source: draft-ietf-httpbis/00/draft-ietf-httpbis-p1-messaging-00.xml @ 67

Last change on this file since 67 was 67, checked in by julian.reschke@…, 12 years ago

Record the day of submission (2007-12-20) so that future runs of xml2rfc can produce the correct front matter.

  • Property svn:eol-style set to native
File size: 169.7 KB
Line 
1<?xml version="1.0" encoding="UTF-8"?>
2<!--
3    This XML document is the output of clean-for-DTD.xslt; a tool that strips
4    extensions to RFC2629(bis) from documents for processing with xml2rfc.
5-->
6<?rfc toc="yes" ?>
7<?rfc symrefs="yes" ?>
8<?rfc sortrefs="yes" ?>
9<?rfc compact="yes"?>
10<?rfc subcompact="no" ?>
11<?rfc linkmailto="no" ?>
12<?rfc editing="no" ?>
13<!DOCTYPE rfc
14  PUBLIC "" "rfc2629.dtd">
15<rfc obsoletes="2068, 2616" category="std" ipr="full3978" docName="draft-ietf-httpbis-p1-messaging-00">
16<front>
17
18  <title abbrev="HTTP/1.1">HTTP/1.1, part 1: URIs, Connections, and Message Parsing</title>
19
20  <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
21    <organization abbrev="Day Software">Day Software</organization>
22    <address>
23      <postal>
24        <street>23 Corporate Plaza DR, Suite 280</street>
25        <city>Newport Beach</city>
26        <region>CA</region>
27        <code>92660</code>
28        <country>USA</country>
29      </postal>
30      <phone>+1-949-706-5300</phone>
31      <facsimile>+1-949-706-5305</facsimile>
32      <email>fielding@gbiv.com</email>
33      <uri>http://roy.gbiv.com/</uri>
34    </address>
35  </author>
36
37  <author initials="J." surname="Gettys" fullname="Jim Gettys">
38    <organization>One Laptop per Child</organization>
39    <address>
40      <postal>
41        <street>21 Oak Knoll Road</street>
42        <city>Carlisle</city>
43        <region>MA</region>
44        <code>01741</code>
45        <country>USA</country>
46      </postal>
47      <email>jg@laptop.org</email>
48      <uri>http://www.laptop.org/</uri>
49    </address>
50  </author>
51 
52  <author initials="J." surname="Mogul" fullname="Jeffrey C. Mogul">
53    <organization abbrev="HP">Hewlett-Packard Company</organization>
54    <address>
55      <postal>
56        <street>HP Labs, Large Scale Systems Group</street>
57        <street>1501 Page Mill Road, MS 1177</street>
58        <city>Palo Alto</city>
59        <region>CA</region>
60        <code>94304</code>
61        <country>USA</country>
62      </postal>
63      <email>JeffMogul@acm.org</email>
64    </address>
65  </author>
66
67  <author initials="H." surname="Frystyk" fullname="Henrik Frystyk Nielsen">
68    <organization abbrev="Microsoft">Microsoft Corporation</organization>
69    <address>
70      <postal>
71        <street>1 Microsoft Way</street>
72        <city>Redmond</city>
73        <region>WA</region>
74        <code>98052</code>
75        <country>USA</country>
76      </postal>
77      <email>henrikn@microsoft.com</email>
78    </address>
79  </author>
80
81  <author initials="L." surname="Masinter" fullname="Larry Masinter">
82    <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
83    <address>
84      <postal>
85        <street>345 Park Ave</street>
86        <city>San Jose</city>
87        <region>CA</region>
88        <code>95110</code>
89        <country>USA</country>
90      </postal>
91      <email>LMM@acm.org</email>
92      <uri>http://larry.masinter.net/</uri>
93    </address>
94  </author>
95 
96  <author initials="P." surname="Leach" fullname="Paul J. Leach">
97    <organization abbrev="Microsoft">Microsoft Corporation</organization>
98    <address>
99      <postal>
100        <street>1 Microsoft Way</street>
101        <city>Redmond</city>
102        <region>WA</region>
103        <code>98052</code>
104      </postal>
105      <email>paulle@microsoft.com</email>
106    </address>
107  </author>
108   
109  <author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
110    <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
111    <address>
112      <postal>
113        <street>MIT Computer Science and Artificial Intelligence Laboratory</street>
114        <street>The Stata Center, Building 32</street>
115        <street>32 Vassar Street</street>
116        <city>Cambridge</city>
117        <region>MA</region>
118        <code>02139</code>
119        <country>USA</country>
120      </postal>
121      <email>timbl@w3.org</email>
122      <uri>http://www.w3.org/People/Berners-Lee/</uri>
123    </address>
124  </author>
125
126  <date month="December" year="2007" day="20"/>
127
128<abstract>
129<t>
130   The Hypertext Transfer Protocol (HTTP) is an application-level
131   protocol for distributed, collaborative, hypermedia information
132   systems. HTTP has been in use by the World Wide Web global information
133   initiative since 1990. This document is Part 1 of the seven-part specification
134   that defines the protocol referred to as "HTTP/1.1" and, taken together,
135   obsoletes RFC 2616.  Part 1 provides an overview of HTTP and
136   its associated terminology, defines the "http" and "https" Uniform
137   Resource Identifier (URI) schemes, defines the generic message syntax
138   and parsing requirements for HTTP message frames, and describes
139   general security concerns for implementations.
