to white-out invalid or unwanted protocol
1324   elements during in-place message filtering.
1325</t>
1326<t anchor="rule.RWS">
1327   The RWS rule is used when at least one linear whitespace octet is required
1328   to separate field tokens. A sender &SHOULD; generate RWS as a single SP.
1329</t>
1330<t anchor="rule.BWS">
1331   The BWS rule is used where the grammar allows optional whitespace only for
1332   historical reasons. A sender &MUST-NOT; generate BWS in messages.
1333   A recipient &MUST; parse for such bad whitespace and remove it before
1334   interpreting the protocol element.
1335</t>
1336<t anchor="rule.whitespace">
1337  <x:anchor-alias value="BWS"/>
1338  <x:anchor-alias value="OWS"/>
1339  <x:anchor-alias value="RWS"/>
1340</t>
1341<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="OWS"/><iref primary="true" item="Grammar" subitem="RWS"/><iref primary="true" item="Grammar" subitem="BWS"/>
1342  <x:ref>OWS</x:ref>            = *( <x:ref>SP</x:ref> / <x:ref>HTAB</x:ref> )
1343                 ; optional whitespace
1344  <x:ref>RWS</x:ref>            = 1*( <x:ref>SP</x:ref> / <x:ref>HTAB</x:ref> )
1345                 ; required whitespace
1346  <x:ref>BWS</x:ref>            = <x:ref>OWS</x:ref>
1347                 ; "bad" whitespace
1348</artwork></figure>
1349</section>
1350
1351<section title="Field Parsing" anchor="field.parsing">
1352<t>
1353   Messages are parsed using a generic algorithm, independent of the
1354   individual header field names. The contents within a given field value are
1355   not parsed until a later stage of message interpretation (usually after the
1356   message's entire header section has been processed).
1357   Consequently, this specification does not use ABNF rules to define each
1358   "Field-Name: Field Value" pair, as was done in previous editions.
1359   Instead, this specification uses ABNF rules which are named according to
1360   each registered field name, wherein the rule defines the valid grammar for
1361   that field's corresponding field values (i.e., after the field-value
1362   has been extracted from the header section by a generic field parser).
1363</t>
1364<t>
1365   No whitespace is allowed between the header field-name and colon.
1366   In the past, differences in the handling of such whitespace have led to
1367   security vulnerabilities in request routing and response handling.
1368   A server &MUST; reject any received request message that contains
1369   whitespace between a header field-name and colon with a response code of
1370   <x:ref>400 (Bad Request)</x:ref>. A proxy &MUST; remove any such whitespace
1371   from a response message before forwarding the message downstream.
1372</t>
1373<t>
1374   A field value might be preceded and/or followed by optional whitespace
1375   (OWS); a single SP preceding the field-value is preferred for consistent
1376   readability by humans.
1377   The field value does not include any leading or trailing white space: OWS
1378   occurring before the first non-whitespace octet of the field value or after
1379   the last non-whitespace octet of the field value ought to be excluded by
1380   parsers when extracting the field value from a header field.
1381</t>
1382<t>
1383   Historically, HTTP header field values could be extended over multiple
1384   lines by preceding each extra line with at least one space or horizontal
1385   tab (obs-fold). This specification deprecates such line folding except
1386   within the message/http media type
1387   (<xref target="internet.media.type.message.http"/>).
1388   A sender &MUST-NOT; generate a message that includes line folding
1389   (i.e., that has any field-value that contains a match to the
1390   <x:ref>obs-fold</x:ref> rule) unless the message is intended for packaging
1391   within the message/http media type.
1392</t>
1393<t>
1394   A server that receives an <x:ref>obs-fold</x:ref> in a request message that
1395   is not within a message/http container &MUST; either reject the message by
1396   sending a <x:ref>400 (Bad Request)</x:ref>, preferably with a
1397   representation explaining that obsolete line folding is unacceptable, or
1398   replace each received <x:ref>obs-fold</x:ref> with one or more
1399   <x:ref>SP</x:ref> octets prior to interpreting the field value or
1400   forwarding the message downstream.
1401</t>
1402<t>
1403   A proxy or gateway that receives an <x:ref>obs-fold</x:ref> in a response
1404   message that is not within a message/http container &MUST; either discard
1405   the message and replace it with a <x:ref>502 (Bad Gateway)</x:ref>
1406   response, preferably with a representation explaining that unacceptable
1407   line folding was received, or replace each received <x:ref>obs-fold</x:ref>
1408   with one or more <x:ref>SP</x:ref> octets prior to interpreting the field
1409   value or forwarding the message downstream.
1410</t>
1411<t>
1412   A user agent that receives an <x:ref>obs-fold</x:ref> in a response message
1413   that is not within a message/http container &MUST; replace each received
1414   <x:ref>obs-fold</x:ref> with one or more <x:ref>SP</x:ref> octets prior to
1415   interpreting the field value.
1416</t>
1417<t>
1418   Historically, HTTP has allowed field content with text in the ISO-8859-1
1419   <xref target="ISO-8859-1"/> charset, supporting other charsets only
1420   through use of <xref target="RFC2047"/> encoding.
1421   In practice, most HTTP header field values use only a subset of the
1422   US-ASCII charset <xref target="USASCII"/>. Newly defined
1423   header fields &SHOULD; limit their field values to US-ASCII octets.
1424   A recipient &SHOULD; treat other octets in field content (obs-text) as
1425   opaque data.
1426</t>
1427</section>
1428
1429<section title="Field Limits" anchor="field.limits">
1430<t>
1431   HTTP does not place a pre-defined limit on the length of each header field
1432   or on the length of the header section as a whole, as described in
1433   <xref target="conformance"/>. Various ad-hoc limitations on individual
1434   header field length are found in practice, often depending on the specific
1435   field semantics.
1436</t>
1437<t>
1438   A server that receives a request header field, or set of fields, larger
1439   than it wishes to process &MUST; respond with an appropriate
1440   <x:ref>4xx (Client Error)</x:ref> status code. Ignoring such header fields
1441   would increase the server's vulnerability to request smuggling attacks
1442   (<xref target="request.smuggling"/>).
1443</t>
1444<t>
1445   A client &MAY; discard or truncate received header fields that are larger
1446   than the client wishes to process if the field semantics are such that the
1447   dropped value(s) can be safely ignored without changing the
1448   message framing or response semantics.
1449</t>
1450</section>
1451
1452<section title="Field value components" anchor="field.components">
1453<t anchor="rule.token.separators">
1454  <x:anchor-alias value="tchar"/>
1455  <x:anchor-alias value="token"/>
1456  <iref item="Delimiters"/>
1457   Most HTTP header field values are defined using common syntax components
1458   (token, quoted-string, and comment) separated by whitespace or specific
1459   delimiting characters. Delimiters are chosen from the set of US-ASCII
1460   visual characters not allowed in a <x:ref>token</x:ref>
1461   (DQUOTE and "(),/:;&lt;=>?@[\]{}").
1462</t>
1463<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="token"/><iref primary="true" item="Grammar" subitem="tchar"/>
1464  <x:ref>token</x:ref>          = 1*<x:ref>tchar</x:ref>
1465<!--
1466  NOTE: the definition of tchar and the prose above about special characters need to match!
1467 -->
1468  <x:ref>tchar</x:ref>          = "!" / "#" / "$" / "%" / "&amp;" / "'" / "*"
1469                 / "+" / "-" / "." / "^" / "_" / "`" / "|" / "~"
1470                 / <x:ref>DIGIT</x:ref> / <x:ref>ALPHA</x:ref>
1471                 ; any <x:ref>VCHAR</x:ref>, except delimiters
1472</artwork></figure>
1473<t anchor="rule.quoted-string">
1474  <x:anchor-alias value="quoted-string"/>
1475  <x:anchor-alias value="qdtext"/>
1476  <x:anchor-alias value="obs-text"/>
1477   A string of text is parsed as a single value if it is quoted using
1478   double-quote marks.
1479</t>
1480<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="quoted-string"/><iref primary="true" item="Grammar" subitem="qdtext"/><iref primary="true" item="Grammar" subitem="obs-text"/>
1481  <x:ref>quoted-string</x:ref>  = <x:ref>DQUOTE</x:ref> *( <x:ref>qdtext</x:ref> / <x:ref>quoted-pair</x:ref> ) <x:ref>DQUOTE</x:ref>
1482  <x:ref>qdtext</x:ref>         = <x:ref>HTAB</x:ref> / <x:ref>SP</x:ref> /%x21 / %x23-5B / %x5D-7E / <x:ref>obs-text</x:ref>
1483  <x:ref>obs-text</x:ref>       = %x80-FF
1484</artwork></figure>
1485<t anchor="rule.comment">
1486  <x:anchor-alias value="comment"/>
1487  <x:anchor-alias value="ctext"/>
1488   Comments can be included in some HTTP header fields by surrounding
1489   the comment text with parentheses. Comments are only allowed in
1490   fields containing "comment" as part of their field value definition.
1491</t>
1492<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="comment"/><iref primary="true" item="Grammar" subitem="ctext"/>
1493  <x:ref>comment</x:ref>        = "(" *( <x:ref>ctext</x:ref> / <x:ref>quoted-pair</x:ref> / <x:ref>comment</x:ref> ) ")"
1494  <x:ref>ctext</x:ref>          = <x:ref>HTAB</x:ref> / <x:ref>SP</x:ref> / %x21-27 / %x2A-5B / %x5D-7E / <x:ref>obs-text</x:ref>
1495</artwork></figure>
1496<t anchor="rule.quoted-pair">
1497  <x:anchor-alias value="quoted-pair"/>
1498   The backslash octet ("\") can be used as a single-octet
1499   quoting mechanism within quoted-string and comment constructs.
1500   Recipients that process the value of a quoted-string &MUST; handle a
1501   quoted-pair as if it were replaced by the octet following the backslash.
1502</t>
1503<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="quoted-pair"/>
1504  <x:ref>quoted-pair</x:ref>    = "\" ( <x:ref>HTAB</x:ref> / <x:ref>SP</x:ref> / <x:ref>VCHAR</x:ref> / <x:ref>obs-text</x:ref> )
1505</artwork></figure>
1506<t>
1507   A sender &SHOULD-NOT; generate a quoted-pair in a quoted-string except
1508   where necessary to quote DQUOTE and backslash octets occurring within that
1509   string.
1510   A sender &SHOULD-NOT; generate a quoted-pair in a comment except
1511   where necessary to quote parentheses ["(" and ")"] and backslash octets
1512   occurring within that comment.
1513</t>
1514</section>
1515
1516</section>
1517
1518<section title="Message Body" anchor="message.body">
1519  <x:anchor-alias value="message-body"/>
1520<t>
1521   The message body (if any) of an HTTP message is used to carry the
1522   payload body of that request or response.  The message body is
1523   identical to the payload body unless a transfer coding has been
1524   applied, as described in <xref target="header.transfer-encoding"/>.
1525</t>
1526<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="message-body"/>
1527  <x:ref>message-body</x:ref> = *OCTET
1528</artwork></figure>
1529<t>
1530   The rules for when a message body is allowed in a message differ for
1531   requests and responses.
1532</t>
1533<t>
1534   The presence of a message body in a request is signaled by a
1535   <x:ref>Content-Length</x:ref> or <x:ref>Transfer-Encoding</x:ref> header
1536   field. Request message framing is independent of method semantics,
1537   even if the method does not define any use for a message body.
1538</t>
1539<t>
1540   The presence of a message body in a response depends on both
1541   the request method to which it is responding and the response
1542   status code (<xref target="status.line"/>).
1543   Responses to the HEAD request method (&HEAD;) never include a message body
1544   because the associated response header fields (e.g.,
1545   <x:ref>Transfer-Encoding</x:ref>, <x:ref>Content-Length</x:ref>, etc.),
1546   if present, indicate only what their values would have been if the request
1547   method had been GET (&GET;).
1548   <x:ref>2xx (Successful)</x:ref> responses to a CONNECT request method
1549   (&CONNECT;) switch to tunnel mode instead of having a message body.
1550   All <x:ref>1xx (Informational)</x:ref>, <x:ref>204 (No Content)</x:ref>, and
1551   <x:ref>304 (Not Modified)</x:ref> responses do not include a message body.
1552   All other responses do include a message body, although the body
1553   might be of zero length.
1554</t>
1555
1556<section title="Transfer-Encoding" anchor="header.transfer-encoding">
1557  <iref primary="true" item="Transfer-Encoding header field" x:for-anchor=""/>
1558  <iref item="chunked (Coding Format)"/>
1559  <x:anchor-alias value="Transfer-Encoding"/>
1560<t>
1561   The Transfer-Encoding header field lists the transfer coding names
1562   corresponding to the sequence of transfer codings that have been
1563   (or will be) applied to the payload body in order to form the message body.
1564   Transfer codings are defined in <xref target="transfer.codings"/>.
1565</t>
1566<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Transfer-Encoding"/>
1567  <x:ref>Transfer-Encoding</x:ref> = 1#<x:ref>transfer-coding</x:ref>
1568</artwork></figure>
1569<t>
1570   Transfer-Encoding is analogous to the Content-Transfer-Encoding field of
1571   MIME, which was designed to enable safe transport of binary data over a
1572   7-bit transport service (<xref target="RFC2045" x:fmt="," x:sec="6"/>).
1573   However, safe transport has a different focus for an 8bit-clean transfer
1574   protocol. In HTTP's case, Transfer-Encoding is primarily intended to
1575   accurately delimit a dynamically generated payload and to distinguish
1576   payload encodings that are only applied for transport efficiency or
1577   security from those that are characteristics of the selected resource.
1578</t>
1579<t>
1580   A recipient &MUST; be able to parse the chunked transfer coding
1581   (<xref target="chunked.encoding"/>) because it plays a crucial role in
1582   framing messages when the payload body size is not known in advance.
1583   A sender &MUST-NOT; apply chunked more than once to a message body
1584   (i.e., chunking an already chunked message is not allowed).
1585   If any transfer coding other than chunked is applied to a request payload
1586   body, the sender &MUST; apply chunked as the final transfer coding to
1587   ensure that the message is properly framed.
1588   If any transfer coding other than chunked is applied to a response payload
1589   body, the sender &MUST; either apply chunked as the final transfer coding
1590   or terminate the message by closing the connection.
1591</t>
1592<figure><preamble>
1593   For example,
1594</preamble><artwork type="example">
1595  Transfer-Encoding: gzip, chunked
1596</artwork><postamble>
1597   indicates that the payload body has been compressed using the gzip
1598   coding and then chunked using the chunked coding while forming the
1599   message body.
1600</postamble></figure>
1601<t>
1602   Unlike <x:ref>Content-Encoding</x:ref> (&content-codings;),
1603   Transfer-Encoding is a property of the message, not of the representation, and
1604   any recipient along the request/response chain &MAY; decode the received
1605   transfer coding(s) or apply additional transfer coding(s) to the message
1606   body, assuming that corresponding changes are made to the Transfer-Encoding
1607   field-value. Additional information about the encoding parameters can be
1608   provided by other header fields not defined by this specification.
1609</t>
1610<t>
1611   Transfer-Encoding &MAY; be sent in a response to a HEAD request or in a
1612   <x:ref>304 (Not Modified)</x:ref> response (&status-304;) to a GET request,
1613   neither of which includes a message body,
1614   to indicate that the origin server would have applied a transfer coding
1615   to the message body if the request had been an unconditional GET.
1616   This indication is not required, however, because any recipient on
1617   the response chain (including the origin server) can remove transfer
1618   codings when they are not needed.
1619</t>
1620<t>
1621   A server &MUST-NOT; send a Transfer-Encoding header field in any response
1622   with a status code of
1623   <x:ref>1xx (Informational)</x:ref> or <x:ref>204 (No Content)</x:ref>.
1624   A server &MUST-NOT; send a Transfer-Encoding header field in any
1625   <x:ref>2xx (Successful)</x:ref> response to a CONNECT request (&CONNECT;).
1626</t>
1627<t>
1628   Transfer-Encoding was added in HTTP/1.1.  It is generally assumed that
1629   implementations advertising only HTTP/1.0 support will not understand
1630   how to process a transfer-encoded payload.
1631   A client &MUST-NOT; send a request containing Transfer-Encoding unless it
1632   knows the server will handle HTTP/1.1 (or later) requests; such knowledge
1633   might be in the form of specific user configuration or by remembering the
1634   version of a prior received response.
1635   A server &MUST-NOT; send a response containing Transfer-Encoding unless
1636   the corresponding request indicates HTTP/1.1 (or later).
1637</t>
1638<t>
1639   A server that receives a request message with a transfer coding it does
1640   not understand &SHOULD; respond with <x:ref>501 (Not Implemented)</x:ref>.
1641</t>
1642</section>
1643
1644<section title="Content-Length" anchor="header.content-length">
1645  <iref primary="true" item="Content-Length header field" x:for-anchor=""/>
1646  <x:anchor-alias value="Content-Length"/>
1647<t>
1648   When a message does not have a <x:ref>Transfer-Encoding</x:ref> header
1649   field, a Content-Length header field can provide the anticipated size,
1650   as a decimal number of octets, for a potential payload body.
1651   For messages that do include a payload body, the Content-Length field-value
1652   provides the framing information necessary for determining where the body
1653   (and message) ends.  For messages that do not include a payload body, the
1654   Content-Length indicates the size of the selected representation
1655   (&representation;).
1656</t>
1657<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Content-Length"/>
1658  <x:ref>Content-Length</x:ref> = 1*<x:ref>DIGIT</x:ref>
1659</artwork></figure>
1660<t>
1661   An example is
1662</t>
1663<figure><artwork type="example">
1664  Content-Length: 3495
1665</artwork></figure>
1666<t>
1667   A sender &MUST-NOT; send a Content-Length header field in any message that
1668   contains a <x:ref>Transfer-Encoding</x:ref> header field.
1669</t>
1670<t>
1671   A user agent &SHOULD; send a Content-Length in a request message when no
1672   <x:ref>Transfer-Encoding</x:ref> is sent and the request method defines
1673   a meaning for an enclosed payload body. For example, a Content-Length
1674   header field is normally sent in a POST request even when the value is
1675   0 (indicating an empty payload body).  A user agent &SHOULD-NOT; send a
1676   Content-Length header field when the request message does not contain a
1677   payload body and the method semantics do not anticipate such a body.
1678</t>
1679<t>
1680   A server &MAY; send a Content-Length header field in a response to a HEAD
1681   request (&HEAD;); a server &MUST-NOT; send Content-Length in such a
1682   response unless its field-value equals the decimal number of octets that
1683   would have been sent in the payload body of a response if the same
1684   request had used the GET method.
1685</t>
1686<t>
1687   A server &MAY; send a Content-Length header field in a
1688   <x:ref>304 (Not Modified)</x:ref> response to a conditional GET request
1689   (&status-304;); a server &MUST-NOT; send Content-Length in such a
1690   response unless its field-value equals the decimal number of octets that
1691   would have been sent in the payload body of a <x:ref>200 (OK)</x:ref>
1692   response to the same request.
1693</t>
1694<t>
1695   A server &MUST-NOT; send a Content-Length header field in any response
1696   with a status code of
1697   <x:ref>1xx (Informational)</x:ref> or <x:ref>204 (No Content)</x:ref>.
1698   A server &MUST-NOT; send a Content-Length header field in any
1699   <x:ref>2xx (Successful)</x:ref> response to a CONNECT request (&CONNECT;).
1700</t>
1701<t>
1702   Aside from the cases defined above, in the absence of Transfer-Encoding,
1703   an origin server &SHOULD; send a Content-Length header field when the
1704   payload body size is known prior to sending the complete header section.
1705   This will allow downstream recipients to measure transfer progress,
1706   know when a received message is complete, and potentially reuse the
1707   connection for additional requests.
1708</t>
1709<t>
1710   Any Content-Length field value greater than or equal to zero is valid.
1711   Since there is no predefined limit to the length of a payload, a
1712   recipient &MUST; anticipate potentially large decimal numerals and
1713   prevent parsing errors due to integer conversion overflows
1714   (<xref target="attack.protocol.element.length"/>).
1715</t>
1716<t>
1717   If a message is received that has multiple Content-Length header fields
1718   with field-values consisting of the same decimal value, or a single
1719   Content-Length header field with a field value containing a list of
1720   identical decimal values (e.g., "Content-Length: 42, 42"), indicating that
1721   duplicate Content-Length header fields have been generated or combined by an
1722   upstream message processor, then the recipient &MUST; either reject the
1723   message as invalid or replace the duplicated field-values with a single
1724   valid Content-Length field containing that decimal value prior to
1725   determining the message body length or forwarding the message.
1726</t>
1727<x:note>
1728  <t>
1729   &Note; HTTP's use of Content-Length for message framing differs
1730   significantly from the same field's use in MIME, where it is an optional
1731   field used only within the "message/external-body" media-type.
1732  </t>
1733</x:note>
1734</section>
1735
1736<section title="Message Body Length" anchor="message.body.length">
1737  <iref item="chunked (Coding Format)"/>
1738<t>
1739   The length of a message body is determined by one of the following
1740   (in order of precedence):
1741</t>
1742<t>
1743  <list style="numbers">
1744    <x:lt><t>
1745     Any response to a HEAD request and any response with a
1746     <x:ref>1xx (Informational)</x:ref>, <x:ref>204 (No Content)</x:ref>, or
1747     <x:ref>304 (Not Modified)</x:ref> status code is always
1748     terminated by the first empty line after the header fields, regardless of
1749     the header fields present in the message, and thus cannot contain a
1750     message body.
1751    </t></x:lt>
1752    <x:lt><t>
1753     Any <x:ref>2xx (Successful)</x:ref> response to a CONNECT request implies that the
1754     connection will become a tunnel immediately after the empty line that
1755     concludes the header fields.  A client &MUST; ignore any
1756     <x:ref>Content-Length</x:ref> or <x:ref>Transfer-Encoding</x:ref> header
1757     fields received in such a message.
1758    </t></x:lt>
1759    <x:lt><t>
1760     If a <x:ref>Transfer-Encoding</x:ref> header field is present
1761     and the chunked transfer coding (<xref target="chunked.encoding"/>)
1762     is the final encoding, the message body length is determined by reading
1763     and decoding the chunked data until the transfer coding indicates the
1764     data is complete.
1765    </t>
1766    <t>
1767     If a <x:ref>Transfer-Encoding</x:ref> header field is present in a
1768     response and the chunked transfer coding is not the final encoding, the
1769     message body length is determined by reading the connection until it is
1770     closed by the server.
1771     If a <x:ref>Transfer-Encoding</x:ref> header field is present in a request and the
1772     chunked transfer coding is not the final encoding, the message body
1773     length cannot be determined reliably; the server &MUST; respond with
1774     the <x:ref>400 (Bad Request)</x:ref> status code and then close the connection.
1775    </t>
1776    <t>
1777     If a message is received with both a <x:ref>Transfer-Encoding</x:ref>
1778     and a <x:ref>Content-Length</x:ref> header field, the Transfer-Encoding
1779     overrides the Content-Length. Such a message might indicate an attempt to
1780     perform request smuggling (<xref target="request.smuggling"/>) or
1781     response splitting (<xref target="response.splitting"/>) and ought to be
1782     handled as an error. A sender &MUST; remove the received Content-Length
1783     field prior to forwarding such a message downstream.
1784    </t></x:lt>
1785    <x:lt><t>
1786     If a message is received without <x:ref>Transfer-Encoding</x:ref> and with
1787     either multiple <x:ref>Content-Length</x:ref> header fields having
1788     differing field-values or a single Content-Length header field having an
1789     invalid value, then the message framing is invalid and
1790     the recipient &MUST; treat it as an unrecoverable error.
1791     If this is a request message, the server &MUST; respond with
1792     a <x:ref>400 (Bad Request)</x:ref> status code and then close the connection.
1793     If this is a response message received by a proxy,
1794     the proxy &MUST; close the connection to the server, discard the received
1795     response, and send a <x:ref>502 (Bad Gateway)</x:ref> response to the
1796     client.