140</t>
141</abstract>
142
143<note title="Editorial Note (To be removed by RFC Editor)">
144  <t>
145    This version of the HTTP specification contains only minimal editorial
146    changes from <xref target="RFC2616"/> (abstract, introductory paragraph,
147    and authors' addresses).  All other changes are due to partitioning the
148    original into seven mostly independent parts.  The intent is for readers
149    of future drafts to able to use draft 00 as the basis for comparison
150    when the WG makes later changes to the specification text.  This draft
151    will shortly be followed by draft 01 (containing the first round of changes
152    that have already been agreed to on the mailing list). There is no point in
153    reviewing this draft other than to verify that the partitioning has been
154    done correctly.  Roy T. Fielding, Yves Lafon, and Julian Reschke
155    will be the editors after draft 00 is submitted.
156  </t>
157  <t>
158    Discussion of this draft should take place on the HTTPBIS working group
159    mailing list (ietf-http-wg@w3.org). The current issues list is
160    at <eref target="http://www3.tools.ietf.org/wg/httpbis/trac/report/11"/>
161    and related documents (including fancy diffs) can be found at
162    <eref target="http://www3.tools.ietf.org/wg/httpbis/"/>.
163  </t>
164</note>
165</front>
166<middle>
167<section title="Introduction" anchor="introduction">
168<t>
169   This document will define aspects of HTTP related to overall network
170   operation, message framing, interaction with transport protocols, and
171   URI schemes. Right now it only includes the extracted relevant sections
172   of <xref target="RFC2616"/>.
173</t>
174<section title="Purpose" anchor="intro.purpose">
175<t>
176   The Hypertext Transfer Protocol (HTTP) is an application-level
177   protocol for distributed, collaborative, hypermedia information
178   systems. HTTP has been in use by the World-Wide Web global
179   information initiative since 1990. The first version of HTTP,
180   referred to as HTTP/0.9, was a simple protocol for raw data transfer
181   across the Internet. HTTP/1.0, as defined by RFC 1945 <xref target="RFC1945"/>, improved
182   the protocol by allowing messages to be in the format of MIME-like
183   messages, containing metainformation about the data transferred and
184   modifiers on the request/response semantics. However, HTTP/1.0 does
185   not sufficiently take into consideration the effects of hierarchical
186   proxies, caching, the need for persistent connections, or virtual
187   hosts. In addition, the proliferation of incompletely-implemented
188   applications calling themselves "HTTP/1.0" has necessitated a
189   protocol version change in order for two communicating applications
190   to determine each other's true capabilities.
191</t>
192<t>
193   This specification defines the protocol referred to as "HTTP/1.1".
194   This protocol includes more stringent requirements than HTTP/1.0 in
195   order to ensure reliable implementation of its features.
196</t>
197<t>
198   Practical information systems require more functionality than simple
199   retrieval, including search, front-end update, and annotation. HTTP
200   allows an open-ended set of methods and headers that indicate the
201   purpose of a request <xref target="RFC2324"/>. It builds on the discipline of reference
202   provided by the Uniform Resource Identifier (URI) <xref target="RFC1630"/>, as a location
203   (URL) <xref target="RFC1738"/> or name (URN) <xref target="RFC1737"/>, for indicating the resource to which a
204   method is to be applied. Messages are passed in a format similar to
205   that used by Internet mail <xref target="RFC822"/> as defined by the Multipurpose
206   Internet Mail Extensions (MIME) <xref target="RFC2045"/>.
207</t>
208<t>
209   HTTP is also used as a generic protocol for communication between
210   user agents and proxies/gateways to other Internet systems, including
211   those supported by the SMTP <xref target="RFC821"/>, NNTP <xref target="RFC977"/>, FTP <xref target="RFC959"/>, Gopher <xref target="RFC1436"/>,
212   and WAIS <xref target="WAIS"/> protocols. In this way, HTTP allows basic hypermedia
213   access to resources available from diverse applications.
214</t>
215</section>
216
217<section title="Requirements" 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 RFC 2119 <xref target="RFC2119"/>.
222</t>
223<t>
224   An implementation is not compliant if it fails to satisfy one or more
225   of the MUST or REQUIRED level requirements for the protocols it
226   implements. An implementation that satisfies all the MUST or REQUIRED
227   level and all the SHOULD level requirements for its protocols is said
228   to be "unconditionally compliant"; one that satisfies all the MUST
229   level requirements but not all the SHOULD level requirements for its
230   protocols is said to be "conditionally compliant."
231</t>
232</section>
233
234<section title="Terminology" anchor="intro.terminology">
235<t>
236   This specification uses a number of terms to refer to the roles
237   played by participants in, and objects of, the HTTP communication.
238</t>
239<t>
240  <iref item="connection"/>
241  connection
242  <list>
243    <t>
244      A transport layer virtual circuit established between two programs
245      for the purpose of communication.
246    </t>
247  </list>
248</t>
249<t>
250  <iref item="message"/>
251  message
252  <list>
253    <t>
254      The basic unit of HTTP communication, consisting of a structured
255      sequence of octets matching the syntax defined in <xref target="http.message"/> and
256      transmitted via the connection.
257    </t>
258  </list>
259</t>
260<t>
261  <iref item="request"/>
262  request
263  <list>
264    <t>
265      An HTTP request message, as defined in <xref target="request"/>.
266    </t>
267  </list>
268</t>
269<t>
270  <iref item="response"/>
271  response
272  <list>
273    <t>
274      An HTTP response message, as defined in <xref target="response"/>.
275    </t>
276  </list>
277</t>
278<t>
279  <iref item="resource"/>
280  resource
281  <list>
282    <t>
283      A network data object or service that can be identified by a URI,
284      as defined in <xref target="uri"/>. Resources may be available in multiple
285      representations (e.g. multiple languages, data formats, size, and
286      resolutions) or vary in other ways.
287    </t>
288  </list>
289</t>
290<t>
291  <iref item="entity"/>
292  entity
293  <list>
294    <t>
295      The information transferred as the payload of a request or
296      response. An entity consists of metainformation in the form of
297      entity-header fields and content in the form of an entity-body, as
298      described in Section 3 of <xref target="Part3"/>.
299    </t>
300  </list>
301</t>
302<t>
303  <iref item="representation"/>
304  representation
305  <list>
306    <t>
307      An entity included with a response that is subject to content
308      negotiation, as described in Section 4 of <xref target="Part3"/>. There may exist multiple
309      representations associated with a particular response status.
310    </t>
311  </list>
312</t>
313<t>
314  <iref item="content negotiation"/>
315  content negotiation
316  <list>
317    <t>
318      The mechanism for selecting the appropriate representation when
319      servicing a request, as described in Section 4 of <xref target="Part3"/>. The
320      representation of entities in any response can be negotiated
321      (including error responses).
322    </t>
323  </list>
324</t>
325<t>
326  <iref item="variant"/>
327  variant
328  <list>
329    <t>
330      A resource may have one, or more than one, representation(s)
331      associated with it at any given instant. Each of these
332      representations is termed a `varriant'.  Use of the term `variant'
333      does not necessarily imply that the resource is subject to content
334      negotiation.
335    </t>
336  </list>
337</t>
338<t>
339  <iref item="client"/>
340  client
341  <list>
342    <t>
343      A program that establishes connections for the purpose of sending
344      requests.
345    </t>
346  </list>
347</t>
348<t>
349  <iref item="user agent"/>
350  user agent
351  <list>
352    <t>
353      The client which initiates a request. These are often browsers,
354      editors, spiders (web-traversing robots), or other end user tools.
355    </t>
356  </list>
357</t>
358<t>
359  <iref item="server"/>
360  server
361  <list>
362    <t>
363      An application program that accepts connections in order to
364      service requests by sending back responses. Any given program may
365      be capable of being both a client and a server; our use of these
366      terms refers only to the role being performed by the program for a
367      particular connection, rather than to the program's capabilities
368      in general. Likewise, any server may act as an origin server,
369      proxy, gateway, or tunnel, switching behavior based on the nature
370      of each request.
371    </t>
372  </list>
373</t>
374<t>
375  <iref item="origin server"/>
376  origin server
377  <list>
378    <t>
379      The server on which a given resource resides or is to be created.
380    </t>
381  </list>
382</t>
383<t>
384  <iref item="proxy"/>
385  proxy
386  <list>
387    <t>
388      An intermediary program which acts as both a server and a client
389      for the purpose of making requests on behalf of other clients.