1797     If this is a response message received by a user agent,
1798     the user agent &MUST; close the connection to the server and discard the
1799     received response.
1800    </t></x:lt>
1801    <x:lt><t>
1802     If a valid <x:ref>Content-Length</x:ref> header field is present without
1803     <x:ref>Transfer-Encoding</x:ref>, its decimal value defines the
1804     expected message body length in octets.
1805     If the sender closes the connection or the recipient times out before the
1806     indicated number of octets are received, the recipient &MUST; consider
1807     the message to be incomplete and close the connection.
1808    </t></x:lt>
1809    <x:lt><t>
1810     If this is a request message and none of the above are true, then the
1811     message body length is zero (no message body is present).
1812    </t></x:lt>
1813    <x:lt><t>
1814     Otherwise, this is a response message without a declared message body
1815     length, so the message body length is determined by the number of octets
1816     received prior to the server closing the connection.
1817    </t></x:lt>
1818  </list>
1819</t>
1820<t>
1821   Since there is no way to distinguish a successfully completed,
1822   close-delimited message from a partially-received message interrupted
1823   by network failure, a server &SHOULD; generate encoding or
1824   length-delimited messages whenever possible.  The close-delimiting
1825   feature exists primarily for backwards compatibility with HTTP/1.0.
1826</t>
1827<t>
1828   A server &MAY; reject a request that contains a message body but
1829   not a <x:ref>Content-Length</x:ref> by responding with
1830   <x:ref>411 (Length Required)</x:ref>.
1831</t>
1832<t>
1833   Unless a transfer coding other than chunked has been applied,
1834   a client that sends a request containing a message body &SHOULD;
1835   use a valid <x:ref>Content-Length</x:ref> header field if the message body
1836   length is known in advance, rather than the chunked transfer coding, since some
1837   existing services respond to chunked with a <x:ref>411 (Length Required)</x:ref>
1838   status code even though they understand the chunked transfer coding.  This
1839   is typically because such services are implemented via a gateway that
1840   requires a content-length in advance of being called and the server
1841   is unable or unwilling to buffer the entire request before processing.
1842</t>
1843<t>
1844   A user agent that sends a request containing a message body &MUST; send a
1845   valid <x:ref>Content-Length</x:ref> header field if it does not know the
1846   server will handle HTTP/1.1 (or later) requests; such knowledge can be in
1847   the form of specific user configuration or by remembering the version of a
1848   prior received response.
1849</t>
1850<t>
1851   If the final response to the last request on a connection has been
1852   completely received and there remains additional data to read, a user agent
1853   &MAY; discard the remaining data or attempt to determine if that data
1854   belongs as part of the prior response body, which might be the case if the
1855   prior message's Content-Length value is incorrect. A client &MUST-NOT;
1856   process, cache, or forward such extra data as a separate response, since
1857   such behavior would be vulnerable to cache poisoning.
1858</t>
1859</section>
1860</section>
1861
1862<section anchor="incomplete.messages" title="Handling Incomplete Messages">
1863<t>
1864   A server that receives an incomplete request message, usually due to a
1865   canceled request or a triggered time-out exception, &MAY; send an error
1866   response prior to closing the connection.
1867</t>
1868<t>
1869   A client that receives an incomplete response message, which can occur
1870   when a connection is closed prematurely or when decoding a supposedly
1871   chunked transfer coding fails, &MUST; record the message as incomplete.
1872   Cache requirements for incomplete responses are defined in
1873   &cache-incomplete;.
1874</t>
1875<t>
1876   If a response terminates in the middle of the header section (before the
1877   empty line is received) and the status code might rely on header fields to
1878   convey the full meaning of the response, then the client cannot assume
1879   that meaning has been conveyed; the client might need to repeat the
1880   request in order to determine what action to take next.
1881</t>
1882<t>
1883   A message body that uses the chunked transfer coding is
1884   incomplete if the zero-sized chunk that terminates the encoding has not
1885   been received.  A message that uses a valid <x:ref>Content-Length</x:ref> is
1886   incomplete if the size of the message body received (in octets) is less than
1887   the value given by Content-Length.  A response that has neither chunked
1888   transfer coding nor Content-Length is terminated by closure of the
1889   connection, and thus is considered complete regardless of the number of
1890   message body octets received, provided that the header section was received
1891   intact.
1892</t>
1893</section>
1894
1895<section title="Message Parsing Robustness" anchor="message.robustness">
1896<t>
1897   Older HTTP/1.0 user agent implementations might send an extra CRLF
1898   after a POST request as a workaround for some early server
1899   applications that failed to read message body content that was
1900   not terminated by a line-ending. An HTTP/1.1 user agent &MUST-NOT;
1901   preface or follow a request with an extra CRLF.  If terminating
1902   the request message body with a line-ending is desired, then the
1903   user agent &MUST; count the terminating CRLF octets as part of the
1904   message body length.
1905</t>
1906<t>
1907   In the interest of robustness, a server that is expecting to receive and
1908   parse a request-line &SHOULD; ignore at least one empty line (CRLF)
1909   received prior to the request-line.
1910</t>
1911<t>
1912   Although the line terminator for the start-line and header
1913   fields is the sequence CRLF, a recipient &MAY; recognize a
1914   single LF as a line terminator and ignore any preceding CR.
1915</t>
1916<t>
1917   Although the request-line and status-line grammar rules require that each
1918   of the component elements be separated by a single SP octet, recipients
1919   &MAY; instead parse on whitespace-delimited word boundaries and, aside
1920   from the CRLF terminator, treat any form of whitespace as the SP separator
1921   while ignoring preceding or trailing whitespace;
1922   such whitespace includes one or more of the following octets:
1923   SP, HTAB, VT (%x0B), FF (%x0C), or bare CR.
1924   However, lenient parsing can result in security vulnerabilities if there
1925   are multiple recipients of the message and each has its own unique
1926   interpretation of robustness (see <xref target="request.smuggling"/>).
1927</t>
1928<t>
1929   When a server listening only for HTTP request messages, or processing
1930   what appears from the start-line to be an HTTP request message,
1931   receives a sequence of octets that does not match the HTTP-message
1932   grammar aside from the robustness exceptions listed above, the
1933   server &SHOULD; respond with a <x:ref>400 (Bad Request)</x:ref> response. 
1934</t>
1935</section>
1936</section>
1937
1938<section title="Transfer Codings" anchor="transfer.codings">
1939  <x:anchor-alias value="transfer-coding"/>
1940  <x:anchor-alias value="transfer-extension"/>
1941<t>
1942   Transfer coding names are used to indicate an encoding
1943   transformation that has been, can be, or might need to be applied to a
1944   payload body in order to ensure "safe transport" through the network.
1945   This differs from a content coding in that the transfer coding is a
1946   property of the message rather than a property of the representation
1947   that is being transferred.
1948</t>
1949<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="transfer-coding"/><iref primary="true" item="Grammar" subitem="transfer-extension"/>
1950  <x:ref>transfer-coding</x:ref>    = "chunked" ; <xref target="chunked.encoding"/>
1951                     / "compress" ; <xref target="compress.coding"/>
1952                     / "deflate" ; <xref target="deflate.coding"/>
1953                     / "gzip" ; <xref target="gzip.coding"/>
1954                     / <x:ref>transfer-extension</x:ref>
1955  <x:ref>transfer-extension</x:ref> = <x:ref>token</x:ref> *( <x:ref>OWS</x:ref> ";" <x:ref>OWS</x:ref> <x:ref>transfer-parameter</x:ref> )
1956</artwork></figure>
1957<t anchor="rule.parameter">
1958  <x:anchor-alias value="transfer-parameter"/>
1959   Parameters are in the form of a name or name=value pair.
1960</t>
1961<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="transfer-parameter"/>
1962  <x:ref>transfer-parameter</x:ref> = <x:ref>token</x:ref> <x:ref>BWS</x:ref> "=" <x:ref>BWS</x:ref> ( <x:ref>token</x:ref> / <x:ref>quoted-string</x:ref> )
1963</artwork></figure>
1964<t>
1965   All transfer-coding names are case-insensitive and ought to be registered
1966   within the HTTP Transfer Coding registry, as defined in
1967   <xref target="transfer.coding.registry"/>.
1968   They are used in the <x:ref>TE</x:ref> (<xref target="header.te"/>) and
1969   <x:ref>Transfer-Encoding</x:ref> (<xref target="header.transfer-encoding"/>)
1970   header fields.
1971</t>
1972
1973<section title="Chunked Transfer Coding" anchor="chunked.encoding">
1974  <iref primary="true" item="chunked (Coding Format)"/>
1975  <x:anchor-alias value="chunk"/>
1976  <x:anchor-alias value="chunked-body"/>
1977  <x:anchor-alias value="chunk-data"/>
1978  <x:anchor-alias value="chunk-size"/>
1979  <x:anchor-alias value="last-chunk"/>
1980<t>
1981   The chunked transfer coding wraps the payload body in order to transfer it
1982   as a series of chunks, each with its own size indicator, followed by an
1983   &OPTIONAL; trailer containing header fields. Chunked enables content
1984   streams of unknown size to be transferred as a sequence of length-delimited
1985   buffers, which enables the sender to retain connection persistence and the
1986   recipient to know when it has received the entire message.
1987</t>
1988<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="chunked-body"><!--terminal production--></iref><iref primary="true" item="Grammar" subitem="chunk"/><iref primary="true" item="Grammar" subitem="chunk-size"/><iref primary="true" item="Grammar" subitem="last-chunk"/><iref primary="false" item="Grammar" subitem="trailer-part"/><iref primary="false" item="Grammar" subitem="chunk-ext"/><iref primary="true" item="Grammar" subitem="chunk-data"/>
1989  <x:ref>chunked-body</x:ref>   = *<x:ref>chunk</x:ref>
1990                   <x:ref>last-chunk</x:ref>
1991                   <x:ref>trailer-part</x:ref>
1992                   <x:ref>CRLF</x:ref>
1993 
1994  <x:ref>chunk</x:ref>          = <x:ref>chunk-size</x:ref> [ <x:ref>chunk-ext</x:ref> ] <x:ref>CRLF</x:ref>
1995                   <x:ref>chunk-data</x:ref> <x:ref>CRLF</x:ref>
1996  <x:ref>chunk-size</x:ref>     = 1*<x:ref>HEXDIG</x:ref>
1997  <x:ref>last-chunk</x:ref>     = 1*("0") [ <x:ref>chunk-ext</x:ref> ] <x:ref>CRLF</x:ref>
1998 
1999  <x:ref>chunk-data</x:ref>     = 1*<x:ref>OCTET</x:ref> ; a sequence of chunk-size octets
2000</artwork></figure>
2001<t>
2002   The chunk-size field is a string of hex digits indicating the size of
2003   the chunk-data in octets. The chunked transfer coding is complete when a
2004   chunk with a chunk-size of zero is received, possibly followed by a
2005   trailer, and finally terminated by an empty line.
2006</t>
2007<t>
2008   A recipient &MUST; be able to parse and decode the chunked transfer coding.
2009</t>
2010
2011<section title="Chunk Extensions" anchor="chunked.extension">
2012  <x:anchor-alias value="chunk-ext"/>
2013  <x:anchor-alias value="chunk-ext-name"/>
2014  <x:anchor-alias value="chunk-ext-val"/>
2015<t>
2016   The chunked encoding allows each chunk to include zero or more chunk
2017   extensions, immediately following the <x:ref>chunk-size</x:ref>, for the
2018   sake of supplying per-chunk metadata (such as a signature or hash),
2019   mid-message control information, or randomization of message body size.
2020</t>
2021<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="chunked-body"><!--terminal production--></iref><iref primary="true" item="Grammar" subitem="chunk-ext"/><iref primary="true" item="Grammar" subitem="chunk-ext-name"/><iref primary="true" item="Grammar" subitem="chunk-ext-val"/>
2022  <x:ref>chunk-ext</x:ref>      = *( ";" <x:ref>chunk-ext-name</x:ref> [ "=" <x:ref>chunk-ext-val</x:ref> ] )
2023
2024  <x:ref>chunk-ext-name</x:ref> = <x:ref>token</x:ref>
2025  <x:ref>chunk-ext-val</x:ref>  = <x:ref>token</x:ref> / <x:ref>quoted-string</x:ref>
2026</artwork></figure>
2027<t>
2028   The chunked encoding is specific to each connection and is likely to be
2029   removed or recoded by each recipient (including intermediaries) before any
2030   higher-level application would have a chance to inspect the extensions.
2031   Hence, use of chunk extensions is generally limited to specialized HTTP
2032   services such as "long polling" (where client and server can have shared
2033   expectations regarding the use of chunk extensions) or for padding within
2034   an end-to-end secured connection.
2035</t>
2036<t>
2037   A recipient &MUST; ignore unrecognized chunk extensions.
2038   A server ought to limit the total length of chunk extensions received in a
2039   request to an amount reasonable for the services provided, in the same way
2040   that it applies length limitations and timeouts for other parts of a
2041   message, and generate an appropriate <x:ref>4xx (Client Error)</x:ref>
2042   response if that amount is exceeded.
2043</t>
2044</section>
2045
2046<section title="Chunked Trailer Part" anchor="chunked.trailer.part">
2047  <x:anchor-alias value="trailer-part"/>
2048<t>
2049   A trailer allows the sender to include additional fields at the end of a
2050   chunked message in order to supply metadata that might be dynamically
2051   generated while the message body is sent, such as a message integrity
2052   check, digital signature, or post-processing status. The trailer fields are
2053   identical to header fields, except they are sent in a chunked trailer
2054   instead of the message's header section.
2055</t>
2056<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="trailer-part"/><iref primary="false" item="Grammar" subitem="header-field"/>
2057  <x:ref>trailer-part</x:ref>   = *( <x:ref>header-field</x:ref> <x:ref>CRLF</x:ref> )
2058</artwork></figure>
2059<t>
2060   A sender &MUST-NOT; generate a trailer that contains a field necessary for
2061   message framing (e.g., <x:ref>Transfer-Encoding</x:ref> and
2062   <x:ref>Content-Length</x:ref>), routing (e.g., <x:ref>Host</x:ref>),
2063   request modifiers (e.g., controls and conditionals in
2064   &request-header-fields;), authentication (e.g., see <xref target="Part7"/>
2065   and <xref target="RFC6265"/>), response control data (e.g., see
2066   &response-control-data;), or determining how to process the payload
2067   (e.g., <x:ref>Content-Encoding</x:ref>, <x:ref>Content-Type</x:ref>,
2068   <x:ref>Content-Range</x:ref>, and <x:ref>Trailer</x:ref>).
2069</t>
2070<t>
2071   When a chunked message containing a non-empty trailer is received, the
2072   recipient &MAY; process the fields (aside from those forbidden above)
2073   as if they were appended to the message's header section.
2074   A recipient &MUST; ignore (or consider as an error) any fields that are
2075   forbidden to be sent in a trailer, since processing them as if they were
2076   present in the header section might bypass external security filters.
2077</t>
2078<t>
2079   Unless the request includes a <x:ref>TE</x:ref> header field indicating
2080   "trailers" is acceptable, as described in <xref target="header.te"/>, a
2081   server &SHOULD-NOT; generate trailer fields that it believes are necessary
2082   for the user agent to receive. Without a TE containing "trailers", the
2083   server ought to assume that the trailer fields might be silently discarded
2084   along the path to the user agent. This requirement allows intermediaries to
2085   forward a de-chunked message to an HTTP/1.0 recipient without buffering the
2086   entire response.
2087</t>
2088</section>
2089
2090<section title="Decoding Chunked" anchor="decoding.chunked">
2091<t>
2092   A process for decoding the chunked transfer coding
2093   can be represented in pseudo-code as:
2094</t>
2095<figure><artwork type="code">
2096  length := 0
2097  read chunk-size, chunk-ext (if any), and CRLF
2098  while (chunk-size &gt; 0) {
2099     read chunk-data and CRLF
2100     append chunk-data to decoded-body
2101     length := length + chunk-size
2102     read chunk-size, chunk-ext (if any), and CRLF
2103  }
2104  read trailer field
2105  while (trailer field is not empty) {
2106     if (trailer field is allowed to be sent in a trailer) {
2107         append trailer field to existing header fields
2108     }
2109     read trailer-field
2110  }
2111  Content-Length := length
2112  Remove "chunked" from Transfer-Encoding
2113  Remove Trailer from existing header fields
2114</artwork></figure>
2115</section>
2116</section>
2117
2118<section title="Compression Codings" anchor="compression.codings">
2119<t>
2120   The codings defined below can be used to compress the payload of a
2121   message.
2122</t>
2123
2124<section title="Compress Coding" anchor="compress.coding">
2125<iref item="compress (Coding Format)"/>
2126<t>
2127   The "compress" coding is an adaptive Lempel-Ziv-Welch (LZW) coding
2128   <xref target="Welch"/> that is commonly produced by the UNIX file
2129   compression program "compress".
2130   A recipient &SHOULD; consider "x-compress" to be equivalent to "compress".
2131</t>
2132</section>
2133
2134<section title="Deflate Coding" anchor="deflate.coding">
2135<iref item="deflate (Coding Format)"/>
2136<t>
2137   The "deflate" coding is a "zlib" data format <xref target="RFC1950"/>
2138   containing a "deflate" compressed data stream <xref target="RFC1951"/>
2139   that uses a combination of the Lempel-Ziv (LZ77) compression algorithm and
2140   Huffman coding.
2141</t>
2142<x:note>
2143  <t>
2144    &Note; Some non-conformant implementations send the "deflate"
2145    compressed data without the zlib wrapper.
2146   </t>
2147</x:note>
2148</section>
2149
2150<section title="Gzip Coding" anchor="gzip.coding">
2151<iref item="gzip (Coding Format)"/>
2152<t>
2153   The "gzip" coding is an LZ77 coding with a 32 bit CRC that is commonly
2154   produced by the gzip file compression program <xref target="RFC1952"/>.
2155   A recipient &SHOULD; consider "x-gzip" to be equivalent to "gzip".
2156</t>
2157</section>
2158
2159</section>
2160
2161<section title="TE" anchor="header.te">
2162  <iref primary="true" item="TE header field" x:for-anchor=""/>
2163  <x:anchor-alias value="TE"/>
2164  <x:anchor-alias value="t-codings"/>
2165  <x:anchor-alias value="t-ranking"/>
2166  <x:anchor-alias value="rank"/>
2167<t>
2168   The "TE" header field in a request indicates what transfer codings,
2169   besides chunked, the client is willing to accept in response, and
2170   whether or not the client is willing to accept trailer fields in a
2171   chunked transfer coding.
2172</t>
2173<t>
2174   The TE field-value consists of a comma-separated list of transfer coding
2175   names, each allowing for optional parameters (as described in
2176   <xref target="transfer.codings"/>), and/or the keyword "trailers".
2177   A client &MUST-NOT; send the chunked transfer coding name in TE;
2178   chunked is always acceptable for HTTP/1.1 recipients.
2179</t>
2180<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="TE"/><iref primary="true" item="Grammar" subitem="t-codings"/><iref primary="true" item="Grammar" subitem="t-ranking"/><iref primary="true" item="Grammar" subitem="rank"/>
2181  <x:ref>TE</x:ref>        = #<x:ref>t-codings</x:ref>
2182  <x:ref>t-codings</x:ref> = "trailers" / ( <x:ref>transfer-coding</x:ref> [ <x:ref>t-ranking</x:ref> ] )
2183  <x:ref>t-ranking</x:ref> = <x:ref>OWS</x:ref> ";" <x:ref>OWS</x:ref> "q=" <x:ref>rank</x:ref>
2184  <x:ref>rank</x:ref>      = ( "0" [ "." 0*3<x:ref>DIGIT</x:ref> ] )
2185             / ( "1" [ "." 0*3("0") ] )
2186</artwork></figure>
2187<t>
2188   Three examples of TE use are below.
2189</t>
2190<figure><artwork type="example">
2191  TE: deflate
2192  TE:
2193  TE: trailers, deflate;q=0.5
2194</artwork></figure>
2195<t>
2196   The presence of the keyword "trailers" indicates that the client is willing
2197   to accept trailer fields in a chunked transfer coding, as defined in
2198   <xref target="chunked.trailer.part"/>, on behalf of itself and any downstream
2199   clients. For requests from an intermediary, this implies that either:
2200   (a) all downstream clients are willing to accept trailer fields in the
2201   forwarded response; or,
2202   (b) the intermediary will attempt to buffer the response on behalf of
2203   downstream recipients.
2204   Note that HTTP/1.1 does not define any means to limit the size of a
2205   chunked response such that an intermediary can be assured of buffering the
2206   entire response.
2207</t>
2208<t>
2209   When multiple transfer codings are acceptable, the client &MAY; rank the
2210   codings by preference using a case-insensitive "q" parameter (similar to
2211   the qvalues used in content negotiation fields, &qvalue;). The rank value
2212   is a real number in the range 0 through 1, where 0.001 is the least
2213   preferred and 1 is the most preferred; a value of 0 means "not acceptable".
2214</t>
2215<t>
2216   If the TE field-value is empty or if no TE field is present, the only
2217   acceptable transfer coding is chunked. A message with no transfer coding
2218   is always acceptable.
2219</t>
2220<t>
2221   Since the TE header field only applies to the immediate connection,
2222   a sender of TE &MUST; also send a "TE" connection option within the
2223   <x:ref>Connection</x:ref> header field (<xref target="header.connection"/>)
2224   in order to prevent the TE field from being forwarded by intermediaries
2225   that do not support its semantics.
2226</t>
2227</section>
2228
2229<section title="Trailer" anchor="header.trailer">
2230  <iref primary="true" item="Trailer header field" x:for-anchor=""/>
2231  <x:anchor-alias value="Trailer"/>
2232<t>
2233   When a message includes a message body encoded with the chunked
2234   transfer coding and the sender desires to send metadata in the form of
2235   trailer fields at the end of the message, the sender &SHOULD; generate a
2236   <x:ref>Trailer</x:ref> header field before the message body to indicate
2237   which fields will be present in the trailers. This allows the recipient
2238   to prepare for receipt of that metadata before it starts processing the body,
2239   which is useful if the message is being streamed and the recipient wishes
2240   to confirm an integrity check on the fly.
2241</t>
2242<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Trailer"/><iref primary="false" item="Grammar" subitem="field-name"/>
2243  <x:ref>Trailer</x:ref> = 1#<x:ref>field-name</x:ref>
2244</artwork></figure>
2245</section>
2246</section>
2247
2248<section title="Message Routing" anchor="message.routing">
2249<t>
2250   HTTP request message routing is determined by each client based on the
2251   target resource, the client's proxy configuration, and
2252   establishment or reuse of an inbound connection.  The corresponding
2253   response routing follows the same connection chain back to the client.
2254</t>
2255
2256<section title="Identifying a Target Resource" anchor="target-resource">
2257  <iref primary="true" item="target resource"/>
2258  <iref primary="true" item="target URI"/>
2259  <x:anchor-alias value="target resource"/>
2260  <x:anchor-alias value="target URI"/>
2261<t>
2262   HTTP is used in a wide variety of applications, ranging from
2263   general-purpose computers to home appliances.  In some cases,
2264   communication options are hard-coded in a client's configuration.
2265   However, most HTTP clients rely on the same resource identification
2266   mechanism and configuration techniques as general-purpose Web browsers.
2267</t>
2268<t>
2269   HTTP communication is initiated by a user agent for some purpose.
2270   The purpose is a combination of request semantics, which are defined in
2271   <xref target="Part2"/>, and a target resource upon which to apply those
2272   semantics.  A URI reference (<xref target="uri"/>) is typically used as
2273   an identifier for the "<x:dfn>target resource</x:dfn>", which a user agent
2274   would resolve to its absolute form in order to obtain the
2275   "<x:dfn>target URI</x:dfn>".  The target URI
2276   excludes the reference's fragment component, if any,
2277   since fragment identifiers are reserved for client-side processing
2278   (<xref target="RFC3986" x:fmt="," x:sec="3.5"/>).