390      Requests are serviced internally or by passing them on, with
391      possible translation, to other servers. A proxy MUST implement
392      both the client and server requirements of this specification. A
393      "transparent proxy" is a proxy that does not modify the request or
394      response beyond what is required for proxy authentication and
395      identification. A "non-transparent proxy" is a proxy that modifies
396      the request or response in order to provide some added service to
397      the user agent, such as group annotation services, media type
398      transformation, protocol reduction, or anonymity filtering. Except
399      where either transparent or non-transparent behavior is explicitly
400      stated, the HTTP proxy requirements apply to both types of
401      proxies.
402    </t>
403  </list>
404</t>
405<t>
406  <iref item="gateway"/>
407  gateway
408  <list>
409    <t>
410      A server which acts as an intermediary for some other server.
411      Unlike a proxy, a gateway receives requests as if it were the
412      origin server for the requested resource; the requesting client
413      may not be aware that it is communicating with a gateway.
414    </t>
415  </list>
416</t>
417<t>
418  <iref item="tunnel"/>
419  tunnel
420  <list>
421    <t>
422      An intermediary program which is acting as a blind relay between
423      two connections. Once active, a tunnel is not considered a party
424      to the HTTP communication, though the tunnel may have been
425      initiated by an HTTP request. The tunnel ceases to exist when both
426      ends of the relayed connections are closed.
427    </t>
428  </list>
429</t>
430<t>
431  <iref item="cache"/>
432  cache
433  <list>
434    <t>
435      A program's local store of response messages and the subsystem
436      that controls its message storage, retrieval, and deletion. A
437      cache stores cacheable responses in order to reduce the response
438      time and network bandwidth consumption on future, equivalent
439      requests. Any client or server may include a cache, though a cache
440      cannot be used by a server that is acting as a tunnel.
441    </t>
442  </list>
443</t>
444<t>
445  <iref item="cacheable"/>
446  cacheable
447  <list>
448    <t>
449      A response is cacheable if a cache is allowed to store a copy of
450      the response message for use in answering subsequent requests. The
451      rules for determining the cacheability of HTTP responses are
452      defined in Section 2 of <xref target="Part6"/>. Even if a resource is cacheable, there may
453      be additional constraints on whether a cache can use the cached
454      copy for a particular request.
455    </t>
456  </list>
457</t>
458<t>
459  <iref item="upstream"/>
460  <iref item="downstream"/>
461  upstream/downstream
462  <list>
463    <t>
464      Upstream and downstream describe the flow of a message: all
465      messages flow from upstream to downstream.
466    </t>
467  </list>
468</t>
469<t>
470  <iref item="inbound"/>
471  <iref item="outbound"/>
472  inbound/outbound
473  <list>
474    <t>
475      Inbound and outbound refer to the request and response paths for
476      messages: "inbound" means "traveling toward the origin server",
477      and "outbound" means "traveling toward the user agent"
478    </t>
479  </list>
480</t>
481</section>
482
483<section title="Overall Operation" anchor="intro.overall.operation">
484<t>
485   The HTTP protocol is a request/response protocol. A client sends a
486   request to the server in the form of a request method, URI, and
487   protocol version, followed by a MIME-like message containing request
488   modifiers, client information, and possible body content over a
489   connection with a server. The server responds with a status line,
490   including the message's protocol version and a success or error code,
491   followed by a MIME-like message containing server information, entity
492   metainformation, and possible entity-body content. The relationship
493   between HTTP and MIME is described in Appendix A of <xref target="Part3"/>.
494</t>
495<t>
496   Most HTTP communication is initiated by a user agent and consists of
497   a request to be applied to a resource on some origin server. In the
498   simplest case, this may be accomplished via a single connection (v)
499   between the user agent (UA) and the origin server (O).
500</t>
501<figure><artwork type="drawing"><![CDATA[
502       request chain ------------------------>
503    UA -------------------v------------------- O
504       <----------------------- response chain
505]]></artwork></figure>
506<t>
507   A more complicated situation occurs when one or more intermediaries
508   are present in the request/response chain. There are three common
509   forms of intermediary: proxy, gateway, and tunnel. A proxy is a
510   forwarding agent, receiving requests for a URI in its absolute form,
511   rewriting all or part of the message, and forwarding the reformatted
512   request toward the server identified by the URI. A gateway is a
513   receiving agent, acting as a layer above some other server(s) and, if
514   necessary, translating the requests to the underlying server's
515   protocol. A tunnel acts as a relay point between two connections
516   without changing the messages; tunnels are used when the
517   communication needs to pass through an intermediary (such as a
518   firewall) even when the intermediary cannot understand the contents
519   of the messages.
520</t>
521<figure><artwork type="drawing"><![CDATA[
522       request chain -------------------------------------->
523    UA -----v----- A -----v----- B -----v----- C -----v----- O
524       <------------------------------------- response chain
525]]></artwork></figure>
526<t>
527   The figure above shows three intermediaries (A, B, and C) between the
528   user agent and origin server. A request or response message that
529   travels the whole chain will pass through four separate connections.
530   This distinction is important because some HTTP communication options
531   may apply only to the connection with the nearest, non-tunnel
532   neighbor, only to the end-points of the chain, or to all connections
533   along the chain. Although the diagram is linear, each participant may
534   be engaged in multiple, simultaneous communications. For example, B
535   may be receiving requests from many clients other than A, and/or
536   forwarding requests to servers other than C, at the same time that it
537   is handling A's request.
538</t>
539<t>
540   Any party to the communication which is not acting as a tunnel may
541   employ an internal cache for handling requests. The effect of a cache
542   is that the request/response chain is shortened if one of the
543   participants along the chain has a cached response applicable to that
544   request. The following illustrates the resulting chain if B has a
545   cached copy of an earlier response from O (via C) for a request which
546   has not been cached by UA or A.
547</t>
548<figure><artwork type="drawing"><![CDATA[
549          request chain ---------->
550       UA -----v----- A -----v----- B - - - - - - C - - - - - - O
551          <--------- response chain
552]]></artwork></figure>
553<t>
554   Not all responses are usefully cacheable, and some requests may
555   contain modifiers which place special requirements on cache behavior.
556   HTTP requirements for cache behavior and cacheable responses are
557   defined in Section 2 of <xref target="Part6"/>.
558</t>
559<t>
560   In fact, there are a wide variety of architectures and configurations
561   of caches and proxies currently being experimented with or deployed
562   across the World Wide Web. These systems include national hierarchies
563   of proxy caches to save transoceanic bandwidth, systems that
564   broadcast or multicast cache entries, organizations that distribute
565   subsets of cached data via CD-ROM, and so on. HTTP systems are used
566   in corporate intranets over high-bandwidth links, and for access via
567   PDAs with low-power radio links and intermittent connectivity. The
568   goal of HTTP/1.1 is to support the wide diversity of configurations
569   already deployed while introducing protocol constructs that meet the
570   needs of those who build web applications that require high
571   reliability and, failing that, at least reliable indications of
572   failure.
573</t>
574<t>
575   HTTP communication usually takes place over TCP/IP connections. The
576   default port is TCP 80 <xref target="RFC1700"/>, but other ports can be used. This does
577   not preclude HTTP from being implemented on top of any other protocol
578   on the Internet, or on other networks. HTTP only presumes a reliable
579   transport; any protocol that provides such guarantees can be used;
580   the mapping of the HTTP/1.1 request and response structures onto the
581   transport data units of the protocol in question is outside the scope
582   of this specification.
583</t>
584<t>
585   In HTTP/1.0, most implementations used a new connection for each
586   request/response exchange. In HTTP/1.1, a connection may be used for
587   one or more request/response exchanges, although connections may be
588   closed for a variety of reasons (see <xref target="persistent.connections"/>).
589</t>
590</section>
591</section>
592
593<section title="Notational Conventions and Generic Grammar" anchor="notation">
594
595<section title="Augmented BNF" anchor="notation.abnf">
596<t>
597   All of the mechanisms specified in this document are described in
598   both prose and an augmented Backus-Naur Form (BNF) similar to that
599   used by RFC 822 <xref target="RFC822"/>. Implementors will need to be familiar with the
600   notation in order to understand this specification. The augmented BNF
601   includes the following constructs:
602</t>
603<t>
604   name = definition
605  <list>
606    <t>
607      The name of a rule is simply the name itself (without any
608      enclosing "&lt;" and "&gt;") and is separated from its definition by the
609      equal "=" character. White space is only significant in that
610      indentation of continuation lines is used to indicate a rule
611      definition that spans more than one line. Certain basic rules are
612      in uppercase, such as SP, LWS, HT, CRLF, DIGIT, ALPHA, etc. Angle
613      brackets are used within definitions whenever their presence will
614      facilitate discerning the use of rule names.
615    </t>
616  </list>
617</t>
618<t>
619   "literal"
620  <list>
621    <t>
622      Quotation marks surround literal text. Unless stated otherwise,
623      the text is case-insensitive.