2279</t>
2280</section>
2281
2282<section title="Connecting Inbound" anchor="connecting.inbound">
2283<t>
2284   Once the target URI is determined, a client needs to decide whether
2285   a network request is necessary to accomplish the desired semantics and,
2286   if so, where that request is to be directed.
2287</t>
2288<t>
2289   If the client has a cache <xref target="Part6"/> and the request can be
2290   satisfied by it, then the request is
2291   usually directed there first.
2292</t>
2293<t>
2294   If the request is not satisfied by a cache, then a typical client will
2295   check its configuration to determine whether a proxy is to be used to
2296   satisfy the request.  Proxy configuration is implementation-dependent,
2297   but is often based on URI prefix matching, selective authority matching,
2298   or both, and the proxy itself is usually identified by an "http" or
2299   "https" URI.  If a proxy is applicable, the client connects inbound by
2300   establishing (or reusing) a connection to that proxy.
2301</t>
2302<t>
2303   If no proxy is applicable, a typical client will invoke a handler routine,
2304   usually specific to the target URI's scheme, to connect directly
2305   to an authority for the target resource.  How that is accomplished is
2306   dependent on the target URI scheme and defined by its associated
2307   specification, similar to how this specification defines origin server
2308   access for resolution of the "http" (<xref target="http.uri"/>) and
2309   "https" (<xref target="https.uri"/>) schemes.
2310</t>
2311<t>
2312   HTTP requirements regarding connection management are defined in
2313   <xref target="connection.management"/>.
2314</t>
2315</section>
2316
2317<section title="Request Target" anchor="request-target">
2318<t>
2319   Once an inbound connection is obtained,
2320   the client sends an HTTP request message (<xref target="http.message"/>)
2321   with a request-target derived from the target URI.
2322   There are four distinct formats for the request-target, depending on both
2323   the method being requested and whether the request is to a proxy.
2324</t>   
2325<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="request-target"/><iref primary="false" item="Grammar" subitem="origin-form"/><iref primary="false" item="Grammar" subitem="absolute-form"/><iref primary="false" item="Grammar" subitem="authority-form"/><iref primary="false" item="Grammar" subitem="asterisk-form"/>
2326  <x:ref>request-target</x:ref> = <x:ref>origin-form</x:ref>
2327                 / <x:ref>absolute-form</x:ref>
2328                 / <x:ref>authority-form</x:ref>
2329                 / <x:ref>asterisk-form</x:ref>
2330</artwork></figure>
2331
2332<section title="origin-form" anchor="origin-form">
2333   <iref item="origin-form (of request-target)"/>
2334<t>
2335   The most common form of request-target is the <x:dfn>origin-form</x:dfn>.
2336</t>
2337<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="origin-form"/>
2338  <x:ref>origin-form</x:ref>    = <x:ref>absolute-path</x:ref> [ "?" <x:ref>query</x:ref> ]
2339</artwork></figure>
2340<t>
2341   When making a request directly to an origin server, other than a CONNECT
2342   or server-wide OPTIONS request (as detailed below),
2343   a client &MUST; send only the absolute path and query components of
2344   the target URI as the request-target.
2345   If the target URI's path component is empty, the client &MUST; send
2346   "/" as the path within the origin-form of request-target.
2347   A <x:ref>Host</x:ref> header field is also sent, as defined in
2348   <xref target="header.host"/>.
2349</t>
2350<t>
2351   For example, a client wishing to retrieve a representation of the resource
2352   identified as
2353</t>
2354<figure><artwork x:indent-with="  " type="example">
2355http://www.example.org/where?q=now
2356</artwork></figure>
2357<t>
2358   directly from the origin server would open (or reuse) a TCP connection
2359   to port 80 of the host "www.example.org" and send the lines:
2360</t>
2361<figure><artwork type="message/http; msgtype=&#34;request&#34;" x:indent-with="  ">
2362GET /where?q=now HTTP/1.1
2363Host: www.example.org
2364</artwork></figure>
2365<t>
2366   followed by the remainder of the request message.
2367</t>
2368</section>
2369
2370<section title="absolute-form" anchor="absolute-form">
2371   <iref item="absolute-form (of request-target)"/>
2372<t>
2373   When making a request to a proxy, other than a CONNECT or server-wide
2374   OPTIONS request (as detailed below), a client &MUST; send the target URI
2375   in <x:dfn>absolute-form</x:dfn> as the request-target.
2376</t>
2377<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="absolute-form"/>
2378  <x:ref>absolute-form</x:ref>  = <x:ref>absolute-URI</x:ref>
2379</artwork></figure>
2380<t>
2381   The proxy is requested to either service that request from a valid cache,
2382   if possible, or make the same request on the client's behalf to either
2383   the next inbound proxy server or directly to the origin server indicated
2384   by the request-target.  Requirements on such "forwarding" of messages are
2385   defined in <xref target="message.forwarding"/>.
2386</t>
2387<t>
2388   An example absolute-form of request-line would be:
2389</t>
2390<figure><artwork type="message/http; msgtype=&#34;request&#34;" x:indent-with="  ">
2391GET http://www.example.org/pub/WWW/TheProject.html HTTP/1.1
2392</artwork></figure>
2393<t>
2394   To allow for transition to the absolute-form for all requests in some
2395   future version of HTTP, a server &MUST; accept the absolute-form
2396   in requests, even though HTTP/1.1 clients will only send them in requests
2397   to proxies.
2398</t>
2399</section>
2400
2401<section title="authority-form" anchor="authority-form">
2402   <iref item="authority-form (of request-target)"/>
2403<t>
2404   The <x:dfn>authority-form</x:dfn> of request-target is only used for
2405   CONNECT requests (&CONNECT;).
2406</t>
2407<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="authority-form"/>
2408  <x:ref>authority-form</x:ref> = <x:ref>authority</x:ref>
2409</artwork></figure>
2410<t>
2411   When making a CONNECT request to establish a
2412   tunnel through one or more proxies, a client &MUST; send only the target
2413   URI's authority component (excluding any userinfo and its "@" delimiter) as
2414   the request-target. For example,
2415</t>
2416<figure><artwork type="message/http; msgtype=&#34;request&#34;" x:indent-with="  ">
2417CONNECT www.example.com:80 HTTP/1.1
2418</artwork></figure>
2419</section>
2420
2421<section title="asterisk-form" anchor="asterisk-form">
2422   <iref item="asterisk-form (of request-target)"/>
2423<t>
2424   The <x:dfn>asterisk-form</x:dfn> of request-target is only used for a server-wide
2425   OPTIONS request (&OPTIONS;).
2426</t>
2427<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="asterisk-form"/>
2428  <x:ref>asterisk-form</x:ref>  = "*"
2429</artwork></figure>
2430<t>
2431   When a client wishes to request OPTIONS
2432   for the server as a whole, as opposed to a specific named resource of
2433   that server, the client &MUST; send only "*" (%x2A) as the request-target.
2434   For example,
2435</t>
2436<figure><artwork type="message/http; msgtype=&#34;request&#34;" x:indent-with="  ">
2437OPTIONS * HTTP/1.1
2438</artwork></figure>
2439<t>
2440   If a proxy receives an OPTIONS request with an absolute-form of
2441   request-target in which the URI has an empty path and no query component,
2442   then the last proxy on the request chain &MUST; send a request-target
2443   of "*" when it forwards the request to the indicated origin server.
2444</t>
2445<figure><preamble>  
2446   For example, the request
2447</preamble><artwork type="message/http; msgtype=&#34;request&#34;" x:indent-with="  ">
2448OPTIONS http://www.example.org:8001 HTTP/1.1
2449</artwork></figure>
2450<figure><preamble>  
2451  would be forwarded by the final proxy as
2452</preamble><artwork type="message/http; msgtype=&#34;request&#34;" x:indent-with="  ">
2453OPTIONS * HTTP/1.1
2454Host: www.example.org:8001
2455</artwork>
2456<postamble>
2457   after connecting to port 8001 of host "www.example.org".
2458</postamble>
2459</figure>
2460</section>
2461</section>
2462
2463<section title="Host" anchor="header.host">
2464  <iref primary="true" item="Host header field" x:for-anchor=""/>
2465  <x:anchor-alias value="Host"/>
2466<t>
2467   The "Host" header field in a request provides the host and port
2468   information from the target URI, enabling the origin
2469   server to distinguish among resources while servicing requests
2470   for multiple host names on a single IP address.
2471</t>
2472<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Host"/>
2473  <x:ref>Host</x:ref> = <x:ref>uri-host</x:ref> [ ":" <x:ref>port</x:ref> ] ; <xref target="http.uri"/>
2474</artwork></figure>
2475<t>
2476   A client &MUST; send a Host header field in all HTTP/1.1 request messages.
2477   If the target URI includes an authority component, then a client &MUST;
2478   send a field-value for Host that is identical to that authority
2479   component, excluding any userinfo subcomponent and its "@" delimiter
2480   (<xref target="http.uri"/>).
2481   If the authority component is missing or undefined for the target URI,
2482   then a client &MUST; send a Host header field with an empty field-value.
2483</t>
2484<t>
2485   Since the Host field-value is critical information for handling a request,
2486   a user agent &SHOULD; generate Host as the first header field following the
2487   request-line.
2488</t>
2489<t>
2490   For example, a GET request to the origin server for
2491   &lt;http://www.example.org/pub/WWW/&gt; would begin with:
2492</t>
2493<figure><artwork type="message/http; msgtype=&#34;request&#34;" x:indent-with="  ">
2494GET /pub/WWW/ HTTP/1.1
2495Host: www.example.org
2496</artwork></figure>
2497<t>
2498   A client &MUST; send a Host header field in an HTTP/1.1 request even
2499   if the request-target is in the absolute-form, since this
2500   allows the Host information to be forwarded through ancient HTTP/1.0
2501   proxies that might not have implemented Host.
2502</t>
2503<t>
2504   When a proxy receives a request with an absolute-form of
2505   request-target, the proxy &MUST; ignore the received
2506   Host header field (if any) and instead replace it with the host
2507   information of the request-target.  A proxy that forwards such a request
2508   &MUST; generate a new Host field-value based on the received
2509   request-target rather than forward the received Host field-value.
2510</t>
2511<t>
2512   Since the Host header field acts as an application-level routing
2513   mechanism, it is a frequent target for malware seeking to poison
2514   a shared cache or redirect a request to an unintended server.
2515   An interception proxy is particularly vulnerable if it relies on
2516   the Host field-value for redirecting requests to internal
2517   servers, or for use as a cache key in a shared cache, without
2518   first verifying that the intercepted connection is targeting a
2519   valid IP address for that host.
2520</t>
2521<t>
2522   A server &MUST; respond with a <x:ref>400 (Bad Request)</x:ref> status code
2523   to any HTTP/1.1 request message that lacks a Host header field and
2524   to any request message that contains more than one Host header field
2525   or a Host header field with an invalid field-value.
2526</t>
2527</section>
2528
2529<section title="Effective Request URI" anchor="effective.request.uri">
2530  <iref primary="true" item="effective request URI"/>
2531  <x:anchor-alias value="effective request URI"/>
2532<t>
2533   Since the request-target often contains only part of the user agent's
2534   target URI, a server reconstructs the intended target as an
2535   "<x:dfn>effective request URI</x:dfn>" to properly service the request.
2536   This reconstruction involves both the server's local configuration and
2537   information communicated in the <x:ref>request-target</x:ref>,
2538   <x:ref>Host</x:ref> header field, and connection context.
2539</t>
2540<t>
2541   For a user agent, the effective request URI is the target URI.
2542</t>
2543<t>
2544   If the <x:ref>request-target</x:ref> is in <x:ref>absolute-form</x:ref>,
2545   the effective request URI is the same as the request-target. Otherwise, the
2546   effective request URI is constructed as follows:
2547<list style="empty">
2548<t>
2549   If the server's configuration (or outbound gateway) provides a fixed URI
2550   <x:ref>scheme</x:ref>, that scheme is used for the effective request URI.
2551   Otherwise, if the request is received over a TLS-secured TCP connection,
2552   the effective request URI's scheme is "https"; if not, the scheme is "http".
2553</t>
2554<t>
2555   If the server's configuration (or outbound gateway) provides a fixed URI
2556   <x:ref>authority</x:ref> component, that authority is used for the
2557   effective request URI. If not, then if the request-target is in
2558   <x:ref>authority-form</x:ref>, the effective request URI's authority
2559   component is the same as the request-target.
2560   If not, then if a <x:ref>Host</x:ref> header field is supplied with a
2561   non-empty field-value, the authority component is the same as the
2562   Host field-value. Otherwise, the authority component is assigned
2563   the default name configured for the server and, if the connection's
2564   incoming TCP port number differs from the default port for the effective
2565   request URI's scheme, then a colon (":") and the incoming port number (in
2566   decimal form) are appended to the authority component.
2567</t>
2568<t>
2569   If the request-target is in <x:ref>authority-form</x:ref> or
2570   <x:ref>asterisk-form</x:ref>, the effective request URI's combined
2571   <x:ref>path</x:ref> and <x:ref>query</x:ref> component is empty. Otherwise,
2572   the combined <x:ref>path</x:ref> and <x:ref>query</x:ref> component is the
2573   same as the request-target.
2574</t>
2575<t>
2576   The components of the effective request URI, once determined as above, can
2577   be combined into <x:ref>absolute-URI</x:ref> form by concatenating the
2578   scheme, "://", authority, and combined path and query component.
2579</t>
2580</list>
2581</t>
2582<figure>
2583<preamble>
2584   Example 1: the following message received over an insecure TCP connection
2585</preamble>
2586<artwork type="example" x:indent-with="  ">
2587GET /pub/WWW/TheProject.html HTTP/1.1
2588Host: www.example.org:8080
2589</artwork>
2590</figure>
2591<figure>
2592<preamble>
2593  has an effective request URI of
2594</preamble>
2595<artwork type="example" x:indent-with="  ">
2596http://www.example.org:8080/pub/WWW/TheProject.html
2597</artwork>
2598</figure>
2599<figure>
2600<preamble>
2601   Example 2: the following message received over a TLS-secured TCP connection
2602</preamble>
2603<artwork type="example" x:indent-with="  ">
2604OPTIONS * HTTP/1.1
2605Host: www.example.org
2606</artwork>
2607</figure>
2608<figure>
2609<preamble>
2610  has an effective request URI of
2611</preamble>
2612<artwork type="example" x:indent-with="  ">
2613https://www.example.org
2614</artwork>
2615</figure>
2616<t>
2617   Recipients of an HTTP/1.0 request that lacks a <x:ref>Host</x:ref> header
2618   field might need to use heuristics (e.g., examination of the URI path for
2619   something unique to a particular host) in order to guess the
2620   effective request URI's authority component.
2621</t>
2622<t>
2623   Once the effective request URI has been constructed, an origin server needs
2624   to decide whether or not to provide service for that URI via the connection
2625   in which the request was received. For example, the request might have been
2626   misdirected, deliberately or accidentally, such that the information within
2627   a received <x:ref>request-target</x:ref> or <x:ref>Host</x:ref> header
2628   field differs from the host or port upon which the connection has been
2629   made. If the connection is from a trusted gateway, that inconsistency might
2630   be expected; otherwise, it might indicate an attempt to bypass security
2631   filters, trick the server into delivering non-public content, or poison a
2632   cache. See <xref target="security.considerations"/> for security
2633   considerations regarding message routing.
2634</t>
2635</section>
2636
2637<section title="Associating a Response to a Request" anchor="associating.response.to.request">
2638<t>
2639   HTTP does not include a request identifier for associating a given
2640   request message with its corresponding one or more response messages.
2641   Hence, it relies on the order of response arrival to correspond exactly
2642   to the order in which requests are made on the same connection.
2643   More than one response message per request only occurs when one or more
2644   informational responses (<x:ref>1xx</x:ref>, see &status-1xx;) precede a
2645   final response to the same request.
2646</t>
2647<t>
2648   A client that has more than one outstanding request on a connection &MUST;
2649   maintain a list of outstanding requests in the order sent and &MUST;
2650   associate each received response message on that connection to the highest
2651   ordered request that has not yet received a final (non-<x:ref>1xx</x:ref>)
2652   response.
2653</t>
2654</section>
2655
2656<section title="Message Forwarding" anchor="message.forwarding">
2657<t>
2658   As described in <xref target="intermediaries"/>, intermediaries can serve
2659   a variety of roles in the processing of HTTP requests and responses.
2660   Some intermediaries are used to improve performance or availability.
2661   Others are used for access control or to filter content.
2662   Since an HTTP stream has characteristics similar to a pipe-and-filter
2663   architecture, there are no inherent limits to the extent an intermediary
2664   can enhance (or interfere) with either direction of the stream.
2665</t>
2666<t>
2667   An intermediary not acting as a tunnel &MUST; implement the
2668   <x:ref>Connection</x:ref> header field, as specified in
2669   <xref target="header.connection"/>, and exclude fields from being forwarded
2670   that are only intended for the incoming connection.
2671</t>
2672<t>
2673   An intermediary &MUST-NOT; forward a message to itself unless it is
2674   protected from an infinite request loop. In general, an intermediary ought
2675   to recognize its own server names, including any aliases, local variations,
2676   or literal IP addresses, and respond to such requests directly.
2677</t>
2678
2679<section title="Via" anchor="header.via">
2680  <iref primary="true" item="Via header field" x:for-anchor=""/>
2681  <x:anchor-alias value="pseudonym"/>
2682  <x:anchor-alias value="received-by"/>
2683  <x:anchor-alias value="received-protocol"/>
2684  <x:anchor-alias value="Via"/>
2685<t>
2686   The "Via" header field indicates the presence of intermediate protocols and
2687   recipients between the user agent and the server (on requests) or between
2688   the origin server and the client (on responses), similar to the
2689   "Received" header field in email
2690   (<xref target="RFC5322" x:fmt="of" x:sec="3.6.7"/>).
2691   Via can be used for tracking message forwards,
2692   avoiding request loops, and identifying the protocol capabilities of
2693   senders along the request/response chain.
2694</t>
2695<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Via"/><iref primary="true" item="Grammar" subitem="received-protocol"/><iref primary="true" item="Grammar" subitem="protocol-name"/><iref primary="true" item="Grammar" subitem="protocol-version"/><iref primary="true" item="Grammar" subitem="received-by"/><iref primary="true" item="Grammar" subitem="pseudonym"/>
2696  <x:ref>Via</x:ref> = 1#( <x:ref>received-protocol</x:ref> <x:ref>RWS</x:ref> <x:ref>received-by</x:ref> [ <x:ref>RWS</x:ref> <x:ref>comment</x:ref> ] )
2697
2698  <x:ref>received-protocol</x:ref> = [ <x:ref>protocol-name</x:ref> "/" ] <x:ref>protocol-version</x:ref>
2699                      ; see <xref target="header.upgrade"/>
2700  <x:ref>received-by</x:ref>       = ( <x:ref>uri-host</x:ref> [ ":" <x:ref>port</x:ref> ] ) / <x:ref>pseudonym</x:ref>
2701  <x:ref>pseudonym</x:ref>         = <x:ref>token</x:ref>
2702</artwork></figure>
2703<t>
2704   Multiple Via field values represent each proxy or gateway that has
2705   forwarded the message. Each intermediary appends its own information
2706   about how the message was received, such that the end result is ordered
2707   according to the sequence of forwarding recipients.
2708</t>
2709<t>
2710   A proxy &MUST; send an appropriate Via header field, as described below, in
2711   each message that it forwards.
2712   An HTTP-to-HTTP gateway &MUST; send an appropriate Via header field in
2713   each inbound request message and &MAY; send a Via header field in
2714   forwarded response messages.
2715</t>
2716<t>
2717   For each intermediary, the received-protocol indicates the protocol and
2718   protocol version used by the upstream sender of the message. Hence, the
2719   Via field value records the advertised protocol capabilities of the
2720   request/response chain such that they remain visible to downstream
2721   recipients; this can be useful for determining what backwards-incompatible
2722   features might be safe to use in response, or within a later request, as
2723   described in <xref target="http.version"/>. For brevity, the protocol-name
2724   is omitted when the received protocol is HTTP.
2725</t>
2726<t>
2727   The received-by portion of the field value is normally the host and optional
2728   port number of a recipient server or client that subsequently forwarded the
2729   message.
2730   However, if the real host is considered to be sensitive information, a
2731   sender &MAY; replace it with a pseudonym. If a port is not provided,
2732   a recipient &MAY; interpret that as meaning it was received on the default
2733   TCP port, if any, for the received-protocol.
2734</t>
2735<t>
2736   A sender &MAY; generate comments in the Via header field to identify the
2737   software of each recipient, analogous to the <x:ref>User-Agent</x:ref> and
2738   <x:ref>Server</x:ref> header fields. However, all comments in the Via field
2739   are optional and a recipient &MAY; remove them prior to forwarding the
2740   message.
2741</t>
2742<t>
2743   For example, a request message could be sent from an HTTP/1.0 user
2744   agent to an internal proxy code-named "fred", which uses HTTP/1.1 to
2745   forward the request to a public proxy at p.example.net, which completes
2746   the request by forwarding it to the origin server at www.example.com.
2747   The request received by www.example.com would then have the following
2748   Via header field:
2749</t>
2750<figure><artwork type="example">
2751  Via: 1.0 fred, 1.1 p.example.net
2752</artwork></figure>
2753<t>
2754   An intermediary used as a portal through a network firewall
2755   &SHOULD-NOT; forward the names and ports of hosts within the firewall
2756   region unless it is explicitly enabled to do so. If not enabled, such an
2757   intermediary &SHOULD; replace each received-by host of any host behind the
2758   firewall by an appropriate pseudonym for that host.
2759</t>
2760<t>
2761   An intermediary &MAY; combine an ordered subsequence of Via header
2762   field entries into a single such entry if the entries have identical
2763   received-protocol values. For example,
2764</t>
2765<figure><artwork type="example">
2766  Via: 1.0 ricky, 1.1 ethel, 1.1 fred, 1.0 lucy
2767</artwork></figure>
2768<t>
2769  could be collapsed to
2770</t>
2771<figure><artwork type="example">
2772  Via: 1.0 ricky, 1.1 mertz, 1.0 lucy
2773</artwork></figure>
2774<t>
2775   A sender &SHOULD-NOT; combine multiple entries unless they are all
2776   under the same organizational control and the hosts have already been
2777   replaced by pseudonyms. A sender &MUST-NOT; combine entries that
2778   have different received-protocol values.
2779</t>
2780</section>
2781
2782<section title="Transformations" anchor="message.transformations">
2783   <iref primary="true" item="transforming proxy"/>
2784   <iref primary="true" item="non-transforming proxy"/>
2785<t>
2786   Some intermediaries include features for transforming messages and their
2787   payloads. A proxy might, for example, convert between image formats in
2788   order to save cache space or to reduce the amount of traffic on a slow
2789   link. However, operational problems might occur when these transformations
2790   are applied to payloads intended for critical applications, such as medical
2791   imaging or scientific data analysis, particularly when integrity checks or
2792   digital signatures are used to ensure that the payload received is
2793   identical to the original.
2794</t>
2795<t>
2796   An HTTP-to-HTTP proxy is called a "<x:dfn>transforming proxy</x:dfn>"
2797   if it is designed or configured to modify messages in a semantically
2798   meaningful way (i.e., modifications, beyond those required by normal
2799   HTTP processing, that change the message in a way that would be
2800   significant to the original sender or potentially significant to
2801   downstream recipients).  For example, a transforming proxy might be
2802   acting as a shared annotation server (modifying responses to include
2803   references to a local annotation database), a malware filter, a
2804   format transcoder, or a privacy filter. Such transformations are presumed
2805   to be desired by whichever client (or client organization) selected the
2806   proxy.
2807</t>
2808<t>
2809   If a proxy receives a request-target with a host name that is not a
2810   fully qualified domain name, it &MAY; add its own domain to the host name
2811   it received when forwarding the request.  A proxy &MUST-NOT; change the
2812   host name if the request-target contains a fully qualified domain name.
2813</t>
2814<t>
2815   A proxy &MUST-NOT; modify the "absolute-path" and "query" parts of the
2816   received request-target when forwarding it to the next inbound server,
2817   except as noted above to replace an empty path with "/" or "*".