624    </t>
625  </list>
626</t>
627<t>
628   rule1 | rule2
629  <list>
630    <t>
631      Elements separated by a bar ("|") are alternatives, e.g., "yes |
632      no" will accept yes or no.
633    </t>
634  </list>
635</t>
636<t>
637   (rule1 rule2)
638  <list>
639    <t>
640      Elements enclosed in parentheses are treated as a single element.
641      Thus, "(elem (foo | bar) elem)" allows the token sequences "elem
642      foo elem" and "elem bar elem".
643    </t>
644  </list>
645</t>
646<t>
647   *rule
648  <list>
649    <t>
650      The character "*" preceding an element indicates repetition. The
651      full form is "&lt;n&gt;*&lt;m&gt;element" indicating at least &lt;n&gt; and at most
652      &lt;m&gt; occurrences of element. Default values are 0 and infinity so
653      that "*(element)" allows any number, including zero; "1*element"
654      requires at least one; and "1*2element" allows one or two.
655    </t>
656  </list>
657</t>
658<t>
659   [rule]
660  <list>
661    <t>
662      Square brackets enclose optional elements; "[foo bar]" is
663      equivalent to "*1(foo bar)".
664    </t>
665  </list>
666</t>
667<t>
668   N rule
669  <list>
670    <t>
671      Specific repetition: "&lt;n&gt;(element)" is equivalent to
672      "&lt;n&gt;*&lt;n&gt;(element)"; that is, exactly &lt;n&gt; occurrences of (element).
673      Thus 2DIGIT is a 2-digit number, and 3ALPHA is a string of three
674      alphabetic characters.
675    </t>
676  </list>
677</t>
678<t>
679   #rule
680  <list>
681    <t>
682      A construct "#" is defined, similar to "*", for defining lists of
683      elements. The full form is "&lt;n&gt;#&lt;m&gt;element" indicating at least
684      &lt;n&gt; and at most &lt;m&gt; elements, each separated by one or more commas
685      (",") and OPTIONAL linear white space (LWS). This makes the usual
686      form of lists very easy; a rule such as
687    </t>
688    <t>
689         ( *LWS element *( *LWS "," *LWS element ))
690    </t>
691    <t>
692      can be shown as
693    </t>
694    <t>
695         1#element
696    </t>
697    <t>
698      Wherever this construct is used, null elements are allowed, but do
699      not contribute to the count of elements present. That is,
700      "(element), , (element) " is permitted, but counts as only two
701      elements. Therefore, where at least one element is required, at
702      least one non-null element MUST be present. Default values are 0
703      and infinity so that "#element" allows any number, including zero;
704      "1#element" requires at least one; and "1#2element" allows one or
705      two.
706    </t>
707  </list>
708</t>
709<t>
710   ; comment
711  <list>
712    <t>
713      A semi-colon, set off some distance to the right of rule text,
714      starts a comment that continues to the end of line. This is a
715      simple way of including useful notes in parallel with the
716      specifications.
717    </t>
718  </list>
719</t>
720<t>
721   implied *LWS
722  <list>
723    <t>
724      The grammar described by this specification is word-based. Except
725      where noted otherwise, linear white space (LWS) can be included
726      between any two adjacent words (token or quoted-string), and
727      between adjacent words and separators, without changing the
728      interpretation of a field. At least one delimiter (LWS and/or
729      separators) MUST exist between any two tokens (for the definition
730      of "token" below), since they would otherwise be interpreted as a
731      single token.
732    </t>
733  </list>
734</t>
735</section>
736
737<section title="Basic Rules" anchor="basic.rules">
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760<t>
761   The following rules are used throughout this specification to
762   describe basic parsing constructs. The US-ASCII coded character set
763   is defined by ANSI X3.4-1986 <xref target="USASCII"/>.