2818</t>
2819<t>
2820   A proxy &MAY; modify the message body through application
2821   or removal of a transfer coding (<xref target="transfer.codings"/>).
2822</t>
2823<t>
2824   A proxy &MUST-NOT; transform the payload (&payload;) of a message that
2825   contains a no-transform cache-control directive (&header-cache-control;).
2826</t>
2827<t>
2828   A proxy &MAY; transform the payload of a message
2829   that does not contain a no-transform cache-control directive.
2830   A proxy that transforms a payload &MUST; add a <x:ref>Warning</x:ref>
2831   header field with the warn-code of 214 ("Transformation Applied")
2832   if one is not already in the message (see &header-warning;).
2833   A proxy that transforms the payload of a <x:ref>200 (OK)</x:ref> response
2834   can further inform downstream recipients that a transformation has been
2835   applied by changing the response status code to
2836   <x:ref>203 (Non-Authoritative Information)</x:ref> (&status-203;).
2837</t>
2838<t>
2839   A proxy &SHOULD-NOT; modify header fields that provide information about
2840   the end points of the communication chain, the resource state, or the
2841   selected representation (other than the payload) unless the field's
2842   definition specifically allows such modification or the modification is
2843   deemed necessary for privacy or security.
2844</t>
2845</section>
2846</section>
2847</section>
2848
2849<section title="Connection Management" anchor="connection.management">
2850<t>
2851   HTTP messaging is independent of the underlying transport or
2852   session-layer connection protocol(s).  HTTP only presumes a reliable
2853   transport with in-order delivery of requests and the corresponding
2854   in-order delivery of responses.  The mapping of HTTP request and
2855   response structures onto the data units of an underlying transport
2856   protocol is outside the scope of this specification.
2857</t>
2858<t>
2859   As described in <xref target="connecting.inbound"/>, the specific
2860   connection protocols to be used for an HTTP interaction are determined by
2861   client configuration and the <x:ref>target URI</x:ref>.
2862   For example, the "http" URI scheme
2863   (<xref target="http.uri"/>) indicates a default connection of TCP
2864   over IP, with a default TCP port of 80, but the client might be
2865   configured to use a proxy via some other connection, port, or protocol.
2866</t>
2867<t>
2868   HTTP implementations are expected to engage in connection management,
2869   which includes maintaining the state of current connections,
2870   establishing a new connection or reusing an existing connection,
2871   processing messages received on a connection, detecting connection
2872   failures, and closing each connection.
2873   Most clients maintain multiple connections in parallel, including
2874   more than one connection per server endpoint.
2875   Most servers are designed to maintain thousands of concurrent connections,
2876   while controlling request queues to enable fair use and detect
2877   denial of service attacks.
2878</t>
2879
2880<section title="Connection" anchor="header.connection">
2881  <iref primary="true" item="Connection header field" x:for-anchor=""/>
2882  <iref primary="true" item="close" x:for-anchor=""/>
2883  <x:anchor-alias value="Connection"/>
2884  <x:anchor-alias value="connection-option"/>
2885  <x:anchor-alias value="close"/>
2886<t>
2887   The "Connection" header field allows the sender to indicate desired
2888   control options for the current connection.  In order to avoid confusing
2889   downstream recipients, a proxy or gateway &MUST; remove or replace any
2890   received connection options before forwarding the message.
2891</t>
2892<t>
2893   When a header field aside from Connection is used to supply control
2894   information for or about the current connection, the sender &MUST; list
2895   the corresponding field-name within the "Connection" header field.
2896   A proxy or gateway &MUST; parse a received Connection
2897   header field before a message is forwarded and, for each
2898   connection-option in this field, remove any header field(s) from
2899   the message with the same name as the connection-option, and then
2900   remove the Connection header field itself (or replace it with the
2901   intermediary's own connection options for the forwarded message).
2902</t>
2903<t>
2904   Hence, the Connection header field provides a declarative way of
2905   distinguishing header fields that are only intended for the
2906   immediate recipient ("hop-by-hop") from those fields that are
2907   intended for all recipients on the chain ("end-to-end"), enabling the
2908   message to be self-descriptive and allowing future connection-specific
2909   extensions to be deployed without fear that they will be blindly
2910   forwarded by older intermediaries.
2911</t>
2912<t>
2913   The Connection header field's value has the following grammar:
2914</t>
2915<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Connection"/><iref primary="true" item="Grammar" subitem="connection-option"/>
2916  <x:ref>Connection</x:ref>        = 1#<x:ref>connection-option</x:ref>
2917  <x:ref>connection-option</x:ref> = <x:ref>token</x:ref>
2918</artwork></figure>
2919<t>
2920   Connection options are case-insensitive.
2921</t>
2922<t>
2923   A sender &MUST-NOT; send a connection option corresponding to a header
2924   field that is intended for all recipients of the payload.
2925   For example, <x:ref>Cache-Control</x:ref> is never appropriate as a
2926   connection option (&header-cache-control;).
2927</t>
2928<t>
2929   The connection options do not always correspond to a header field
2930   present in the message, since a connection-specific header field
2931   might not be needed if there are no parameters associated with a
2932   connection option. In contrast, a connection-specific header field that
2933   is received without a corresponding connection option usually indicates
2934   that the field has been improperly forwarded by an intermediary and
2935   ought to be ignored by the recipient.
2936</t>
2937<t>
2938   When defining new connection options, specification authors ought to survey
2939   existing header field names and ensure that the new connection option does
2940   not share the same name as an already deployed header field.
2941   Defining a new connection option essentially reserves that potential
2942   field-name for carrying additional information related to the
2943   connection option, since it would be unwise for senders to use
2944   that field-name for anything else.
2945</t>
2946<t>
2947   The "<x:dfn>close</x:dfn>" connection option is defined for a
2948   sender to signal that this connection will be closed after completion of
2949   the response. For example,
2950</t>
2951<figure><artwork type="example">
2952  Connection: close
2953</artwork></figure>
2954<t>
2955   in either the request or the response header fields indicates that the
2956   sender is going to close the connection after the current request/response
2957   is complete (<xref target="persistent.tear-down"/>).
2958</t>
2959<t>
2960   A client that does not support <x:ref>persistent connections</x:ref> &MUST;
2961   send the "close" connection option in every request message.
2962</t>
2963<t>
2964   A server that does not support <x:ref>persistent connections</x:ref> &MUST;
2965   send the "close" connection option in every response message that
2966   does not have a <x:ref>1xx (Informational)</x:ref> status code.
2967</t>
2968</section>
2969
2970<section title="Establishment" anchor="persistent.establishment">
2971<t>
2972   It is beyond the scope of this specification to describe how connections
2973   are established via various transport or session-layer protocols.
2974   Each connection applies to only one transport link.
2975</t>
2976</section>
2977
2978<section title="Persistence" anchor="persistent.connections">
2979   <x:anchor-alias value="persistent connections"/>
2980<t>
2981   HTTP/1.1 defaults to the use of "<x:dfn>persistent connections</x:dfn>",
2982   allowing multiple requests and responses to be carried over a single
2983   connection. The "<x:ref>close</x:ref>" connection-option is used to signal
2984   that a connection will not persist after the current request/response.
2985   HTTP implementations &SHOULD; support persistent connections.
2986</t>
2987<t>
2988   A recipient determines whether a connection is persistent or not based on
2989   the most recently received message's protocol version and
2990   <x:ref>Connection</x:ref> header field (if any):
2991   <list style="symbols">
2992     <t>If the <x:ref>close</x:ref> connection option is present, the
2993        connection will not persist after the current response; else,</t>
2994     <t>If the received protocol is HTTP/1.1 (or later), the connection will
2995        persist after the current response; else,</t>
2996     <t>If the received protocol is HTTP/1.0, the "keep-alive"
2997        connection option is present, the recipient is not a proxy, and
2998        the recipient wishes to honor the HTTP/1.0 "keep-alive" mechanism,
2999        the connection will persist after the current response; otherwise,</t>
3000     <t>The connection will close after the current response.</t>
3001   </list>
3002</t>
3003<t>
3004   A client &MAY; send additional requests on a persistent connection until it
3005   sends or receives a <x:ref>close</x:ref> connection option or receives an
3006   HTTP/1.0 response without a "keep-alive" connection option.
3007</t>
3008<t>
3009   In order to remain persistent, all messages on a connection need to
3010   have a self-defined message length (i.e., one not defined by closure
3011   of the connection), as described in <xref target="message.body"/>.
3012   A server &MUST; read the entire request message body or close
3013   the connection after sending its response, since otherwise the
3014   remaining data on a persistent connection would be misinterpreted
3015   as the next request.  Likewise,
3016   a client &MUST; read the entire response message body if it intends
3017   to reuse the same connection for a subsequent request.
3018</t>
3019<t>
3020   A proxy server &MUST-NOT; maintain a persistent connection with an
3021   HTTP/1.0 client (see <xref x:sec="19.7.1" x:fmt="of" target="RFC2068"/> for
3022   information and discussion of the problems with the Keep-Alive header field
3023   implemented by many HTTP/1.0 clients).
3024</t>
3025<t>
3026   See <xref target="compatibility.with.http.1.0.persistent.connections"/>
3027   for more information on backward compatibility with HTTP/1.0 clients.
3028</t>
3029
3030<section title="Retrying Requests" anchor="persistent.retrying.requests">
3031<t>
3032   Connections can be closed at any time, with or without intention.
3033   Implementations ought to anticipate the need to recover
3034   from asynchronous close events.
3035</t>
3036<t>
3037   When an inbound connection is closed prematurely, a client &MAY; open a new
3038   connection and automatically retransmit an aborted sequence of requests if
3039   all of those requests have idempotent methods (&idempotent-methods;).
3040   A proxy &MUST-NOT; automatically retry non-idempotent requests.
3041</t>
3042<t>
3043   A user agent &MUST-NOT; automatically retry a request with a non-idempotent
3044   method unless it has some means to know that the request semantics are
3045   actually idempotent, regardless of the method, or some means to detect that
3046   the original request was never applied. For example, a user agent that
3047   knows (through design or configuration) that a POST request to a given
3048   resource is safe can repeat that request automatically.
3049   Likewise, a user agent designed specifically to operate on a version
3050   control repository might be able to recover from partial failure conditions
3051   by checking the target resource revision(s) after a failed connection,
3052   reverting or fixing any changes that were partially applied, and then
3053   automatically retrying the requests that failed.
3054</t>
3055<t>
3056   A client &SHOULD-NOT; automatically retry a failed automatic retry.
3057</t>
3058</section>
3059
3060<section title="Pipelining" anchor="pipelining">
3061   <x:anchor-alias value="pipeline"/>
3062<t>
3063   A client that supports persistent connections &MAY; "<x:dfn>pipeline</x:dfn>"
3064   its requests (i.e., send multiple requests without waiting for each
3065   response). A server &MAY; process a sequence of pipelined requests in
3066   parallel if they all have safe methods (&safe-methods;), but &MUST; send
3067   the corresponding responses in the same order that the requests were
3068   received.
3069</t>
3070<t>
3071   A client that pipelines requests &SHOULD; retry unanswered requests if the
3072   connection closes before it receives all of the corresponding responses.
3073   When retrying pipelined requests after a failed connection (a connection
3074   not explicitly closed by the server in its last complete response), a
3075   client &MUST-NOT; pipeline immediately after connection establishment,
3076   since the first remaining request in the prior pipeline might have caused
3077   an error response that can be lost again if multiple requests are sent on a
3078   prematurely closed connection (see the TCP reset problem described in
3079   <xref target="persistent.tear-down"/>).
3080</t>
3081<t>
3082   Idempotent methods (&idempotent-methods;) are significant to pipelining
3083   because they can be automatically retried after a connection failure.
3084   A user agent &SHOULD-NOT; pipeline requests after a non-idempotent method,
3085   until the final response status code for that method has been received,
3086   unless the user agent has a means to detect and recover from partial
3087   failure conditions involving the pipelined sequence.
3088</t>
3089<t>
3090   An intermediary that receives pipelined requests &MAY; pipeline those
3091   requests when forwarding them inbound, since it can rely on the outbound
3092   user agent(s) to determine what requests can be safely pipelined. If the
3093   inbound connection fails before receiving a response, the pipelining
3094   intermediary &MAY; attempt to retry a sequence of requests that have yet
3095   to receive a response if the requests all have idempotent methods;
3096   otherwise, the pipelining intermediary &SHOULD; forward any received
3097   responses and then close the corresponding outbound connection(s) so that
3098   the outbound user agent(s) can recover accordingly.
3099</t>
3100</section>
3101</section>
3102   
3103<section title="Concurrency" anchor="persistent.concurrency">
3104<t>
3105   A client ought to limit the number of simultaneous open
3106   connections that it maintains to a given server.
3107</t>
3108<t>
3109   Previous revisions of HTTP gave a specific number of connections as a
3110   ceiling, but this was found to be impractical for many applications. As a
3111   result, this specification does not mandate a particular maximum number of
3112   connections, but instead encourages clients to be conservative when opening
3113   multiple connections.
3114</t>
3115<t>
3116   Multiple connections are typically used to avoid the "head-of-line
3117   blocking" problem, wherein a request that takes significant server-side
3118   processing and/or has a large payload blocks subsequent requests on the
3119   same connection. However, each connection consumes server resources.
3120   Furthermore, using multiple connections can cause undesirable side effects
3121   in congested networks.
3122</t>
3123<t>
3124   Note that a server might reject traffic that it deems abusive or
3125   characteristic of a denial of service attack, such as an excessive number
3126   of open connections from a single client.
3127</t>
3128</section>
3129
3130<section title="Failures and Time-outs" anchor="persistent.failures">
3131<t>
3132   Servers will usually have some time-out value beyond which they will
3133   no longer maintain an inactive connection. Proxy servers might make
3134   this a higher value since it is likely that the client will be making
3135   more connections through the same proxy server. The use of persistent
3136   connections places no requirements on the length (or existence) of
3137   this time-out for either the client or the server.
3138</t>
3139<t>
3140   A client or server that wishes to time-out &SHOULD; issue a graceful close
3141   on the connection. Implementations &SHOULD; constantly monitor open
3142   connections for a received closure signal and respond to it as appropriate,
3143   since prompt closure of both sides of a connection enables allocated system
3144   resources to be reclaimed.
3145</t>
3146<t>
3147   A client, server, or proxy &MAY; close the transport connection at any
3148   time. For example, a client might have started to send a new request
3149   at the same time that the server has decided to close the "idle"
3150   connection. From the server's point of view, the connection is being
3151   closed while it was idle, but from the client's point of view, a
3152   request is in progress.
3153</t>
3154<t>
3155   A server &SHOULD; sustain persistent connections, when possible, and allow
3156   the underlying
3157   transport's flow control mechanisms to resolve temporary overloads, rather
3158   than terminate connections with the expectation that clients will retry.
3159   The latter technique can exacerbate network congestion.
3160</t>
3161<t>
3162   A client sending a message body &SHOULD; monitor
3163   the network connection for an error response while it is transmitting
3164   the request. If the client sees a response that indicates the server does
3165   not wish to receive the message body and is closing the connection, the
3166   client &SHOULD; immediately cease transmitting the body and close its side
3167   of the connection.
3168</t>
3169</section>
3170   
3171<section title="Tear-down" anchor="persistent.tear-down">
3172  <iref primary="false" item="Connection header field" x:for-anchor=""/>
3173  <iref primary="false" item="close" x:for-anchor=""/>
3174<t>
3175   The <x:ref>Connection</x:ref> header field
3176   (<xref target="header.connection"/>) provides a "<x:ref>close</x:ref>"
3177   connection option that a sender &SHOULD; send when it wishes to close
3178   the connection after the current request/response pair.
3179</t>
3180<t>
3181   A client that sends a <x:ref>close</x:ref> connection option &MUST-NOT;
3182   send further requests on that connection (after the one containing
3183   <x:ref>close</x:ref>) and &MUST; close the connection after reading the
3184   final response message corresponding to this request.
3185</t>
3186<t>
3187   A server that receives a <x:ref>close</x:ref> connection option &MUST;
3188   initiate a close of the connection (see below) after it sends the
3189   final response to the request that contained <x:ref>close</x:ref>.
3190   The server &SHOULD; send a <x:ref>close</x:ref> connection option
3191   in its final response on that connection. The server &MUST-NOT; process
3192   any further requests received on that connection.
3193</t>
3194<t>
3195   A server that sends a <x:ref>close</x:ref> connection option &MUST;
3196   initiate a close of the connection (see below) after it sends the
3197   response containing <x:ref>close</x:ref>. The server &MUST-NOT; process
3198   any further requests received on that connection.
3199</t>
3200<t>
3201   A client that receives a <x:ref>close</x:ref> connection option &MUST;
3202   cease sending requests on that connection and close the connection
3203   after reading the response message containing the close; if additional
3204   pipelined requests had been sent on the connection, the client &SHOULD-NOT;
3205   assume that they will be processed by the server.
3206</t>
3207<t>
3208   If a server performs an immediate close of a TCP connection, there is a
3209   significant risk that the client will not be able to read the last HTTP
3210   response.  If the server receives additional data from the client on a
3211   fully-closed connection, such as another request that was sent by the
3212   client before receiving the server's response, the server's TCP stack will
3213   send a reset packet to the client; unfortunately, the reset packet might
3214   erase the client's unacknowledged input buffers before they can be read
3215   and interpreted by the client's HTTP parser.
3216</t>
3217<t>
3218   To avoid the TCP reset problem, servers typically close a connection in
3219   stages. First, the server performs a half-close by closing only the write
3220   side of the read/write connection. The server then continues to read from
3221   the connection until it receives a corresponding close by the client, or
3222   until the server is reasonably certain that its own TCP stack has received
3223   the client's acknowledgement of the packet(s) containing the server's last
3224   response. Finally, the server fully closes the connection.
3225</t>
3226<t>
3227   It is unknown whether the reset problem is exclusive to TCP or might also
3228   be found in other transport connection protocols.
3229</t>
3230</section>
3231
3232<section title="Upgrade" anchor="header.upgrade">
3233  <iref primary="true" item="Upgrade header field" x:for-anchor=""/>
3234  <x:anchor-alias value="Upgrade"/>
3235  <x:anchor-alias value="protocol"/>
3236  <x:anchor-alias value="protocol-name"/>
3237  <x:anchor-alias value="protocol-version"/>
3238<t>
3239   The "Upgrade" header field is intended to provide a simple mechanism
3240   for transitioning from HTTP/1.1 to some other protocol on the same
3241   connection.  A client &MAY; send a list of protocols in the Upgrade
3242   header field of a request to invite the server to switch to one or
3243   more of those protocols, in order of descending preference, before sending
3244   the final response. A server &MAY; ignore a received Upgrade header field
3245   if it wishes to continue using the current protocol on that connection.
3246   Upgrade cannot be used to insist on a protocol change.
3247</t>
3248<figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Upgrade"/>
3249  <x:ref>Upgrade</x:ref>          = 1#<x:ref>protocol</x:ref>
3250
3251  <x:ref>protocol</x:ref>         = <x:ref>protocol-name</x:ref> ["/" <x:ref>protocol-version</x:ref>]
3252  <x:ref>protocol-name</x:ref>    = <x:ref>token</x:ref>
3253  <x:ref>protocol-version</x:ref> = <x:ref>token</x:ref>
3254</artwork></figure>
3255<t>
3256   A server that sends a <x:ref>101 (Switching Protocols)</x:ref> response
3257   &MUST; send an Upgrade header field to indicate the new protocol(s) to
3258   which the connection is being switched; if multiple protocol layers are
3259   being switched, the sender &MUST; list the protocols in layer-ascending
3260   order. A server &MUST-NOT; switch to a protocol that was not indicated by
3261   the client in the corresponding request's Upgrade header field.
3262   A server &MAY; choose to ignore the order of preference indicated by the
3263   client and select the new protocol(s) based on other factors, such as the
3264   nature of the request or the current load on the server.
3265</t>
3266<t>
3267   A server that sends a <x:ref>426 (Upgrade Required)</x:ref> response
3268   &MUST; send an Upgrade header field to indicate the acceptable protocols,
3269   in order of descending preference.
3270</t>
3271<t>
3272   A server &MAY; send an Upgrade header field in any other response to
3273   advertise that it implements support for upgrading to the listed protocols,
3274   in order of descending preference, when appropriate for a future request.
3275</t>
3276<figure><preamble>
3277   The following is a hypothetical example sent by a client:
3278</preamble><artwork type="message/http; msgtype=&#34;request&#34;" x:indent-with="  ">
3279GET /hello.txt HTTP/1.1
3280Host: www.example.com
3281Connection: upgrade
3282Upgrade: HTTP/2.0, SHTTP/1.3, IRC/6.9, RTA/x11
3283
3284</artwork></figure>
3285<t>
3286   The capabilities and nature of the
3287   application-level communication after the protocol change is entirely
3288   dependent upon the new protocol(s) chosen. However, immediately after
3289   sending the 101 response, the server is expected to continue responding to
3290   the original request as if it had received its equivalent within the new
3291   protocol (i.e., the server still has an outstanding request to satisfy
3292   after the protocol has been changed, and is expected to do so without
3293   requiring the request to be repeated).
3294</t>
3295<t>
3296   For example, if the Upgrade header field is received in a GET request
3297   and the server decides to switch protocols, it first responds
3298   with a <x:ref>101 (Switching Protocols)</x:ref> message in HTTP/1.1 and
3299   then immediately follows that with the new protocol's equivalent of a
3300   response to a GET on the target resource.  This allows a connection to be
3301   upgraded to protocols with the same semantics as HTTP without the
3302   latency cost of an additional round-trip.  A server &MUST-NOT; switch
3303   protocols unless the received message semantics can be honored by the new
3304   protocol; an OPTIONS request can be honored by any protocol.
3305</t>
3306<figure><preamble>
3307   The following is an example response to the above hypothetical request:
3308</preamble><artwork type="message/http; msgtype=&#34;response&#34;" x:indent-with="  ">
3309HTTP/1.1 101 Switching Protocols
3310Connection: upgrade
3311Upgrade: HTTP/2.0
3312
3313[... data stream switches to HTTP/2.0 with an appropriate response
3314(as defined by new protocol) to the "GET /hello.txt" request ...]
3315</artwork></figure>
3316<t>
3317   When Upgrade is sent, the sender &MUST; also send a
3318   <x:ref>Connection</x:ref> header field (<xref target="header.connection"/>)
3319   that contains an "upgrade" connection option, in order to prevent Upgrade
3320   from being accidentally forwarded by intermediaries that might not implement
3321   the listed protocols.  A server &MUST; ignore an Upgrade header field that
3322   is received in an HTTP/1.0 request.
3323</t>
3324<t>
3325   A client cannot begin using an upgraded protocol on the connection until
3326   it has completely sent the request message (i.e., the client can't change
3327   the protocol it is sending in the middle of a message).
3328   If a server receives both Upgrade and an <x:ref>Expect</x:ref> header field
3329   with the "100-continue" expectation (&header-expect;), the
3330   server &MUST; send a <x:ref>100 (Continue)</x:ref> response before sending
3331   a <x:ref>101 (Switching Protocols)</x:ref> response.
3332</t>
3333<t>
3334   The Upgrade header field only applies to switching protocols on top of the
3335   existing connection; it cannot be used to switch the underlying connection
3336   (transport) protocol, nor to switch the existing communication to a
3337   different connection. For those purposes, it is more appropriate to use a
3338   <x:ref>3xx (Redirection)</x:ref> response (&status-3xx;).
3339</t>
3340<t>
3341   This specification only defines the protocol name "HTTP" for use by
3342   the family of Hypertext Transfer Protocols, as defined by the HTTP
3343   version rules of <xref target="http.version"/> and future updates to this
3344   specification. Additional tokens ought to be registered with IANA using the
3345   registration procedure defined in <xref target="upgrade.token.registry"/>.
3346</t>
3347</section>
3348</section>
3349
3350<section title="ABNF list extension: #rule" anchor="abnf.extension">
3351<t>
3352   A #rule extension to the ABNF rules of <xref target="RFC5234"/> is used to
3353   improve readability in the definitions of some header field values.