764</t>
765<figure><iref primary="true" item="Grammar" subitem="OCTET"/><iref primary="true" item="Grammar" subitem="CHAR"/><iref primary="true" item="Grammar" subitem="UPALPHA"/><iref primary="true" item="Grammar" subitem="LOALPHA"/><iref primary="true" item="Grammar" subitem="ALPHA"/><iref primary="true" item="Grammar" subitem="DIGIT"/><iref primary="true" item="Grammar" subitem="CTL"/><iref primary="true" item="Grammar" subitem="CR"/><iref primary="true" item="Grammar" subitem="LF"/><iref primary="true" item="Grammar" subitem="SP"/><iref primary="true" item="Grammar" subitem="HT"/><artwork type="abnf2616"><![CDATA[
766    OCTET          = <any 8-bit sequence of data>
767    CHAR           = <any US-ASCII character (octets 0 - 127)>
768    UPALPHA        = <any US-ASCII uppercase letter "A".."Z">
769    LOALPHA        = <any US-ASCII lowercase letter "a".."z">
770    ALPHA          = UPALPHA | LOALPHA
771    DIGIT          = <any US-ASCII digit "0".."9">
772    CTL            = <any US-ASCII control character
773                     (octets 0 - 31) and DEL (127)>
774    CR             = <US-ASCII CR, carriage return (13)>
775    LF             = <US-ASCII LF, linefeed (10)>
776    SP             = <US-ASCII SP, space (32)>
777    HT             = <US-ASCII HT, horizontal-tab (9)>
778    <">            = <US-ASCII double-quote mark (34)>
779]]></artwork></figure>
780<t>
781   HTTP/1.1 defines the sequence CR LF as the end-of-line marker for all
782   protocol elements except the entity-body (see <xref target="tolerant.applications"/> for
783   tolerant applications). The end-of-line marker within an entity-body
784   is defined by its associated media type, as described in <xref target="Part3"/>.
785</t>
786<figure><iref primary="true" item="Grammar" subitem="CRLF"/><artwork type="abnf2616"><![CDATA[
787    CRLF           = CR LF
788]]></artwork></figure>
789<t>
790   HTTP/1.1 header field values can be folded onto multiple lines if the
791   continuation line begins with a space or horizontal tab. All linear
792   white space, including folding, has the same semantics as SP. A
793   recipient MAY replace any linear white space with a single SP before
794   interpreting the field value or forwarding the message downstream.
795</t>
796<figure><iref primary="true" item="Grammar" subitem="LWS"/><artwork type="abnf2616"><![CDATA[
797    LWS            = [CRLF] 1*( SP | HT )
798]]></artwork></figure>
799<t>
800   The TEXT rule is only used for descriptive field contents and values
801   that are not intended to be interpreted by the message parser. Words
802   of *TEXT MAY contain characters from character sets other than ISO-8859-1
803   <xref target="ISO-8859"/> only when encoded according to the rules of RFC 2047
804   <xref target="RFC2047"/>.
805</t>
806<figure><iref primary="true" item="Grammar" subitem="TEXT"/><artwork type="abnf2616"><![CDATA[
807    TEXT           = <any OCTET except CTLs,
808                     but including LWS>
809]]></artwork></figure>
810<t>
811   A CRLF is allowed in the definition of TEXT only as part of a header
812   field continuation. It is expected that the folding LWS will be
813   replaced with a single SP before interpretation of the TEXT value.
814</t>
815<t>
816   Hexadecimal numeric characters are used in several protocol elements.
817</t>
818<figure><iref primary="true" item="Grammar" subitem="HEX"/><artwork type="abnf2616"><![CDATA[
819    HEX            = "A" | "B" | "C" | "D" | "E" | "F"
820                   | "a" | "b" | "c" | "d" | "e" | "f" | DIGIT
821]]></artwork></figure>
822<t>
823   Many HTTP/1.1 header field values consist of words separated by LWS
824   or special characters. These special characters MUST be in a quoted
825   string to be used within a parameter value (as defined in
826   <xref target="transfer.codings"/>).
827</t>
828<figure><iref primary="true" item="Grammar" subitem="token"/><iref primary="true" item="Grammar" subitem="separators"/><artwork type="abnf2616"><![CDATA[
829    token          = 1*<any CHAR except CTLs or separators>
830    separators     = "(" | ")" | "<" | ">" | "@"
831                   | "," | ";" | ":" | "\" | <">
832                   | "/" | "[" | "]" | "?" | "="
833                   | "{" | "}" | SP | HT
834]]></artwork></figure>
835<t>
836   Comments can be included in some HTTP header fields by surrounding
837   the comment text with parentheses. Comments are only allowed in
838   fields containing "comment" as part of their field value definition.
839   In all other fields, parentheses are considered part of the field
840   value.
841</t>
842<figure><iref primary="true" item="Grammar" subitem="comment"/><iref primary="true" item="Grammar" subitem="ctext"/><artwork type="abnf2616"><![CDATA[
843    comment        = "(" *( ctext | quoted-pair | comment ) ")"
844    ctext          = <any TEXT excluding "(" and ")">
845]]></artwork></figure>
846<t>
847   A string of text is parsed as a single word if it is quoted using
848   double-quote marks.
849</t>
850<figure><iref primary="true" item="Grammar" subitem="quoted-string"/><iref primary="true" item="Grammar" subitem="qdtext"/><artwork type="abnf2616"><![CDATA[
851    quoted-string  = ( <"> *(qdtext | quoted-pair ) <"> )
852    qdtext         = <any TEXT except <">>
853]]></artwork></figure>
854<t>
855   The backslash character ("\") MAY be used as a single-character
856   quoting mechanism only within quoted-string and comment constructs.