3354</t>
3355<t>
3356   A construct "#" is defined, similar to "*", for defining comma-delimited
3357   lists of elements. The full form is "&lt;n&gt;#&lt;m&gt;element" indicating
3358   at least &lt;n&gt; and at most &lt;m&gt; elements, each separated by a single
3359   comma (",") and optional whitespace (OWS).   
3360</t>
3361<figure><preamble>
3362   In any production that uses the list construct, a sender &MUST-NOT;
3363   generate empty list elements. In other words, a sender &MUST; generate
3364   lists that satisfy the following syntax:
3365</preamble><artwork type="example">
3366  1#element =&gt; element *( OWS "," OWS element )
3367</artwork></figure>
3368<figure><preamble>
3369   and:
3370</preamble><artwork type="example">
3371  #element =&gt; [ 1#element ]
3372</artwork></figure>
3373<figure><preamble>
3374   and for n &gt;= 1 and m &gt; 1:
3375</preamble><artwork type="example">
3376  &lt;n&gt;#&lt;m&gt;element =&gt; element &lt;n-1&gt;*&lt;m-1&gt;( OWS "," OWS element )
3377</artwork></figure>
3378<t>
3379   For compatibility with legacy list rules, a recipient &MUST; parse and ignore
3380   a reasonable number of empty list elements: enough to handle common mistakes
3381   by senders that merge values, but not so much that they could be used as a
3382   denial of service mechanism. In other words, a recipient &MUST; accept lists
3383   that satisfy the following syntax:
3384</t>
3385<figure><artwork type="example">
3386  #element =&gt; [ ( "," / element ) *( OWS "," [ OWS element ] ) ]
3387 
3388  1#element =&gt; *( "," OWS ) element *( OWS "," [ OWS element ] )
3389</artwork></figure>
3390<t>
3391   Empty elements do not contribute to the count of elements present.
3392   For example, given these ABNF productions:
3393</t>
3394<figure><artwork type="example">
3395  example-list      = 1#example-list-elmt
3396  example-list-elmt = token ; see <xref target="field.components"/>
3397</artwork></figure>
3398<t>
3399   Then the following are valid values for example-list (not including the
3400   double quotes, which are present for delimitation only):
3401</t>
3402<figure><artwork type="example">
3403  "foo,bar"
3404  "foo ,bar,"
3405  "foo , ,bar,charlie   "
3406</artwork></figure>
3407<t>
3408   In contrast, the following values would be invalid, since at least one
3409   non-empty element is required by the example-list production:
3410</t>
3411<figure><artwork type="example">
3412  ""
3413  ","
3414  ",   ,"
3415</artwork></figure>
3416<t>
3417   <xref target="collected.abnf"/> shows the collected ABNF for recipients
3418   after the list constructs have been expanded.
3419</t>
3420</section>
3421
3422<section title="IANA Considerations" anchor="IANA.considerations">
3423
3424<section title="Header Field Registration" anchor="header.field.registration">
3425<t>
3426   HTTP header fields are registered within the Message Header Field Registry
3427   maintained at
3428   <eref target="http://www.iana.org/assignments/message-headers/"/>.
3429</t>
3430<t>
3431   This document defines the following HTTP header fields, so their
3432   associated registry entries shall be updated according to the permanent
3433   registrations below (see <xref target="BCP90"/>):
3434</t>
3435<?BEGININC p1-messaging.iana-headers ?>
3436<!--AUTOGENERATED FROM extract-header-defs.xslt, do not edit manually-->
3437<texttable align="left" suppress-title="true" anchor="iana.header.registration.table">
3438   <ttcol>Header Field Name</ttcol>
3439   <ttcol>Protocol</ttcol>
3440   <ttcol>Status</ttcol>
3441   <ttcol>Reference</ttcol>
3442
3443   <c>Connection</c>
3444   <c>http</c>
3445   <c>standard</c>
3446   <c>
3447      <xref target="header.connection"/>
3448   </c>
3449   <c>Content-Length</c>
3450   <c>http</c>
3451   <c>standard</c>
3452   <c>
3453      <xref target="header.content-length"/>
3454   </c>
3455   <c>Host</c>
3456   <c>http</c>
3457   <c>standard</c>
3458   <c>
3459      <xref target="header.host"/>
3460   </c>
3461   <c>TE</c>
3462   <c>http</c>
3463   <c>standard</c>
3464   <c>
3465      <xref target="header.te"/>
3466   </c>
3467   <c>Trailer</c>
3468   <c>http</c>
3469   <c>standard</c>
3470   <c>
3471      <xref target="header.trailer"/>
3472   </c>
3473   <c>Transfer-Encoding</c>
3474   <c>http</c>
3475   <c>standard</c>
3476   <c>
3477      <xref target="header.transfer-encoding"/>
3478   </c>
3479   <c>Upgrade</c>
3480   <c>http</c>
3481   <c>standard</c>
3482   <c>
3483      <xref target="header.upgrade"/>
3484   </c>
3485   <c>Via</c>
3486   <c>http</c>
3487   <c>standard</c>
3488   <c>
3489      <xref target="header.via"/>
3490   </c>
3491</texttable>
3492<!--(END)-->
3493<?ENDINC p1-messaging.iana-headers ?>
3494<t>
3495   Furthermore, the header field-name "Close" shall be registered as
3496   "reserved", since using that name as an HTTP header field might
3497   conflict with the "close" connection option of the "<x:ref>Connection</x:ref>"
3498   header field (<xref target="header.connection"/>).
3499</t>
3500<texttable align="left" suppress-title="true">
3501   <ttcol>Header Field Name</ttcol>
3502   <ttcol>Protocol</ttcol>
3503   <ttcol>Status</ttcol>
3504   <ttcol>Reference</ttcol>
3505
3506   <c>Close</c>
3507   <c>http</c>
3508   <c>reserved</c>
3509   <c>
3510      <xref target="header.field.registration"/>
3511   </c>
3512</texttable>
3513<t>
3514   The change controller is: "IETF (iesg@ietf.org) - Internet Engineering Task Force".
3515</t>
3516</section>
3517
3518<section title="URI Scheme Registration" anchor="uri.scheme.registration">
3519<t>
3520   IANA maintains the registry of URI Schemes <xref target="BCP115"/> at
3521   <eref target="http://www.iana.org/assignments/uri-schemes/"/>.
3522</t>
3523<t>
3524   This document defines the following URI schemes, so their
3525   associated registry entries shall be updated according to the permanent
3526   registrations below:
3527</t>
3528<texttable align="left" suppress-title="true">
3529   <ttcol>URI Scheme</ttcol>
3530   <ttcol>Description</ttcol>
3531   <ttcol>Reference</ttcol>
3532
3533   <c>http</c>
3534   <c>Hypertext Transfer Protocol</c>
3535   <c><xref target="http.uri"/></c>
3536
3537   <c>https</c>
3538   <c>Hypertext Transfer Protocol Secure</c>
3539   <c><xref target="https.uri"/></c>
3540</texttable>
3541</section>
3542
3543<section title="Internet Media Type Registration" anchor="internet.media.type.http">
3544<t>
3545   IANA maintains the registry of Internet media types <xref target="BCP13"/> at
3546   <eref target="http://www.iana.org/assignments/media-types"/>.
3547</t>
3548<t>
3549   This document serves as the specification for the Internet media types
3550   "message/http" and "application/http". The following is to be registered with
3551   IANA.
3552</t>
3553<section title="Internet Media Type message/http" anchor="internet.media.type.message.http">
3554<iref item="Media Type" subitem="message/http" primary="true"/>
3555<iref item="message/http Media Type" primary="true"/>
3556<t>
3557   The message/http type can be used to enclose a single HTTP request or
3558   response message, provided that it obeys the MIME restrictions for all
3559   "message" types regarding line length and encodings.
3560</t>
3561<t>
3562  <list style="hanging" x:indent="12em">
3563    <t hangText="Type name:">
3564      message
3565    </t>
3566    <t hangText="Subtype name:">
3567      http
3568    </t>
3569    <t hangText="Required parameters:">
3570      N/A
3571    </t>
3572    <t hangText="Optional parameters:">
3573      version, msgtype
3574      <list style="hanging">
3575        <t hangText="version:">
3576          The HTTP-version number of the enclosed message
3577          (e.g., "1.1"). If not present, the version can be
3578          determined from the first line of the body.
3579        </t>
3580        <t hangText="msgtype:">
3581          The message type &mdash; "request" or "response". If not
3582          present, the type can be determined from the first
3583          line of the body.
3584        </t>
3585      </list>
3586    </t>
3587    <t hangText="Encoding considerations:">
3588      only "7bit", "8bit", or "binary" are permitted
3589    </t>
3590    <t hangText="Security considerations:">
3591      see <xref target="security.considerations"/>
3592    </t>
3593    <t hangText="Interoperability considerations:">
3594      N/A
3595    </t>
3596    <t hangText="Published specification:">
3597      This specification (see <xref target="internet.media.type.message.http"/>).
3598    </t>
3599    <t hangText="Applications that use this media type:">
3600      N/A
3601    </t>
3602    <t hangText="Fragment identifier considerations:">
3603      N/A
3604    </t>
3605    <t hangText="Additional information:">
3606      <list style="hanging">
3607        <t hangText="Magic number(s):">N/A</t>
3608        <t hangText="Deprecated alias names for this type:">N/A</t>
3609        <t hangText="File extension(s):">N/A</t>
3610        <t hangText="Macintosh file type code(s):">N/A</t>
3611      </list>
3612    </t>
3613    <t hangText="Person and email address to contact for further information:">
3614      See Authors Section.
3615    </t>
3616    <t hangText="Intended usage:">
3617      COMMON
3618    </t>
3619    <t hangText="Restrictions on usage:">
3620      N/A
3621    </t>
3622    <t hangText="Author:">
3623      See Authors Section.
3624    </t>
3625    <t hangText="Change controller:">
3626      IESG
3627    </t>
3628  </list>
3629</t>
3630</section>
3631<section title="Internet Media Type application/http" anchor="internet.media.type.application.http">
3632<iref item="Media Type" subitem="application/http" primary="true"/>
3633<iref item="application/http Media Type" primary="true"/>
3634<t>
3635   The application/http type can be used to enclose a pipeline of one or more
3636   HTTP request or response messages (not intermixed).
3637</t>
3638<t>
3639  <list style="hanging" x:indent="12em">
3640    <t hangText="Type name:">
3641      application
3642    </t>
3643    <t hangText="Subtype name:">
3644      http
3645    </t>
3646    <t hangText="Required parameters:">
3647      N/A
3648    </t>
3649    <t hangText="Optional parameters:">
3650      version, msgtype
3651      <list style="hanging">
3652        <t hangText="version:">
3653          The HTTP-version number of the enclosed messages
3654          (e.g., "1.1"). If not present, the version can be
3655          determined from the first line of the body.
3656        </t>
3657        <t hangText="msgtype:">
3658          The message type &mdash; "request" or "response". If not
3659          present, the type can be determined from the first
3660          line of the body.
3661        </t>
3662      </list>
3663    </t>
3664    <t hangText="Encoding considerations:">
3665      HTTP messages enclosed by this type
3666      are in "binary" format; use of an appropriate
3667      Content-Transfer-Encoding is required when
3668      transmitted via E-mail.
3669    </t>
3670    <t hangText="Security considerations:">
3671      see <xref target="security.considerations"/>
3672    </t>
3673    <t hangText="Interoperability considerations:">
3674      N/A
3675    </t>
3676    <t hangText="Published specification:">
3677      This specification (see <xref target="internet.media.type.application.http"/>).
3678    </t>
3679    <t hangText="Applications that use this media type:">
3680      N/A
3681    </t>
3682    <t hangText="Fragment identifier considerations:">
3683      N/A
3684    </t>
3685    <t hangText="Additional information:">
3686      <list style="hanging">
3687        <t hangText="Deprecated alias names for this type:">N/A</t>
3688        <t hangText="Magic number(s):">N/A</t>
3689        <t hangText="File extension(s):">N/A</t>
3690        <t hangText="Macintosh file type code(s):">N/A</t>
3691      </list>
3692    </t>
3693    <t hangText="Person and email address to contact for further information:">
3694      See Authors Section.
3695    </t>
3696    <t hangText="Intended usage:">
3697      COMMON
3698    </t>
3699    <t hangText="Restrictions on usage:">
3700      N/A
3701    </t>
3702    <t hangText="Author:">
3703      See Authors Section.
3704    </t>
3705    <t hangText="Change controller:">
3706      IESG
3707    </t>
3708  </list>
3709</t>
3710</section>
3711</section>
3712
3713<section title="Transfer Coding Registry" anchor="transfer.coding.registry">
3714<t>
3715   The HTTP Transfer Coding Registry defines the name space for transfer
3716   coding names. It is maintained at <eref target="http://www.iana.org/assignments/http-parameters"/>.
3717</t>
3718
3719<section title="Procedure" anchor="transfer.coding.registry.procedure">
3720<t>
3721   Registrations &MUST; include the following fields:
3722   <list style="symbols">
3723     <t>Name</t>
3724     <t>Description</t>
3725     <t>Pointer to specification text</t>
3726   </list>
3727</t>
3728<t>
3729   Names of transfer codings &MUST-NOT; overlap with names of content codings
3730   (&content-codings;) unless the encoding transformation is identical, as
3731   is the case for the compression codings defined in
3732   <xref target="compression.codings"/>.
3733</t>
3734<t>
3735   Values to be added to this name space require IETF Review (see
3736   <xref target="RFC5226" x:fmt="of" x:sec="4.1"/>), and &MUST;
3737   conform to the purpose of transfer coding defined in this specification.
3738</t>
3739<t>
3740   Use of program names for the identification of encoding formats
3741   is not desirable and is discouraged for future encodings.
3742</t>
3743</section>
3744
3745<section title="Registration" anchor="transfer.coding.registration">
3746<t>
3747   The HTTP Transfer Coding Registry shall be updated with the registrations
3748   below:
3749</t>
3750<texttable align="left" suppress-title="true" anchor="iana.transfer.coding.registration.table">
3751   <ttcol>Name</ttcol>
3752   <ttcol>Description</ttcol>
3753   <ttcol>Reference</ttcol>
3754   <c>chunked</c>
3755   <c>Transfer in a series of chunks</c>
3756   <c>
3757      <xref target="chunked.encoding"/>
3758   </c>
3759   <c>compress</c>
3760   <c>UNIX "compress" data format <xref target="Welch"/></c>
3761   <c>
3762      <xref target="compress.coding"/>
3763   </c>
3764   <c>deflate</c>
3765   <c>"deflate" compressed data (<xref target="RFC1951"/>) inside
3766   the "zlib" data format (<xref target="RFC1950"/>)
3767   </c>
3768   <c>
3769      <xref target="deflate.coding"/>
3770   </c>
3771   <c>gzip</c>
3772   <c>GZIP file format <xref target="RFC1952"/></c>
3773   <c>
3774      <xref target="gzip.coding"/>
3775   </c>
3776   <c>x-compress</c>
3777   <c>Deprecated (alias for compress)</c>
3778   <c>
3779      <xref target="compress.coding"/>
3780   </c>
3781   <c>x-gzip</c>
3782   <c>Deprecated (alias for gzip)</c>
3783   <c>
3784      <xref target="gzip.coding"/>
3785   </c>
3786</texttable>
3787</section>
3788</section>
3789
3790<section title="Content Coding Registration" anchor="content.coding.registration">
3791<t>
3792   IANA maintains the registry of HTTP Content Codings at
3793   <eref target="http://www.iana.org/assignments/http-parameters"/>.
3794</t>
3795<t>
3796   The HTTP Content Codings Registry shall be updated with the registrations
3797   below:
3798</t>
3799<texttable align="left" suppress-title="true" anchor="iana.content.coding.registration.table">
3800   <ttcol>Name</ttcol>
3801   <ttcol>Description</ttcol>
3802   <ttcol>Reference</ttcol>
3803   <c>compress</c>
3804   <c>UNIX "compress" data format <xref target="Welch"/></c>
3805   <c>
3806      <xref target="compress.coding"/>
3807   </c>
3808   <c>deflate</c>
3809   <c>"deflate" compressed data (<xref target="RFC1951"/>) inside
3810   the "zlib" data format (<xref target="RFC1950"/>)</c>
3811   <c>
3812      <xref target="deflate.coding"/>
3813   </c>
3814   <c>gzip</c>
3815   <c>GZIP file format <xref target="RFC1952"/></c>
3816   <c>
3817      <xref target="gzip.coding"/>
3818   </c>
3819   <c>x-compress</c>
3820   <c>Deprecated (alias for compress)</c>
3821   <c>
3822      <xref target="compress.coding"/>
3823   </c>
3824   <c>x-gzip</c>
3825   <c>Deprecated (alias for gzip)</c>
3826   <c>
3827      <xref target="gzip.coding"/>
3828   </c>
3829</texttable>
3830</section>
3831
3832<section title="Upgrade Token Registry" anchor="upgrade.token.registry">
3833<t>
3834   The HTTP Upgrade Token Registry defines the name space for protocol-name
3835   tokens used to identify protocols in the <x:ref>Upgrade</x:ref> header
3836   field. The registry is maintained at <eref target="http://www.iana.org/assignments/http-upgrade-tokens"/>.
3837</t>
3838
3839<section title="Procedure" anchor="upgrade.token.registry.procedure">  
3840<t>
3841   Each registered protocol name is associated with contact information
3842   and an optional set of specifications that details how the connection
3843   will be processed after it has been upgraded.
3844</t>
3845<t>
3846   Registrations happen on a "First Come First Served" basis (see
3847   <xref target="RFC5226" x:sec="4.1" x:fmt="of"/>) and are subject to the
3848   following rules:
3849  <list style="numbers">
3850    <t>A protocol-name token, once registered, stays registered forever.</t>
3851    <t>The registration &MUST; name a responsible party for the
3852       registration.</t>
3853    <t>The registration &MUST; name a point of contact.</t>
3854    <t>The registration &MAY; name a set of specifications associated with
3855       that token. Such specifications need not be publicly available.</t>
3856    <t>The registration &SHOULD; name a set of expected "protocol-version"
3857       tokens associated with that token at the time of registration.</t>
3858    <t>The responsible party &MAY; change the registration at any time.
3859       The IANA will keep a record of all such changes, and make them
3860       available upon request.</t>
3861    <t>The IESG &MAY; reassign responsibility for a protocol token.
3862       This will normally only be used in the case when a
3863       responsible party cannot be contacted.</t>
3864  </list>
3865</t>
3866<t>
3867   This registration procedure for HTTP Upgrade Tokens replaces that
3868   previously defined in <xref target="RFC2817" x:fmt="of" x:sec="7.2"/>.
3869</t>
3870</section>
3871
3872<section title="Upgrade Token Registration" anchor="upgrade.token.registration">
3873<t>
3874   The "HTTP" entry in the HTTP Upgrade Token Registry shall be updated with
3875   the registration below:
3876</t>
3877<texttable align="left" suppress-title="true">
3878   <ttcol>Value</ttcol>
3879   <ttcol>Description</ttcol>
3880   <ttcol>Expected Version Tokens</ttcol>
3881   <ttcol>Reference</ttcol>
3882
3883   <c>HTTP</c>
3884   <c>Hypertext Transfer Protocol</c>
3885   <c>any DIGIT.DIGIT (e.g, "2.0")</c>
3886   <c><xref target="http.version"/></c>
3887</texttable>
3888<t>
3889   The responsible party is: "IETF (iesg@ietf.org) - Internet Engineering Task Force".
3890</t>
3891</section>
3892</section>
3893
3894</section>
3895
3896<section title="Security Considerations" anchor="security.considerations">
3897<t>
3898   This section is meant to inform developers, information providers, and
3899   users of known security considerations relevant to HTTP message syntax,
3900   parsing, and routing. Security considerations about HTTP semantics and
3901   payloads are addressed in &semantics;.
3902</t>
3903
3904<section title="Establishing Authority" anchor="establishing.authority">
3905  <iref item="authoritative response" primary="true"/>
3906  <iref item="phishing" primary="true"/>
3907<t>
3908   HTTP relies on the notion of an <x:dfn>authoritative response</x:dfn>: a
3909   response that has been determined by (or at the direction of) the authority
3910   identified within the target URI to be the most appropriate response for
3911   that request given the state of the target resource at the time of
3912   response message origination. Providing a response from a non-authoritative
3913   source, such as a shared cache, is often useful to improve performance and
3914   availability, but only to the extent that the source can be trusted or
3915   the distrusted response can be safely used.
3916</t>
3917<t>
3918   Unfortunately, establishing authority can be difficult.
3919   For example, <x:dfn>phishing</x:dfn> is an attack on the user's perception
3920   of authority, where that perception can be misled by presenting similar
3921   branding in hypertext, possibly aided by userinfo obfuscating the authority
3922   component (see <xref target="http.uri"/>).
3923   User agents can reduce the impact of phishing attacks by enabling users to
3924   easily inspect a target URI prior to making an action, by prominently
3925   distinguishing (or rejecting) userinfo when present, and by not sending
3926   stored credentials and cookies when the referring document is from an
3927   unknown or untrusted source.
3928</t>
3929<t>
3930   When a registered name is used in the authority component, the "http" URI
3931   scheme (<xref target="http.uri"/>) relies on the user's local name
3932   resolution service to determine where it can find authoritative responses.
3933   This means that any attack on a user's network host table, cached names, or
3934   name resolution libraries becomes an avenue for attack on establishing
3935   authority. Likewise, the user's choice of server for Domain Name Service
3936   (DNS), and the hierarchy of servers from which it obtains resolution
3937   results, could impact the authenticity of address mappings;
3938   DNSSEC (<xref target="RFC4033"/>) is one way to improve authenticity.
3939</t>
3940<t>
3941   Furthermore, after an IP address is obtained, establishing authority for
3942   an "http" URI is vulnerable to attacks on Internet Protocol routing.
3943</t>
3944<t>
3945   The "https" scheme (<xref target="https.uri"/>) is intended to prevent
3946   (or at least reveal) many of these potential attacks on establishing
3947   authority, provided that the negotiated TLS connection is secured and
3948   the client properly verifies that the communicating server's identity
3949   matches the target URI's authority component
3950   (see <xref target="RFC2818"/>). Correctly implementing such verification
3951   can be difficult (see <xref target="Georgiev"/>).
3952</t>
3953</section>
3954
3955<section title="Risks of Intermediaries" anchor="risks.intermediaries">
3956<t>
3957   By their very nature, HTTP intermediaries are men-in-the-middle, and thus
3958   represent an opportunity for man-in-the-middle attacks. Compromise of
3959   the systems on which the intermediaries run can result in serious security
3960   and privacy problems. Intermediaries might have access to security-related
3961   information, personal information about individual users and
3962   organizations, and proprietary information belonging to users and
3963   content providers. A compromised intermediary, or an intermediary
3964   implemented or configured without regard to security and privacy
3965   considerations, might be used in the commission of a wide range of
3966   potential attacks.
3967</t>
3968<t>
3969   Intermediaries that contain a shared cache are especially vulnerable
3970   to cache poisoning attacks, as described in &cache-poisoning;.
3971</t>
3972<t>
3973   Implementers need to consider the privacy and security
3974   implications of their design and coding decisions, and of the
3975   configuration options they provide to operators (especially the
3976   default configuration).
3977</t>
3978<t>
3979   Users need to be aware that intermediaries are no more trustworthy than
3980   the people who run them; HTTP itself cannot solve this problem.
3981</t>
3982</section>
3983
3984<section title="Attacks via Protocol Element Length" anchor="attack.protocol.element.length">
3985<t>
3986   Because HTTP uses mostly textual, character-delimited fields, parsers are
3987   often vulnerable to attacks based on sending very long (or very slow)
3988   streams of data, particularly where an implementation is expecting a
3989   protocol element with no predefined length.
3990</t>
3991<t>
3992   To promote interoperability, specific recommendations are made for minimum
3993   size limits on request-line (<xref target="request.line"/>)
3994   and header fields (<xref target="header.fields"/>). These are
3995   minimum recommendations, chosen to be supportable even by implementations
3996   with limited resources; it is expected that most implementations will
3997   choose substantially higher limits.
3998</t>
3999<t>
4000   A server can reject messages that
4001   have request-targets that are too long (&status-414;) or request entities
4002   that are too large (&status-4xx;). Additional status codes related to
4003   capacity limits have been defined by extensions to HTTP
4004   <xref target="RFC6585"/>.
4005</t>
4006<t>
4007   Recipients ought to carefully limit the extent to which they process other
4008   protocol elements, including (but not limited to) request methods, response
4009   status phrases, header field-names, numeric values, and body chunks.
4010   Failure to limit such processing can result in buffer overflows, arithmetic
4011   overflows, or increased vulnerability to denial of service attacks.
4012</t>
4013</section>
4014
4015<section title="Response Splitting" anchor="response.splitting">
4016<t>
4017   Response splitting (a.k.a, CRLF injection) is a common technique, used in
4018   various attacks on Web usage, that exploits the line-based nature of HTTP
4019   message framing and the ordered association of requests to responses on
4020   persistent connections <xref target="Klein"/>. This technique can be
4021   particularly damaging when the requests pass through a shared cache.
4022</t>
4023<t>
4024   Response splitting exploits a vulnerability in servers (usually within an
4025   application server) where an attacker can send encoded data within some
4026   parameter of the request that is later decoded and echoed within any of the
4027   response header fields of the response. If the decoded data is crafted to
4028   look like the response has ended and a subsequent response has begun, the
4029   response has been split and the content within the apparent second response
4030   is controlled by the attacker. The attacker can then make any other request
4031   on the same persistent connection and trick the recipients (including
4032   intermediaries) into believing that the second half of the split is an
4033   authoritative answer to the second request.
4034</t>
4035<t>
4036   For example, a parameter within the request-target might be read by an
4037   application server and reused within a redirect, resulting in the same
4038   parameter being echoed in the <x:ref>Location</x:ref> header field of the
4039   response. If the parameter is decoded by the application and not properly
4040   encoded when placed in the response field, the attacker can send encoded
4041   CRLF octets and other content that will make the application's single
4042   response look like two or more responses.
4043</t>
4044<t>
4045   A common defense against response splitting is to filter requests for data
4046   that looks like encoded CR and LF (e.g., "%0D" and "%0A"). However, that
4047   assumes the application server is only performing URI decoding, rather
4048   than more obscure data transformations like charset transcoding, XML entity
4049   translation, base64 decoding, sprintf reformatting, etc.  A more effective
4050   mitigation is to prevent anything other than the server's core protocol
4051   libraries from sending a CR or LF within the header section, which means
4052   restricting the output of header fields to APIs that filter for bad octets
4053   and not allowing application servers to write directly to the protocol
4054   stream.
4055</t>
4056</section>
4057
4058<section title="Request Smuggling" anchor="request.smuggling">
4059<t>
4060   Request smuggling (<xref target="Linhart"/>) is a technique that exploits
4061   differences in protocol parsing among various recipients to hide additional
4062   requests (which might otherwise be blocked or disabled by policy) within an
4063   apparently harmless request.  Like response splitting, request smuggling
4064   can lead to a variety of attacks on HTTP usage.
4065</t>
4066<t>
4067   This specification has introduced new requirements on request parsing,
4068   particularly with regard to message framing in
4069   <xref target="message.body.length"/>, to reduce the effectiveness of
4070   request smuggling.
4071</t>
4072</section>
4073
4074<section title="Message Integrity" anchor="message.integrity">
4075<t>
4076   HTTP does not define a specific mechanism for ensuring message integrity,
4077   instead relying on the error-detection ability of underlying transport
4078   protocols and the use of length or chunk-delimited framing to detect
4079   completeness. Additional integrity mechanisms, such as hash functions or
4080   digital signatures applied to the content, can be selectively added to
4081   messages via extensible metadata header fields. Historically, the lack of
4082   a single integrity mechanism has been justified by the informal nature of
4083   most HTTP communication.  However, the prevalence of HTTP as an information
4084   access mechanism has resulted in its increasing use within environments
4085   where verification of message integrity is crucial.
4086</t>
4087<t>
4088   User agents are encouraged to implement configurable means for detecting
4089   and reporting failures of message integrity such that those means can be
4090   enabled within environments for which integrity is necessary. For example,
4091   a browser being used to view medical history or drug interaction
4092   information needs to indicate to the user when such information is detected
4093   by the protocol to be incomplete, expired, or corrupted during transfer.
4094   Such mechanisms might be selectively enabled via user agent extensions or
4095   the presence of message integrity metadata in a response.
4096   At a minimum, user agents ought to provide some indication that allows a
4097   user to distinguish between a complete and incomplete response message
4098   (<xref target="incomplete.messages"/>) when such verification is desired.
4099</t>
4100</section>
4101
4102<section title="Message Confidentiality" anchor="message.confidentiality">
4103<t>
4104   HTTP relies on underlying transport protocols to provide message
4105   confidentiality when that is desired. HTTP has been specifically designed
4106   to be independent of the transport protocol, such that it can be used
4107   over many different forms of encrypted connection, with the selection of
4108   such transports being identified by the choice of URI scheme or within
4109   user agent configuration.
4110</t>
4111<t>
4112   The "https" scheme can be used to identify resources that require a
4113   confidential connection, as described in <xref target="https.uri"/>.
4114</t>
4115</section>
4116
4117<section title="Privacy of Server Log Information" anchor="privacy.of.server.log.information">
4118<t>
4119   A server is in the position to save personal data about a user's requests
4120   over time, which might identify their reading patterns or subjects of
4121   interest.  In particular, log information gathered at an intermediary
4122   often contains a history of user agent interaction, across a multitude
4123   of sites, that can be traced to individual users.
4124</t>
4125<t>
4126   HTTP log information is confidential in nature; its handling is often
4127   constrained by laws and regulations.  Log information needs to be securely
4128   stored and appropriate guidelines followed for its analysis.
4129   Anonymization of personal information within individual entries helps,
4130   but is generally not sufficient to prevent real log traces from being
4131   re-identified based on correlation with other access characteristics.
4132   As such, access traces that are keyed to a specific client are unsafe to
4133   publish even if the key is pseudonymous.
4134</t>
4135<t>
4136   To minimize the risk of theft or accidental publication, log information
4137   ought to be purged of personally identifiable information, including
4138   user identifiers, IP addresses, and user-provided query parameters,
4139   as soon as that information is no longer necessary to support operational
4140   needs for security, auditing, or fraud control.
4141</t>
4142</section>
4143</section>
4144
4145<section title="Acknowledgments" anchor="acks">
4146<t>
4147   This edition of HTTP/1.1 builds on the many contributions that went into
4148   <xref target="RFC1945" format="none">RFC 1945</xref>,
4149   <xref target="RFC2068" format="none">RFC 2068</xref>,
4150   <xref target="RFC2145" format="none">RFC 2145</xref>, and
4151   <xref target="RFC2616" format="none">RFC 2616</xref>, including
4152   substantial contributions made by the previous authors, editors, and
4153   working group chairs: Tim Berners-Lee, Ari Luotonen, Roy T. Fielding,
4154   Henrik Frystyk Nielsen, Jim Gettys, Jeffrey C. Mogul, Larry Masinter,
4155   and Paul J. Leach. Mark Nottingham oversaw this effort as working group chair.
4156</t>
4157<t>
4158   Since 1999, the following contributors have helped improve the HTTP
4159   specification by reporting bugs, asking smart questions, drafting or
4160   reviewing text, and evaluating open issues:
4161</t>
4162<?BEGININC acks ?>
4163<t>Adam Barth,
4164Adam Roach,
4165Addison Phillips,
4166Adrian Chadd,
4167Adrian Cole,
4168Adrien W. de Croy,
4169Alan Ford,
4170Alan Ruttenberg,
4171Albert Lunde,
4172Alek Storm,
4173Alex Rousskov,
4174Alexandre Morgaut,
4175Alexey Melnikov,
4176Alisha Smith,
4177Amichai Rothman,
4178Amit Klein,
4179Amos Jeffries,
4180Andreas Maier,
4181Andreas Petersson,
4182Andrei Popov,
4183Anil Sharma,
4184Anne van Kesteren,
4185Anthony Bryan,
4186Asbjorn Ulsberg,
4187Ashok Kumar,
4188Balachander Krishnamurthy,
4189Barry Leiba,
4190Ben Laurie,
4191Benjamin Carlyle,
4192Benjamin Niven-Jenkins,
4193Benoit Claise,
4194Bil Corry,
4195Bill Burke,
4196Bjoern Hoehrmann,
4197Bob Scheifler,
4198Boris Zbarsky,
4199Brett Slatkin,
4200Brian Kell,
4201Brian McBarron,
4202Brian Pane,
4203Brian Raymor,
4204Brian Smith,
4205Bruce Perens,
4206Bryce Nesbitt,
4207Cameron Heavon-Jones,
4208Carl Kugler,
4209Carsten Bormann,
4210Charles Fry,
4211Chris Burdess,
4212Chris Newman,
4213Christian Huitema,
4214Cyrus Daboo,
4215Dale Robert Anderson,
4216Dan Wing,
4217Dan Winship,
4218Daniel Stenberg,
4219Darrel Miller,
4220Dave Cridland,
4221Dave Crocker,
4222Dave Kristol,
4223Dave Thaler,
4224David Booth,
4225David Singer,
4226David W. Morris,
4227Diwakar Shetty,
4228Dmitry Kurochkin,
4229Drummond Reed,
4230Duane Wessels,
4231Edward Lee,
4232Eitan Adler,
4233Eliot Lear,
4234Emile Stephan,
4235Eran Hammer-Lahav,
4236Eric D. Williams,
4237Eric J. Bowman,
4238Eric Lawrence,
4239Eric Rescorla,
4240Erik Aronesty,
4241EungJun Yi,
4242Evan Prodromou,
4243Felix Geisendoerfer,
4244Florian Weimer,
4245Frank Ellermann,
4246Fred Akalin,
4247Fred Bohle,
4248Frederic Kayser,
4249Gabor Molnar,
4250Gabriel Montenegro,
4251Geoffrey Sneddon,
4252Gervase Markham,
4253Gili Tzabari,
4254Grahame Grieve,
4255Greg Slepak,
4256Greg Wilkins,
4257Grzegorz Calkowski,
4258Harald Tveit Alvestrand,
4259Harry Halpin,
4260Helge Hess,
4261Henrik Nordstrom,
4262Henry S. Thompson,
4263Henry Story,
4264Herbert van de Sompel,
4265Herve Ruellan,
4266Howard Melman,
4267Hugo Haas,
4268Ian Fette,
4269Ian Hickson,
4270Ido Safruti,
4271Ilari Liusvaara,
4272Ilya Grigorik,
4273Ingo Struck,
4274J. Ross Nicoll,
4275James Cloos,
4276James H. Manger,
4277James Lacey,
4278James M. Snell,
4279Jamie Lokier,
4280Jan Algermissen,
4281Jari Arkko,
4282Jeff Hodges (who came up with the term 'effective Request-URI'),
4283Jeff Pinner,
4284Jeff Walden,
4285Jim Luther,
4286Jitu Padhye,
4287Joe D. Williams,
4288Joe Gregorio,
4289Joe Orton,
4290Joel Jaeggli,
4291John C. Klensin,
4292John C. Mallery,
4293John Cowan,
4294John Kemp,
4295John Panzer,
4296John Schneider,
4297John Stracke,
4298John Sullivan,
4299Jonas Sicking,
4300Jonathan A. Rees,
4301Jonathan Billington,
4302Jonathan Moore,
4303Jonathan Silvera,
4304Jordi Ros,
4305Joris Dobbelsteen,
4306Josh Cohen,
4307Julien Pierre,
4308Jungshik Shin,
4309Justin Chapweske,
4310Justin Erenkrantz,
4311Justin James,
4312Kalvinder Singh,
4313Karl Dubost,
4314Kathleen Moriarty,
4315Keith Hoffman,
4316Keith Moore,
4317Ken Murchison,
4318Koen Holtman,
4319Konstantin Voronkov,
4320Kris Zyp,
4321Leif Hedstrom,
4322Lionel Morand,
4323Lisa Dusseault,
4324Maciej Stachowiak,
4325Manu Sporny,
4326Marc Schneider,
4327Marc Slemko,
4328Mark Baker,
4329Mark Pauley,
4330Mark Watson,
4331Markus Isomaki,
4332Markus Lanthaler,
4333Martin J. Duerst,
4334Martin Musatov,
4335Martin Nilsson,
4336Martin Thomson,
4337Matt Lynch,
4338Matthew Cox,
4339Matthew Kerwin,
4340Max Clark,
4341Menachem Dodge,
4342Meral Shirazipour,
4343Michael Burrows,
4344Michael Hausenblas,
4345Michael Scharf,
4346Michael Sweet,
4347Michael Tuexen,
4348Michael Welzl,
4349Mike Amundsen,
4350Mike Belshe,
4351Mike Bishop,
4352Mike Kelly,
4353Mike Schinkel,
4354Miles Sabin,
4355Murray S. Kucherawy,
4356Mykyta Yevstifeyev,
4357Nathan Rixham,
4358Nicholas Shanks,
4359Nico Williams,
4360Nicolas Alvarez,
4361Nicolas Mailhot,
4362Noah Slater,
4363Osama Mazahir,
4364Pablo Castro,
4365Pat Hayes,
4366Patrick R. McManus,
4367Paul E. Jones,
4368Paul Hoffman,
4369Paul Marquess,
4370Pete Resnick,
4371Peter Lepeska,
4372Peter Occil,
4373Peter Saint-Andre,
4374Peter Watkins,
4375Phil Archer,
4376Phil Hunt,
4377Philippe Mougin,
4378Phillip Hallam-Baker,
4379Piotr Dobrogost,
4380Poul-Henning Kamp,
4381Preethi Natarajan,
4382Rajeev Bector,
4383Ray Polk,
4384Reto Bachmann-Gmuer,
4385Richard Barnes,
4386Richard Cyganiak,
4387Rob Trace,
4388Robby Simpson,
4389Robert Brewer,
4390Robert Collins,
4391Robert Mattson,
4392Robert O'Callahan,
4393Robert Olofsson,
4394Robert Sayre,
4395Robert Siemer,
4396Robert de Wilde,
4397Roberto Javier Godoy,
4398Roberto Peon,
4399Roland Zink,
4400Ronny Widjaja,
4401Ryan Hamilton,
4402S. Mike Dierken,
4403Salvatore Loreto,
4404Sam Johnston,
4405Sam Pullara,
4406Sam Ruby,
4407Saurabh Kulkarni,
4408Scott Lawrence (who maintained the original issues list),
4409Sean B. Palmer,
4410Sean Turner,
4411Sebastien Barnoud,
4412Shane McCarron,
4413Shigeki Ohtsu,
4414Simon Yarde,
4415Stefan Eissing,
4416Stefan Tilkov,
4417Stefanos Harhalakis,
4418Stephane Bortzmeyer,
4419Stephen Farrell,
4420Stephen Kent,
4421Stephen Ludin,
4422Stuart Williams,
4423Subbu Allamaraju,
4424Subramanian Moonesamy,
4425Susan Hares,
4426Sylvain Hellegouarch,
4427Tapan Divekar,
4428Tatsuhiro Tsujikawa,
4429Tatsuya Hayashi,
4430Ted Hardie,
4431Ted Lemon,
4432Thomas Broyer,
4433Thomas Fossati,
4434Thomas Maslen,
4435Thomas Nadeau,
4436Thomas Nordin,
4437Thomas Roessler,
4438Tim Bray,
4439Tim Morgan,
4440Tim Olsen,
4441Tom Zhou,
4442Travis Snoozy,
4443Tyler Close,
4444Vincent Murphy,
4445Wenbo Zhu,
4446Werner Baumann,
4447Wilbur Streett,
4448Wilfredo Sanchez Vega,
4449William A. Rowe Jr.,
4450William Chan,
4451Willy Tarreau,
4452Xiaoshu Wang,
4453Yaron Goland,
4454Yngve Nysaeter Pettersen,
4455Yoav Nir,
4456Yogesh Bang,
4457Yuchung Cheng,
4458Yutaka Oiwa,
4459Yves Lafon (long-time member of the editor team),
4460Zed A. Shaw, and
4461Zhong Yu.
4462</t>
4463<?ENDINC acks ?>
4464<t>
4465   See <xref target="RFC2616" x:fmt="of" x:sec="16"/> for additional
4466   acknowledgements from prior revisions.
4467</t>
4468</section>
4469
4470</middle>
4471<back>
4472
4473<references title="Normative References">
4474
4475<reference anchor="Part2">
4476  <front>
4477    <title>Hypertext Transfer Protocol (HTTP/1.1): Semantics and Content</title>
4478    <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
4479      <organization abbrev="Adobe">Adobe Systems Incorporated</organization>
4480      <address><email>fielding@gbiv.com</email></address>
4481    </author>
4482    <author initials="J. F." surname="Reschke" fullname="Julian F. Reschke" role="editor">
4483      <organization abbrev="greenbytes">greenbytes GmbH</organization>
4484      <address><email>julian.reschke@greenbytes.de</email></address>
4485    </author>
4486    <date month="&ID-MONTH;" year="&ID-YEAR;"/>
4487  </front>
4488  <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p2-semantics-&ID-VERSION;"/>
4489  <x:source href="p2-semantics.xml" basename="p2-semantics">
4490    <x:defines>1xx (Informational)</x:defines>
4491    <x:defines>1xx</x:defines>
4492    <x:defines>100 (Continue)</x:defines>
4493    <x:defines>101 (Switching Protocols)</x:defines>
4494    <x:defines>2xx (Successful)</x:defines>
4495    <x:defines>2xx</x:defines>
4496    <x:defines>200 (OK)</x:defines>
4497    <x:defines>203 (Non-Authoritative Information)</x:defines>
4498    <x:defines>204 (No Content)</x:defines>
4499    <x:defines>3xx (Redirection)</x:defines>
4500    <x:defines>3xx</x:defines>
4501    <x:defines>301 (Moved Permanently)</x:defines>
4502    <x:defines>4xx (Client Error)</x:defines>
4503    <x:defines>4xx</x:defines>
4504    <x:defines>400 (Bad Request)</x:defines>
4505    <x:defines>411 (Length Required)</x:defines>
4506    <x:defines>414 (URI Too Long)</x:defines>
4507    <x:defines>417 (Expectation Failed)</x:defines>
4508    <x:defines>426 (Upgrade Required)</x:defines>
4509    <x:defines>501 (Not Implemented)</x:defines>
4510    <x:defines>502 (Bad Gateway)</x:defines>
4511    <x:defines>505 (HTTP Version Not Supported)</x:defines>
4512    <x:defines>Accept-Encoding</x:defines>
4513    <x:defines>Allow</x:defines>
4514    <x:defines>Content-Encoding</x:defines>
4515    <x:defines>Content-Location</x:defines>
4516    <x:defines>Content-Type</x:defines>
4517    <x:defines>Date</x:defines>
4518    <x:defines>Expect</x:defines>
4519    <x:defines>Location</x:defines>
4520    <x:defines>Server</x:defines>
4521    <x:defines>User-Agent</x:defines>
4522  </x:source>
4523</reference>
4524
4525<reference anchor="Part4">
4526  <front>
4527    <title>Hypertext Transfer Protocol (HTTP/1.1): Conditional Requests</title>
4528    <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
4529      <organization abbrev="Adobe">Adobe Systems Incorporated</organization>
4530      <address><email>fielding@gbiv.com</email></address>
4531    </author>
4532    <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
4533      <organization abbrev="greenbytes">greenbytes GmbH</organization>
4534      <address><email>julian.reschke@greenbytes.de</email></address>
4535    </author>
4536    <date month="&ID-MONTH;" year="&ID-YEAR;" />
4537  </front>
4538  <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p4-conditional-&ID-VERSION;" />
4539  <x:source basename="p4-conditional" href="p4-conditional.xml">
4540    <x:defines>304 (Not Modified)</x:defines>
4541    <x:defines>ETag</x:defines>
4542    <x:defines>Last-Modified</x:defines>
4543  </x:source>
4544</reference>
4545
4546<reference anchor="Part5">
4547  <front>
4548    <title>Hypertext Transfer Protocol (HTTP/1.1): Range Requests</title>
4549    <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
4550      <organization abbrev="Adobe">Adobe Systems Incorporated</organization>
4551      <address><email>fielding@gbiv.com</email></address>
4552    </author>
4553    <author initials="Y." surname="Lafon" fullname="Yves Lafon" role="editor">
4554      <organization abbrev="W3C">World Wide Web Consortium</organization>
4555      <address><email>ylafon@w3.org</email></address>
4556    </author>
4557    <author initials="J. F." surname="Reschke" fullname="Julian F. Reschke" role="editor">
4558      <organization abbrev="greenbytes">greenbytes GmbH</organization>
4559      <address><email>julian.reschke@greenbytes.de</email></address>
4560    </author>
4561    <date month="&ID-MONTH;" year="&ID-YEAR;"/>
4562  </front>
4563  <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p5-range-&ID-VERSION;"/>
4564  <x:source href="p5-range.xml" basename="p5-range">
4565    <x:defines>Content-Range</x:defines>
4566  </x:source>
4567</reference>
4568
4569<reference anchor="Part6">
4570  <front>
4571    <title>Hypertext Transfer Protocol (HTTP/1.1): Caching</title>
4572    <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
4573      <organization abbrev="Adobe">Adobe Systems Incorporated</organization>
4574      <address><email>fielding@gbiv.com</email></address>
4575    </author>
4576    <author initials="M." surname="Nottingham" fullname="Mark Nottingham" role="editor">
4577      <organization>Akamai</organization>
4578      <address><email>mnot@mnot.net</email></address>
4579    </author>
4580    <author initials="J. F." surname="Reschke" fullname="Julian F. Reschke" role="editor">
4581      <organization abbrev="greenbytes">greenbytes GmbH</organization>
4582      <address><email>julian.reschke@greenbytes.de</email></address>
4583    </author>
4584    <date month="&ID-MONTH;" year="&ID-YEAR;"/>
4585  </front>
4586  <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p6-cache-&ID-VERSION;"/>
4587  <x:source href="p6-cache.xml" basename="p6-cache">
4588    <x:defines>Cache-Control</x:defines>
4589    <x:defines>Expires</x:defines>
4590    <x:defines>Warning</x:defines>
4591  </x:source>
4592</reference>
4593
4594<reference anchor="Part7">
4595  <front>
4596    <title>Hypertext Transfer Protocol (HTTP/1.1): Authentication</title>
4597    <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
4598      <organization abbrev="Adobe">Adobe Systems Incorporated</organization>
4599      <address><email>fielding@gbiv.com</email></address>
4600    </author>
4601    <author initials="J. F." surname="Reschke" fullname="Julian F. Reschke" role="editor">
4602      <organization abbrev="greenbytes">greenbytes GmbH</organization>
4603      <address><email>julian.reschke@greenbytes.de</email></address>
4604    </author>
4605    <date month="&ID-MONTH;" year="&ID-YEAR;"/>
4606  </front>
4607  <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p7-auth-&ID-VERSION;"/>
4608  <x:source href="p7-auth.xml" basename="p7-auth">
4609    <x:defines>Proxy-Authenticate</x:defines>
4610    <x:defines>Proxy-Authorization</x:defines>
4611  </x:source>
4612</reference>
4613
4614<reference anchor="RFC5234">
4615  <front>
4616    <title abbrev="ABNF for Syntax Specifications">Augmented BNF for Syntax Specifications: ABNF</title>
4617    <author initials="D." surname="Crocker" fullname="Dave Crocker" role="editor">
4618      <organization>Brandenburg InternetWorking</organization>
4619      <address>
4620        <email>dcrocker@bbiw.net</email>
4621      </address> 
4622    </author>
4623    <author initials="P." surname="Overell" fullname="Paul Overell">
4624      <organization>THUS plc.</organization>
4625      <address>
4626        <email>paul.overell@thus.net</email>
4627      </address>
4628    </author>
4629    <date month="January" year="2008"/>
4630  </front>
4631  <seriesInfo name="STD" value="68"/>
4632  <seriesInfo name="RFC" value="5234"/>
4633</reference>
4634
4635<reference anchor="RFC2119">
4636  <front>
4637    <title>Key words for use in RFCs to Indicate Requirement Levels</title>
4638    <author initials="S." surname="Bradner" fullname="Scott Bradner">
4639      <organization>Harvard University</organization>
4640      <address><email>sob@harvard.edu</email></address>
4641    </author>
4642    <date month="March" year="1997"/>
4643  </front>
4644  <seriesInfo name="BCP" value="14"/>
4645  <seriesInfo name="RFC" value="2119"/>
4646</reference>
4647
4648<reference anchor="RFC3986">
4649 <front>
4650  <title abbrev='URI Generic Syntax'>Uniform Resource Identifier (URI): Generic Syntax</title>
4651  <author initials='T.' surname='Berners-Lee' fullname='Tim Berners-Lee'>
4652    <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
4653    <address>
4654       <email>timbl@w3.org</email>
4655       <uri>http://www.w3.org/People/Berners-Lee/</uri>
4656    </address>
4657  </author>
4658  <author initials='R.' surname='Fielding' fullname='Roy T. Fielding'>
4659    <organization abbrev="Day Software">Day Software</organization>
4660    <address>
4661      <email>fielding@gbiv.com</email>
4662      <uri>http://roy.gbiv.com/</uri>
4663    </address>
4664  </author>
4665  <author initials='L.' surname='Masinter' fullname='Larry Masinter'>
4666    <organization abbrev="Adobe Systems">Adobe Systems Incorporated</organization>
4667    <address>
4668      <email>LMM@acm.org</email>
4669      <uri>http://larry.masinter.net/</uri>
4670    </address>
4671  </author>
4672  <date month='January' year='2005'></date>
4673 </front>
4674 <seriesInfo name="STD" value="66"/>
4675 <seriesInfo name="RFC" value="3986"/>
4676</reference>
4677
4678<reference anchor="RFC0793">
4679  <front>
4680    <title>Transmission Control Protocol</title>
4681    <author initials='J.' surname='Postel' fullname='Jon Postel'>
4682      <organization>University of Southern California (USC)/Information Sciences Institute</organization>
4683    </author>
4684    <date year='1981' month='September' />
4685  </front>
4686  <seriesInfo name='STD' value='7' />
4687  <seriesInfo name='RFC' value='793' />
4688</reference>
4689
4690<reference anchor="USASCII">
4691  <front>
4692    <title>Coded Character Set -- 7-bit American Standard Code for Information Interchange</title>
4693    <author>
4694      <organization>American National Standards Institute</organization>
4695    </author>
4696    <date year="1986"/>
4697  </front>
4698  <seriesInfo name="ANSI" value="X3.4"/>
4699</reference>
4700
4701<reference anchor="RFC1950">
4702  <front>
4703    <title>ZLIB Compressed Data Format Specification version 3.3</title>
4704    <author initials="L.P." surname="Deutsch" fullname="L. Peter Deutsch">
4705      <organization>Aladdin Enterprises</organization>
4706      <address><email>ghost@aladdin.com</email></address>
4707    </author>
4708    <author initials="J-L." surname="Gailly" fullname="Jean-Loup Gailly"/>
4709    <date month="May" year="1996"/>
4710  </front>
4711  <seriesInfo name="RFC" value="1950"/>
4712  <!--<annotation>
4713    RFC 1950 is an Informational RFC, thus it might be less stable than
4714    this specification. On the other hand, this downward reference was
4715    present since the publication of <xref target="RFC2068" x:fmt="none">RFC 2068</xref> in 1997,
4716    therefore it is unlikely to cause problems in practice. See also
4717    <xref target="BCP97"/>.
4718  </annotation>-->
4719</reference>
4720
4721<reference anchor="RFC1951">
4722  <front>
4723    <title>DEFLATE Compressed Data Format Specification version 1.3</title>
4724    <author initials="P." surname="Deutsch" fullname="L. Peter Deutsch">
4725      <organization>Aladdin Enterprises</organization>
4726      <address><email>ghost@aladdin.com</email></address>
4727    </author>
4728    <date month="May" year="1996"/>
4729  </front>
4730  <seriesInfo name="RFC" value="1951"/>
4731  <!--<annotation>
4732    RFC 1951 is an Informational RFC, thus it might be less stable than
4733    this specification. On the other hand, this downward reference was
4734    present since the publication of <xref target="RFC2068" x:fmt="none">RFC 2068</xref> in 1997,
4735    therefore it is unlikely to cause problems in practice. See also
4736    <xref target="BCP97"/>.
4737  </annotation>-->
4738</reference>
4739
4740<reference anchor="RFC1952">
4741  <front>
4742    <title>GZIP file format specification version 4.3</title>
4743    <author initials="P." surname="Deutsch" fullname="L. Peter Deutsch">
4744      <organization>Aladdin Enterprises</organization>
4745      <address><email>ghost@aladdin.com</email></address>
4746    </author>
4747    <author initials="J-L." surname="Gailly" fullname="Jean-Loup Gailly">
4748      <address><email>gzip@prep.ai.mit.edu</email></address>
4749    </author>
4750    <author initials="M." surname="Adler" fullname="Mark Adler">
4751      <address><email>madler@alumni.caltech.edu</email></address>
4752    </author>
4753    <author initials="L.P." surname="Deutsch" fullname="L. Peter Deutsch">
4754      <address><email>ghost@aladdin.com</email></address>
4755    </author>
4756    <author initials="G." surname="Randers-Pehrson" fullname="Glenn Randers-Pehrson">
4757      <address><email>randeg@alumni.rpi.edu</email></address>
4758    </author>
4759    <date month="May" year="1996"/>
4760  </front>
4761  <seriesInfo name="RFC" value="1952"/>
4762  <!--<annotation>
4763    RFC 1952 is an Informational RFC, thus it might be less stable than
4764    this specification. On the other hand, this downward reference was
4765    present since the publication of <xref target="RFC2068" x:fmt="none">RFC 2068</xref> in 1997,
4766    therefore it is unlikely to cause problems in practice. See also
4767    <xref target="BCP97"/>.
4768  </annotation>-->
4769</reference>
4770
4771<reference anchor="Welch">
4772  <front>
4773    <title>A Technique for High Performance Data Compression</title>
4774    <author initials="T.A." surname="Welch" fullname="Terry A. Welch"/>
4775    <date month="June" year="1984"/>
4776  </front>
4777  <seriesInfo name="IEEE Computer" value="17(6)"/>
4778</reference>
4779
4780</references>
4781
4782<references title="Informative References">
4783
4784<reference anchor="ISO-8859-1">
4785  <front>
4786    <title>
4787     Information technology -- 8-bit single-byte coded graphic character sets -- Part 1: Latin alphabet No. 1
4788    </title>
4789    <author>
4790      <organization>International Organization for Standardization</organization>
4791    </author>
4792    <date year="1998"/>
4793  </front>
4794  <seriesInfo name="ISO/IEC" value="8859-1:1998"/>
4795</reference>
4796
4797<reference anchor='RFC1919'>
4798  <front>
4799    <title>Classical versus Transparent IP Proxies</title>
4800    <author initials='M.' surname='Chatel' fullname='Marc Chatel'>
4801      <address><email>mchatel@pax.eunet.ch</email></address>
4802    </author>
4803    <date year='1996' month='March' />
4804  </front>
4805  <seriesInfo name='RFC' value='1919' />
4806</reference>
4807
4808<reference anchor="RFC1945">
4809  <front>
4810    <title abbrev="HTTP/1.0">Hypertext Transfer Protocol -- HTTP/1.0</title>
4811    <author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
4812      <organization>MIT, Laboratory for Computer Science</organization>
4813      <address><email>timbl@w3.org</email></address>
4814    </author>
4815    <author initials="R.T." surname="Fielding" fullname="Roy T. Fielding">
4816      <organization>University of California, Irvine, Department of Information and Computer Science</organization>
4817      <address><email>fielding@ics.uci.edu</email></address>
4818    </author>
4819    <author initials="H.F." surname="Nielsen" fullname="Henrik Frystyk Nielsen">
4820      <organization>W3 Consortium, MIT Laboratory for Computer Science</organization>
4821      <address><email>frystyk@w3.org</email></address>
4822    </author>
4823    <date month="May" year="1996"/>
4824  </front>
4825  <seriesInfo name="RFC" value="1945"/>
4826</reference>
4827
4828<reference anchor="RFC2045">
4829  <front>
4830    <title abbrev="Internet Message Bodies">Multipurpose Internet Mail Extensions (MIME) Part One: Format of Internet Message Bodies</title>
4831    <author initials="N." surname="Freed" fullname="Ned Freed">
4832      <organization>Innosoft International, Inc.</organization>
4833      <address><email>ned@innosoft.com</email></address>
4834    </author>
4835    <author initials="N.S." surname="Borenstein" fullname="Nathaniel S. Borenstein">
4836      <organization>First Virtual Holdings</organization>
4837      <address><email>nsb@nsb.fv.com</email></address>
4838    </author>
4839    <date month="November" year="1996"/>
4840  </front>
4841  <seriesInfo name="RFC" value="2045"/>
4842</reference>
4843
4844<reference anchor="RFC2047">
4845  <front>
4846    <title abbrev="Message Header Extensions">MIME (Multipurpose Internet Mail Extensions) Part Three: Message Header Extensions for Non-ASCII Text</title>
4847    <author initials="K." surname="Moore" fullname="Keith Moore">
4848      <organization>University of Tennessee</organization>
4849      <address><email>moore@cs.utk.edu</email></address>
4850    </author>
4851    <date month="November" year="1996"/>
4852  </front>
4853  <seriesInfo name="RFC" value="2047"/>
4854</reference>
4855
4856<reference anchor="RFC2068">
4857  <front>
4858    <title>Hypertext Transfer Protocol -- HTTP/1.1</title>
4859    <author initials="R." surname="Fielding" fullname="Roy T. Fielding">
4860      <organization>University of California, Irvine, Department of Information and Computer Science</organization>
4861      <address><email>fielding@ics.uci.edu</email></address>
4862    </author>
4863    <author initials="J." surname="Gettys" fullname="Jim Gettys">
4864      <organization>MIT Laboratory for Computer Science</organization>
4865      <address><email>jg@w3.org</email></address>
4866    </author>
4867    <author initials="J." surname="Mogul" fullname="Jeffrey C. Mogul">
4868      <organization>Digital Equipment Corporation, Western Research Laboratory</organization>
4869      <address><email>mogul@wrl.dec.com</email></address>
4870    </author>
4871    <author initials="H." surname="Nielsen" fullname="Henrik Frystyk Nielsen">
4872      <organization>MIT Laboratory for Computer Science</organization>
4873      <address><email>frystyk@w3.org</email></address>
4874    </author>
4875    <author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
4876      <organization>MIT Laboratory for Computer Science</organization>
4877      <address><email>timbl@w3.org</email></address>
4878    </author>
4879    <date month="January" year="1997"/>
4880  </front>
4881  <seriesInfo name="RFC" value="2068"/>
4882</reference>
4883
4884<reference anchor="RFC2145">
4885  <front>
4886    <title abbrev="HTTP Version Numbers">Use and Interpretation of HTTP Version Numbers</title>
4887    <author initials="J.C." surname="Mogul" fullname="Jeffrey C. Mogul">
4888      <organization>Western Research Laboratory</organization>
4889      <address><email>mogul@wrl.dec.com</email></address>
4890    </author>
4891    <author initials="R.T." surname="Fielding" fullname="Roy T. Fielding">
4892      <organization>Department of Information and Computer Science</organization>
4893      <address><email>fielding@ics.uci.edu</email></address>
4894    </author>
4895    <author initials="J." surname="Gettys" fullname="Jim Gettys">
4896      <organization>MIT Laboratory for Computer Science</organization>
4897      <address><email>jg@w3.org</email></address>
4898    </author>
4899    <author initials="H.F." surname="Nielsen" fullname="Henrik Frystyk Nielsen">
4900      <organization>W3 Consortium</organization>
4901      <address><email>frystyk@w3.org</email></address>
4902    </author>
4903    <date month="May" year="1997"/>
4904  </front>
4905  <seriesInfo name="RFC" value="2145"/>
4906</reference>
4907
4908<reference anchor="RFC2616">
4909  <front>
4910    <title>Hypertext Transfer Protocol -- HTTP/1.1</title>
4911    <author initials="R." surname="Fielding" fullname="R. Fielding">
4912      <organization>University of California, Irvine</organization>
4913      <address><email>fielding@ics.uci.edu</email></address>
4914    </author>
4915    <author initials="J." surname="Gettys" fullname="J. Gettys">
4916      <organization>W3C</organization>
4917      <address><email>jg@w3.org</email></address>
4918    </author>
4919    <author initials="J." surname="Mogul" fullname="J. Mogul">
4920      <organization>Compaq Computer Corporation</organization>
4921      <address><email>mogul@wrl.dec.com</email></address>
4922    </author>
4923    <author initials="H." surname="Frystyk" fullname="H. Frystyk">
4924      <organization>MIT Laboratory for Computer Science</organization>
4925      <address><email>frystyk@w3.org</email></address>
4926    </author>
4927    <author initials="L." surname="Masinter" fullname="L. Masinter">
4928      <organization>Xerox Corporation</organization>
4929      <address><email>masinter@parc.xerox.com</email></address>
4930    </author>
4931    <author initials="P." surname="Leach" fullname="P. Leach">
4932      <organization>Microsoft Corporation</organization>
4933      <address><email>paulle@microsoft.com</email></address>
4934    </author>
4935    <author initials="T." surname="Berners-Lee" fullname="T. Berners-Lee">
4936      <organization>W3C</organization>
4937      <address><email>timbl@w3.org</email></address>
4938    </author>
4939    <date month="June" year="1999"/>
4940  </front>
4941  <seriesInfo name="RFC" value="2616"/>
4942</reference>
4943
4944<reference anchor='RFC2817'>
4945  <front>
4946    <title>Upgrading to TLS Within HTTP/1.1</title>
4947    <author initials='R.' surname='Khare' fullname='R. Khare'>
4948      <organization>4K Associates / UC Irvine</organization>
4949      <address><email>rohit@4K-associates.com</email></address>
4950    </author>
4951    <author initials='S.' surname='Lawrence' fullname='S. Lawrence'>
4952      <organization>Agranat Systems, Inc.</organization>
4953      <address><email>lawrence@agranat.com</email></address>
4954    </author>
4955    <date year='2000' month='May' />
4956  </front>
4957  <seriesInfo name='RFC' value='2817' />
4958</reference>
4959
4960<reference anchor='RFC2818'>
4961  <front>
4962    <title>HTTP Over TLS</title>
4963    <author initials='E.' surname='Rescorla' fullname='Eric Rescorla'>
4964      <organization>RTFM, Inc.</organization>
4965      <address><email>ekr@rtfm.com</email></address>
4966    </author>
4967    <date year='2000' month='May' />
4968  </front>
4969  <seriesInfo name='RFC' value='2818' />
4970</reference>
4971
4972<reference anchor='RFC3040'>
4973  <front>
4974    <title>Internet Web Replication and Caching Taxonomy</title>
4975    <author initials='I.' surname='Cooper' fullname='I. Cooper'>
4976      <organization>Equinix, Inc.</organization>
4977    </author>
4978    <author initials='I.' surname='Melve' fullname='I. Melve'>
4979      <organization>UNINETT</organization>
4980    </author>
4981    <author initials='G.' surname='Tomlinson' fullname='G. Tomlinson'>
4982      <organization>CacheFlow Inc.</organization>
4983    </author>
4984    <date year='2001' month='January' />
4985  </front>
4986  <seriesInfo name='RFC' value='3040' />
4987</reference>
4988
4989<reference anchor='BCP90'>
4990  <front>
4991    <title>Registration Procedures for Message Header Fields</title>
4992    <author initials='G.' surname='Klyne' fullname='G. Klyne'>
4993      <organization>Nine by Nine</organization>
4994      <address><email>GK-IETF@ninebynine.org</email></address>
4995    </author>
4996    <author initials='M.' surname='Nottingham' fullname='M. Nottingham'>
4997      <organization>BEA Systems</organization>
4998      <address><email>mnot@pobox.com</email></address>
4999    </author>
5000    <author initials='J.' surname='Mogul' fullname='J. Mogul'>
5001      <organization>HP Labs</organization>
5002      <address><email>JeffMogul@acm.org</email></address>
5003    </author>
5004    <date year='2004' month='September' />
5005  </front>
5006  <seriesInfo name='BCP' value='90' />
5007  <seriesInfo name='RFC' value='3864' />
5008</reference>
5009
5010<reference anchor='RFC4033'>
5011  <front>
5012    <title>DNS Security Introduction and Requirements</title>
5013    <author initials='R.' surname='Arends' fullname='R. Arends'/>
5014    <author initials='R.' surname='Austein' fullname='R. Austein'/>
5015    <author initials='M.' surname='Larson' fullname='M. Larson'/>
5016    <author initials='D.' surname='Massey' fullname='D. Massey'/>
5017    <author initials='S.' surname='Rose' fullname='S. Rose'/>
5018    <date year='2005' month='March' />
5019  </front>
5020  <seriesInfo name='RFC' value='4033' />
5021</reference>
5022
5023<reference anchor="BCP13">
5024  <front>
5025    <title>Media Type Specifications and Registration Procedures</title>
5026    <author initials="N." surname="Freed" fullname="Ned Freed">
5027      <organization>Oracle</organization>
5028      <address>
5029        <email>ned+ietf@mrochek.com</email>
5030      </address>
5031    </author>
5032    <author initials="J." surname="Klensin" fullname="John C. Klensin">
5033      <address>
5034        <email>john+ietf@jck.com</email>
5035      </address>
5036    </author>
5037    <author initials="T." surname="Hansen" fullname="Tony Hansen">
5038      <organization>AT&amp;T Laboratories</organization>
5039      <address>
5040        <email>tony+mtsuffix@maillennium.att.com</email>
5041      </address>
5042    </author>
5043    <date year="2013" month="January"/>
5044  </front>
5045  <seriesInfo name="BCP" value="13"/>
5046  <seriesInfo name="RFC" value="6838"/>
5047</reference>
5048
5049<reference anchor='BCP115'>
5050  <front>
5051    <title>Guidelines and Registration Procedures for New URI Schemes</title>
5052    <author initials='T.' surname='Hansen' fullname='T. Hansen'>
5053      <organization>AT&amp;T Laboratories</organization>
5054      <address>
5055        <email>tony+urireg@maillennium.att.com</email>
5056      </address>
5057    </author>
5058    <author initials='T.' surname='Hardie' fullname='T. Hardie'>
5059      <organization>Qualcomm, Inc.</organization>
5060      <address>
5061        <email>hardie@qualcomm.com</email>
5062      </address>
5063    </author>
5064    <author initials='L.' surname='Masinter' fullname='L. Masinter'>
5065      <organization>Adobe Systems</organization>
5066      <address>
5067        <email>LMM@acm.org</email>
5068      </address>
5069    </author>
5070    <date year='2006' month='February' />
5071  </front>
5072  <seriesInfo name='BCP' value='115' />
5073  <seriesInfo name='RFC' value='4395' />
5074</reference>
5075
5076<reference anchor='RFC4559'>
5077  <front>
5078    <title>SPNEGO-based Kerberos and NTLM HTTP Authentication in Microsoft Windows</title>
5079    <author initials='K.' surname='Jaganathan' fullname='K. Jaganathan'/>
5080    <author initials='L.' surname='Zhu' fullname='L. Zhu'/>
5081    <author initials='J.' surname='Brezak' fullname='J. Brezak'/>
5082    <date year='2006' month='June' />
5083  </front>
5084  <seriesInfo name='RFC' value='4559' />
5085</reference>
5086
5087<reference anchor='RFC5226'>
5088  <front>
5089    <title>Guidelines for Writing an IANA Considerations Section in RFCs</title>
5090    <author initials='T.' surname='Narten' fullname='T. Narten'>
5091      <organization>IBM</organization>
5092      <address><email>narten@us.ibm.com</email></address>
5093    </author>
5094    <author initials='H.' surname='Alvestrand' fullname='H. Alvestrand'>
5095      <organization>Google</organization>
5096      <address><email>Harald@Alvestrand.no</email></address>
5097    </author>
5098    <date year='2008' month='May' />
5099  </front>
5100  <seriesInfo name='BCP' value='26' />
5101  <seriesInfo name='RFC' value='5226' />
5102</reference>
5103
5104<reference anchor='RFC5246'>
5105   <front>
5106      <title>The Transport Layer Security (TLS) Protocol Version 1.2</title>
5107      <author initials='T.' surname='Dierks' fullname='T. Dierks'/>
5108      <author initials='E.' surname='Rescorla' fullname='E. Rescorla'>
5109         <organization>RTFM, Inc.</organization>
5110      </author>
5111      <date year='2008' month='August' />
5112   </front>
5113   <seriesInfo name='RFC' value='5246' />
5114</reference>
5115
5116<reference anchor="RFC5322">
5117  <front>
5118    <title>Internet Message Format</title>
5119    <author initials="P." surname="Resnick" fullname="P. Resnick">
5120      <organization>Qualcomm Incorporated</organization>
5121    </author>
5122    <date year="2008" month="October"/>
5123  </front>
5124  <seriesInfo name="RFC" value="5322"/>
5125</reference>
5126
5127<reference anchor="RFC6265">
5128  <front>
5129    <title>HTTP State Management Mechanism</title>
5130    <author initials="A." surname="Barth" fullname="Adam Barth">
5131      <organization abbrev="U.C. Berkeley">
5132        University of California, Berkeley
5133      </organization>
5134      <address><email>abarth@eecs.berkeley.edu</email></address>
5135    </author>
5136    <date year="2011" month="April" />
5137  </front>
5138  <seriesInfo name="RFC" value="6265"/>
5139</reference>
5140
5141<reference anchor='RFC6585'>
5142  <front>
5143    <title>Additional HTTP Status Codes</title>
5144    <author initials='M.' surname='Nottingham' fullname='M. Nottingham'>
5145      <organization>Rackspace</organization>
5146    </author>
5147    <author initials='R.' surname='Fielding' fullname='R. Fielding'>
5148      <organization>Adobe</organization>
5149    </author>
5150    <date year='2012' month='April' />
5151   </front>
5152   <seriesInfo name='RFC' value='6585' />
5153</reference>
5154
5155<!--<reference anchor='BCP97'>
5156  <front>
5157    <title>Handling Normative References to Standards-Track Documents</title>
5158    <author initials='J.' surname='Klensin' fullname='J. Klensin'>
5159      <address>
5160        <email>klensin+ietf@jck.com</email>
5161      </address>
5162    </author>
5163    <author initials='S.' surname='Hartman' fullname='S. Hartman'>
5164      <organization>MIT</organization>
5165      <address>
5166        <email>hartmans-ietf@mit.edu</email>
5167      </address>
5168    </author>
5169    <date year='2007' month='June' />
5170  </front>
5171  <seriesInfo name='BCP' value='97' />
5172  <seriesInfo name='RFC' value='4897' />
5173</reference>-->
5174
5175<reference anchor="Kri2001" target="http://arxiv.org/abs/cs.SE/0105018">
5176  <front>
5177    <title>HTTP Cookies: Standards, Privacy, and Politics</title>
5178    <author initials="D." surname="Kristol" fullname="David M. Kristol"/>
5179    <date year="2001" month="November"/>
5180  </front>
5181  <seriesInfo name="ACM Transactions on Internet Technology" value="1(2)"/>
5182</reference>
5183
5184<reference anchor="Klein" target="http://packetstormsecurity.com/papers/general/whitepaper_httpresponse.pdf">
5185  <front>
5186    <title>Divide and Conquer - HTTP Response Splitting, Web Cache Poisoning Attacks, and Related Topics</title>
5187    <author initials="A." surname="Klein" fullname="Amit Klein">
5188      <organization>Sanctum, Inc.</organization>
5189    </author>
5190    <date year="2004" month="March"/>
5191  </front>
5192</reference>
5193
5194<reference anchor="Georgiev" target="http://doi.acm.org/10.1145/2382196.2382204">
5195  <front>
5196    <title>The Most Dangerous Code in the World: Validating SSL Certificates in Non-browser Software</title>
5197    <author initials="M." surname="Georgiev" fullname="Martin Georgiev"/>
5198    <author initials="S." surname="Iyengar" fullname="Subodh Iyengar"/>
5199    <author initials="S." surname="Jana" fullname="Suman Jana"/>
5200    <author initials="R." surname="Anubhai" fullname="Rishita Anubhai"/>
5201    <author initials="D." surname="Boneh" fullname="Dan Boneh"/>
5202    <author initials="V." surname="Shmatikov" fullname="Vitaly Shmatikov"/>
5203    <date year="2012" month="October"/>
5204  </front>
5205  <x:prose>In Proceedings of the 2012 ACM Conference on Computer and Communications Security (CCS '12), pp. 38-49</x:prose>
5206</reference>
5207
5208<reference anchor="Linhart" target="http://www.watchfire.com/news/whitepapers.aspx">
5209  <front>
5210    <title>HTTP Request Smuggling</title>
5211    <author initials="C." surname="Linhart" fullname="Chaim Linhart"/>
5212    <author initials="A." surname="Klein" fullname="Amit Klein"/>
5213    <author initials="R." surname="Heled" fullname="Ronen Heled"/>
5214    <author initials="S." surname="Orrin" fullname="Steve Orrin"/>
5215    <date year="2005" month="June"/>
5216  </front>
5217</reference>
5218
5219</references>
5220
5221
5222<section title="HTTP Version History" anchor="compatibility">
5223<t>
5224   HTTP has been in use since 1990. The first version, later referred to as
5225   HTTP/0.9, was a simple protocol for hypertext data transfer across the
5226   Internet, using only a single request method (GET) and no metadata.
5227   HTTP/1.0, as defined by <xref target="RFC1945"/>, added a range of request
5228   methods and MIME-like messaging, allowing for metadata to be transferred
5229   and modifiers placed on the request/response semantics. However,
5230   HTTP/1.0 did not sufficiently take into consideration the effects of
5231   hierarchical proxies, caching, the need for persistent connections, or
5232   name-based virtual hosts. The proliferation of incompletely-implemented
5233   applications calling themselves "HTTP/1.0" further necessitated a
5234   protocol version change in order for two communicating applications
5235   to determine each other's true capabilities.
5236</t>
5237<t>
5238   HTTP/1.1 remains compatible with HTTP/1.0 by including more stringent
5239   requirements that enable reliable implementations, adding only
5240   those features that can either be safely ignored by an HTTP/1.0
5241   recipient or only sent when communicating with a party advertising
5242   conformance with HTTP/1.1.
5243</t>
5244<t>
5245   HTTP/1.1 has been designed to make supporting previous versions easy.
5246   A general-purpose HTTP/1.1 server ought to be able to understand any valid
5247   request in the format of HTTP/1.0, responding appropriately with an
5248   HTTP/1.1 message that only uses features understood (or safely ignored) by
5249   HTTP/1.0 clients. Likewise, an HTTP/1.1 client can be expected to
5250   understand any valid HTTP/1.0 response.
5251</t>
5252<t>
5253   Since HTTP/0.9 did not support header fields in a request, there is no
5254   mechanism for it to support name-based virtual hosts (selection of resource
5255   by inspection of the <x:ref>Host</x:ref> header field).
5256   Any server that implements name-based virtual hosts ought to disable
5257   support for HTTP/0.9. Most requests that appear to be HTTP/0.9 are, in
5258   fact, badly constructed HTTP/1.x requests caused by a client failing to
5259   properly encode the request-target.
5260</t>
5261
5262<section title="Changes from HTTP/1.0" anchor="changes.from.1.0">
5263<t>
5264   This section summarizes major differences between versions HTTP/1.0
5265   and HTTP/1.1.
5266</t>
5267
5268<section title="Multi-homed Web Servers" anchor="changes.to.simplify.multi-homed.web.servers.and.conserve.ip.addresses">
5269<t>
5270   The requirements that clients and servers support the <x:ref>Host</x:ref>
5271   header field (<xref target="header.host"/>), report an error if it is
5272   missing from an HTTP/1.1 request, and accept absolute URIs (<xref target="request-target"/>)
5273   are among the most important changes defined by HTTP/1.1.
5274</t>
5275<t>
5276   Older HTTP/1.0 clients assumed a one-to-one relationship of IP
5277   addresses and servers; there was no other established mechanism for
5278   distinguishing the intended server of a request than the IP address
5279   to which that request was directed. The <x:ref>Host</x:ref> header field was
5280   introduced during the development of HTTP/1.1 and, though it was
5281   quickly implemented by most HTTP/1.0 browsers, additional requirements
5282   were placed on all HTTP/1.1 requests in order to ensure complete
5283   adoption.  At the time of this writing, most HTTP-based services
5284   are dependent upon the Host header field for targeting requests.
5285</t>
5286</section>
5287
5288<section title="Keep-Alive Connections" anchor="compatibility.with.http.1.0.persistent.connections">
5289<t>
5290   In HTTP/1.0, each connection is established by the client prior to the
5291   request and closed by the server after sending the response. However, some
5292   implementations implement the explicitly negotiated ("Keep-Alive") version
5293   of persistent connections described in <xref x:sec="19.7.1" x:fmt="of"
5294   target="RFC2068"/>.
5295</t>
5296<t>
5297   Some clients and servers might wish to be compatible with these previous
5298   approaches to persistent connections, by explicitly negotiating for them
5299   with a "Connection: keep-alive" request header field. However, some
5300   experimental implementations of HTTP/1.0 persistent connections are faulty;
5301   for example, if an HTTP/1.0 proxy server doesn't understand
5302   <x:ref>Connection</x:ref>, it will erroneously forward that header field
5303   to the next inbound server, which would result in a hung connection.
5304</t>
5305<t>
5306   One attempted solution was the introduction of a Proxy-Connection header
5307   field, targeted specifically at proxies. In practice, this was also
5308   unworkable, because proxies are often deployed in multiple layers, bringing
5309   about the same problem discussed above.
5310</t>
5311<t>
5312   As a result, clients are encouraged not to send the Proxy-Connection header
5313   field in any requests.
5314</t>
5315<t>
5316   Clients are also encouraged to consider the use of Connection: keep-alive
5317   in requests carefully; while they can enable persistent connections with
5318   HTTP/1.0 servers, clients using them will need to monitor the
5319   connection for "hung" requests (which indicate that the client ought stop
5320   sending the header field), and this mechanism ought not be used by clients
5321   at all when a proxy is being used.
5322</t>
5323</section>
5324
5325<section title="Introduction of Transfer-Encoding" anchor="introduction.of.transfer-encoding">
5326<t>
5327   HTTP/1.1 introduces the <x:ref>Transfer-Encoding</x:ref> header field
5328   (<xref target="header.transfer-encoding"/>).
5329   Transfer codings need to be decoded prior to forwarding an HTTP message
5330   over a MIME-compliant protocol.
5331</t>
5332</section>
5333
5334</section>
5335
5336<section title="Changes from RFC 2616" anchor="changes.from.rfc.2616">
5337<t>
5338  HTTP's approach to error handling has been explained.
5339  (<xref target="conformance" />)
5340</t>
5341<t>
5342  The HTTP-version ABNF production has been clarified to be case-sensitive.
5343  Additionally, version numbers has been restricted to single digits, due
5344  to the fact that implementations are known to handle multi-digit version
5345  numbers incorrectly.
5346  (<xref target="http.version"/>)
5347</t>
5348<t>
5349  Userinfo (i.e., username and password) are now disallowed in HTTP and
5350  HTTPS URIs, because of security issues related to their transmission on the
5351  wire.
5352  (<xref target="http.uri" />)
5353</t>
5354<t>
5355  The HTTPS URI scheme is now defined by this specification; previously,
5356  it was done in  <xref target="RFC2818" x:fmt="of" x:sec="2.4"/>.
5357  Furthermore, it implies end-to-end security.
5358  (<xref target="https.uri"/>)
5359</t>
5360<t>
5361  HTTP messages can be (and often are) buffered by implementations; despite
5362  it sometimes being available as a stream, HTTP is fundamentally a
5363  message-oriented protocol.
5364  Minimum supported sizes for various protocol elements have been
5365  suggested, to improve interoperability.
5366  (<xref target="http.message" />)
5367</t>
5368<t>
5369  Invalid whitespace around field-names is now required to be rejected,
5370  because accepting it represents a security vulnerability.
5371  The ABNF productions defining header fields now only list the field value.
5372  (<xref target="header.fields"/>)
5373</t>
5374<t>
5375  Rules about implicit linear whitespace between certain grammar productions
5376  have been removed; now whitespace is only allowed where specifically
5377  defined in the ABNF.
5378  (<xref target="whitespace"/>)
5379</t>
5380<t>
5381  Header fields that span multiple lines ("line folding") are deprecated.
5382  (<xref target="field.parsing" />)
5383</t>
5384<t> 
5385  The NUL octet is no longer allowed in comment and quoted-string text, and
5386  handling of backslash-escaping in them has been clarified.
5387  The quoted-pair rule no longer allows escaping control characters other than
5388  HTAB.
5389  Non-ASCII content in header fields and the reason phrase has been obsoleted
5390  and made opaque (the TEXT rule was removed).
5391  (<xref target="field.components"/>)
5392</t> 
5393<t>
5394  Bogus "<x:ref>Content-Length</x:ref>" header fields are now required to be
5395  handled as errors by recipients.
5396  (<xref target="header.content-length"/>)
5397</t>
5398<t>
5399  The algorithm for determining the message body length has been clarified
5400  to indicate all of the special cases (e.g., driven by methods or status
5401  codes) that affect it, and that new protocol elements cannot define such
5402  special cases.
5403  CONNECT is a new, special case in determining message body length.
5404  "multipart/byteranges" is no longer a way of determining message body length
5405  detection.
5406  (<xref target="message.body.length"/>)
5407</t>
5408<t>
5409  The "identity" transfer coding token has been removed.
5410  (Sections <xref format="counter" target="message.body"/> and
5411  <xref format="counter" target="transfer.codings"/>)
5412</t>
5413<t>
5414  Chunk length does not include the count of the octets in the
5415  chunk header and trailer.
5416  Line folding in chunk extensions is  disallowed.
5417  (<xref target="chunked.encoding"/>)
5418</t>
5419<t>
5420  The meaning of the "deflate" content coding has been clarified.
5421  (<xref target="deflate.coding" />)
5422</t>
5423<t>
5424  The segment + query components of RFC 3986 have been used to define the
5425  request-target, instead of abs_path from RFC 1808.
5426  The asterisk-form of the request-target is only allowed with the OPTIONS
5427  method.
5428  (<xref target="request-target"/>)
5429</t>
5430<t>
5431  The term "Effective Request URI" has been introduced.
5432  (<xref target="effective.request.uri" />)
5433</t>
5434<t>
5435  Gateways do not need to generate <x:ref>Via</x:ref> header fields anymore.
5436  (<xref target="header.via"/>)
5437</t>
5438<t>
5439  Exactly when "close" connection options have to be sent has been clarified.
5440  Also, "hop-by-hop" header fields are required to appear in the Connection header
5441  field; just because they're defined as hop-by-hop in this specification
5442  doesn't exempt them.
5443  (<xref target="header.connection"/>)
5444</t>
5445<t>
5446  The limit of two connections per server has been removed.
5447  An idempotent sequence of requests is no longer required to be retried.
5448  The requirement to retry requests under certain circumstances when the
5449  server prematurely closes the connection has been removed.
5450  Also, some extraneous requirements about when servers are allowed to close
5451  connections prematurely have been removed.
5452  (<xref target="persistent.connections"/>)
5453</t>
5454<t>
5455  The semantics of the <x:ref>Upgrade</x:ref> header field is now defined in
5456  responses other than 101 (this was incorporated from <xref
5457  target="RFC2817"/>). Furthermore, the ordering in the field value is now
5458  significant.
5459  (<xref target="header.upgrade"/>)
5460</t>
5461<t>
5462  Empty list elements in list productions (e.g., a list header field containing
5463  ", ,") have been deprecated.
5464  (<xref target="abnf.extension"/>)
5465</t>
5466<t>
5467  Registration of Transfer Codings now requires IETF Review
5468  (<xref target="transfer.coding.registry"/>)
5469</t>
5470<t>
5471  This specification now defines the Upgrade Token Registry, previously
5472  defined in <xref target="RFC2817" x:fmt="of" x:sec="7.2"/>.
5473  (<xref target="upgrade.token.registry"/>)
5474</t>
5475<t>
5476  The expectation to support HTTP/0.9 requests has been removed.
5477  (<xref target="compatibility"/>)
5478</t>
5479<t>
5480  Issues with the Keep-Alive and Proxy-Connection header fields in requests
5481  are pointed out, with use of the latter being discouraged altogether.
5482  (<xref target="compatibility.with.http.1.0.persistent.connections" />)
5483</t>
5484</section>
5485</section>
5486
5487<?BEGININC p1-messaging.abnf-appendix ?>
5488<section xmlns:x="http://purl.org/net/xml2rfc/ext" title="Collected ABNF" anchor="collected.abnf">
5489<figure>
5490<artwork type="abnf" name="p1-messaging.parsed-abnf">
5491<x:ref>BWS</x:ref> = OWS
5492
5493<x:ref>Connection</x:ref> = *( "," OWS ) connection-option *( OWS "," [ OWS
5494 connection-option ] )
5495<x:ref>Content-Length</x:ref> = 1*DIGIT
5496
5497<x:ref>HTTP-message</x:ref> = start-line *( header-field CRLF ) CRLF [ message-body
5498 ]
5499<x:ref>HTTP-name</x:ref> = %x48.54.54.50 ; HTTP
5500<x:ref>HTTP-version</x:ref> = HTTP-name "/" DIGIT "." DIGIT
5501<x:ref>Host</x:ref> = uri-host [ ":" port ]
5502
5503<x:ref>OWS</x:ref> = *( SP / HTAB )
5504
5505<x:ref>RWS</x:ref> = 1*( SP / HTAB )
5506
5507<x:ref>TE</x:ref> = [ ( "," / t-codings ) *( OWS "," [ OWS t-codings ] ) ]
5508<x:ref>Trailer</x:ref> = *( "," OWS ) field-name *( OWS "," [ OWS field-name ] )
5509<x:ref>Transfer-Encoding</x:ref> = *( "," OWS ) transfer-coding *( OWS "," [ OWS
5510 transfer-coding ] )
5511
5512<x:ref>URI-reference</x:ref> = &lt;URI-reference, defined in [RFC3986], Section 4.1&gt;
5513<x:ref>Upgrade</x:ref> = *( "," OWS ) protocol *( OWS "," [ OWS protocol ] )
5514
5515<x:ref>Via</x:ref> = *( "," OWS ) ( received-protocol RWS received-by [ RWS comment
5516 ] ) *( OWS "," [ OWS ( received-protocol RWS received-by [ RWS
5517 comment ] ) ] )
5518
5519<x:ref>absolute-URI</x:ref> = &lt;absolute-URI, defined in [RFC3986], Section 4.3&gt;
5520<x:ref>absolute-form</x:ref> = absolute-URI
5521<x:ref>absolute-path</x:ref> = 1*( "/" segment )
5522<x:ref>asterisk-form</x:ref> = "*"
5523<x:ref>authority</x:ref> = &lt;authority, defined in [RFC3986], Section 3.2&gt;
5524<x:ref>authority-form</x:ref> = authority
5525
5526<x:ref>chunk</x:ref> = chunk-size [ chunk-ext ] CRLF chunk-data CRLF
5527<x:ref>chunk-data</x:ref> = 1*OCTET
5528<x:ref>chunk-ext</x:ref> = *( ";" chunk-ext-name [ "=" chunk-ext-val ] )
5529<x:ref>chunk-ext-name</x:ref> = token
5530<x:ref>chunk-ext-val</x:ref> = token / quoted-string
5531<x:ref>chunk-size</x:ref> = 1*HEXDIG
5532<x:ref>chunked-body</x:ref> = *chunk last-chunk trailer-part CRLF
5533<x:ref>comment</x:ref> = "(" *( ctext / quoted-pair / comment ) ")"
5534<x:ref>connection-option</x:ref> = token
5535<x:ref>ctext</x:ref> = HTAB / SP / %x21-27 ; '!'-'''
5536 / %x2A-5B ; '*'-'['
5537 / %x5D-7E ; ']'-'~'
5538 / obs-text
5539
5540<x:ref>field-content</x:ref> = field-vchar [ 1*( SP / HTAB ) field-vchar ]
5541<x:ref>field-name</x:ref> = token
5542<x:ref>field-value</x:ref> = *( field-content / obs-fold )
5543<x:ref>field-vchar</x:ref> = VCHAR / obs-text
5544<x:ref>fragment</x:ref> = &lt;fragment, defined in [RFC3986], Section 3.5&gt;
5545
5546<x:ref>header-field</x:ref> = field-name ":" OWS field-value OWS
5547<x:ref>http-URI</x:ref> = "http://" authority path-abempty [ "?" query ] [ "#"
5548 fragment ]
5549<x:ref>https-URI</x:ref> = "https://" authority path-abempty [ "?" query ] [ "#"
5550 fragment ]
5551
5552<x:ref>last-chunk</x:ref> = 1*"0" [ chunk-ext ] CRLF
5553
5554<x:ref>message-body</x:ref> = *OCTET
5555<x:ref>method</x:ref> = token
5556
5557<x:ref>obs-fold</x:ref> = CRLF 1*( SP / HTAB )
5558<x:ref>obs-text</x:ref> = %x80-FF
5559<x:ref>origin-form</x:ref> = absolute-path [ "?" query ]
5560
5561<x:ref>partial-URI</x:ref> = relative-part [ "?" query ]
5562<x:ref>path-abempty</x:ref> = &lt;path-abempty, defined in [RFC3986], Section 3.3&gt;
5563<x:ref>port</x:ref> = &lt;port, defined in [RFC3986], Section 3.2.3&gt;
5564<x:ref>protocol</x:ref> = protocol-name [ "/" protocol-version ]
5565<x:ref>protocol-name</x:ref> = token
5566<x:ref>protocol-version</x:ref> = token
5567<x:ref>pseudonym</x:ref> = token
5568
5569<x:ref>qdtext</x:ref> = HTAB / SP / "!" / %x23-5B ; '#'-'['
5570 / %x5D-7E ; ']'-'~'
5571 / obs-text
5572<x:ref>query</x:ref> = &lt;query, defined in [RFC3986], Section 3.4&gt;
5573<x:ref>quoted-pair</x:ref> = "\" ( HTAB / SP / VCHAR / obs-text )
5574<x:ref>quoted-string</x:ref> = DQUOTE *( qdtext / quoted-pair ) DQUOTE
5575
5576<x:ref>rank</x:ref> = ( "0" [ "." *3DIGIT ] ) / ( "1" [ "." *3"0" ] )
5577<x:ref>reason-phrase</x:ref> = *( HTAB / SP / VCHAR / obs-text )
5578<x:ref>received-by</x:ref> = ( uri-host [ ":" port ] ) / pseudonym
5579<x:ref>received-protocol</x:ref> = [ protocol-name "/" ] protocol-version
5580<x:ref>relative-part</x:ref> = &lt;relative-part, defined in [RFC3986], Section 4.2&gt;
5581<x:ref>request-line</x:ref> = method SP request-target SP HTTP-version CRLF
5582<x:ref>request-target</x:ref> = origin-form / absolute-form / authority-form /
5583 asterisk-form
5584
5585<x:ref>scheme</x:ref> = &lt;scheme, defined in [RFC3986], Section 3.1&gt;
5586<x:ref>segment</x:ref> = &lt;segment, defined in [RFC3986], Section 3.3&gt;
5587<x:ref>start-line</x:ref> = request-line / status-line
5588<x:ref>status-code</x:ref> = 3DIGIT
5589<x:ref>status-line</x:ref> = HTTP-version SP status-code SP reason-phrase CRLF
5590
5591<x:ref>t-codings</x:ref> = "trailers" / ( transfer-coding [ t-ranking ] )
5592<x:ref>t-ranking</x:ref> = OWS ";" OWS "q=" rank
5593<x:ref>tchar</x:ref> = "!" / "#" / "$" / "%" / "&amp;" / "'" / "*" / "+" / "-" / "." /
5594 "^" / "_" / "`" / "|" / "~" / DIGIT / ALPHA
5595<x:ref>token</x:ref> = 1*tchar
5596<x:ref>trailer-part</x:ref> = *( header-field CRLF )
5597<x:ref>transfer-coding</x:ref> = "chunked" / "compress" / "deflate" / "gzip" /
5598 transfer-extension
5599<x:ref>transfer-extension</x:ref> = token *( OWS ";" OWS transfer-parameter )
5600<x:ref>transfer-parameter</x:ref> = token BWS "=" BWS ( token / quoted-string )
5601
5602<x:ref>uri-host</x:ref> = &lt;host, defined in [RFC3986], Section 3.2.2&gt;
5603</artwork>
5604</figure>
5605</section>
5606<?ENDINC p1-messaging.abnf-appendix ?>
5607
5608<section title="Change Log (to be removed by RFC Editor before publication)" anchor="change.log">
5609
5610<section title="Since RFC 2616">
5611<t>
5612  Changes up to the IETF Last Call draft are summarized
5613  in <eref target="http://trac.tools.ietf.org/html/draft-ietf-httpbis-p1-messaging-24#appendix-C"/>.
5614</t>
5615</section>
5616
5617<section title="Since draft-ietf-httpbis-p1-messaging-24" anchor="changes.since.24">
5618<t>
5619  Closed issues:
5620  <list style="symbols">
5621    <t>
5622      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/502"/>:
5623      "APPSDIR review of draft-ietf-httpbis-p1-messaging-24"
5624    </t>
5625    <t>
5626      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/507"/>:
5627      "integer value parsing"
5628    </t>
5629    <t>
5630      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/517"/>:
5631      "move IANA registrations to correct draft"
5632    </t>
5633  </list>
5634</t>
5635</section>
5636
5637<section title="Since draft-ietf-httpbis-p1-messaging-25" anchor="changes.since.25">
5638<t>
5639  Closed issues:
5640  <list style="symbols">
5641    <t>
5642      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/526"/>:
5643      "check media type registration templates"
5644    </t>
5645    <t>
5646      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/528"/>:
5647      "Redundant rule quoted-str-nf"
5648    </t>
5649    <t>
5650      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/538"/>:
5651      "add 'stateless' to Abstract"
5652    </t>
5653    <t>
5654      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/540"/>:
5655      "clarify ABNF layering"
5656    </t>
5657    <t>
5658      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/541"/>:
5659      "use of 'word' ABNF production"
5660    </t>
5661    <t>
5662      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/542"/>:
5663      "improve introduction of list rule"
5664    </t>
5665    <t>
5666      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/544"/>:
5667      "moving 2616/2068/2145 to historic"
5668    </t>
5669    <t>
5670      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/549"/>:
5671      "augment security considerations with pointers to current research"
5672    </t>
5673    <t>
5674      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/551"/>:
5675      "intermediaries handling trailers"
5676    </t>
5677    <t>
5678      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/552"/>:
5679      "allow privacy proxies to be conformant"
5680    </t>
5681  </list>
5682</t>
5683<t>
5684  Partly resolved issues:
5685  <list style="symbols">
5686    <t>
5687      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/531"/>:
5688      "IESG ballot on draft-ietf-httpbis-p1-messaging-25"
5689    </t>
5690  </list>
5691</t>
5692</section>
5693</section>
5694
5695</back>
5696</rfc>
Note: See TracBrowser for help on using the repository browser.