857</t>
858<figure><iref primary="true" item="Grammar" subitem="quoted-pair"/><artwork type="abnf2616"><![CDATA[
859    quoted-pair    = "\" CHAR
860]]></artwork></figure>
861</section>
862</section>
863
864<section title="Protocol Parameters" anchor="protocol.parameters">
865
866<section title="HTTP Version" anchor="http.version">
867<t>
868   HTTP uses a "&lt;major&gt;.&lt;minor&gt;" numbering scheme to indicate versions
869   of the protocol. The protocol versioning policy is intended to allow
870   the sender to indicate the format of a message and its capacity for
871   understanding further HTTP communication, rather than the features
872   obtained via that communication. No change is made to the version
873   number for the addition of message components which do not affect
874   communication behavior or which only add to extensible field values.
875   The &lt;minor&gt; number is incremented when the changes made to the
876   protocol add features which do not change the general message parsing
877   algorithm, but which may add to the message semantics and imply
878   additional capabilities of the sender. The &lt;major&gt; number is
879   incremented when the format of a message within the protocol is
880   changed. See RFC 2145 <xref target="RFC2145"/> for a fuller explanation.
881</t>
882<t>
883   The version of an HTTP message is indicated by an HTTP-Version field
884   in the first line of the message.
885</t>
886<figure><iref primary="true" item="Grammar" subitem="HTTP-Version"/><artwork type="abnf2616"><![CDATA[
887       HTTP-Version   = "HTTP" "/" 1*DIGIT "." 1*DIGIT
888]]></artwork></figure>
889<t>
890   Note that the major and minor numbers MUST be treated as separate
891   integers and that each MAY be incremented higher than a single digit.
892   Thus, HTTP/2.4 is a lower version than HTTP/2.13, which in turn is
893   lower than HTTP/12.3. Leading zeros MUST be ignored by recipients and
894   MUST NOT be sent.
895</t>
896<t>
897   An application that sends a request or response message that includes
898   HTTP-Version of "HTTP/1.1" MUST be at least conditionally compliant
899   with this specification. Applications that are at least conditionally
900   compliant with this specification SHOULD use an HTTP-Version of
901   "HTTP/1.1" in their messages, and MUST do so for any message that is
902   not compatible with HTTP/1.0. For more details on when to send
903   specific HTTP-Version values, see RFC 2145 <xref target="RFC2145"/>.
904</t>
905<t>
906   The HTTP version of an application is the highest HTTP version for
907   which the application is at least conditionally compliant.
908</t>
909<t>
910   Proxy and gateway applications need to be careful when forwarding
911   messages in protocol versions different from that of the application.
912   Since the protocol version indicates the protocol capability of the
913   sender, a proxy/gateway MUST NOT send a message with a version
914   indicator which is greater than its actual version. If a higher
915   version request is received, the proxy/gateway MUST either downgrade
916   the request version, or respond with an error, or switch to tunnel
917   behavior.
918</t>
919<t>
920   Due to interoperability problems with HTTP/1.0 proxies discovered
921   since the publication of RFC 2068 <xref target="RFC2068"/>, caching proxies MUST, gateways
922   MAY, and tunnels MUST NOT upgrade the request to the highest version
923   they support. The proxy/gateway's response to that request MUST be in
924   the same major version as the request.
925</t>
926<t>
927  <list>
928    <t>
929      Note: Converting between versions of HTTP may involve modification
930      of header fields required or forbidden by the versions involved.
931    </t>
932  </list>
933</t>
934</section>
935
936<section title="Uniform Resource Identifiers" anchor="uri">
937<t>
938   URIs have been known by many names: WWW addresses, Universal Document
939   Identifiers, Universal Resource Identifiers <xref target="RFC1630"/>, and finally the
940   combination of Uniform Resource Locators (URL) <xref target="RFC1738"/> and Names (URN)
941   <xref target="RFC1737"/>. As far as HTTP is concerned, Uniform Resource Identifiers are
942   simply formatted strings which identify--via name, location, or any
943   other characteristic--a resource.
944</t>
945
946<section title="General Syntax" anchor="general.syntax">
947<t>
948   URIs in HTTP can be represented in absolute form or relative to some
949   known base URI <xref target="RFC1808"/>, depending upon the context of their use. The two
950   forms are differentiated by the fact that absolute URIs always begin
951   with a scheme name followed by a colon. For definitive information on
952   URL syntax and semantics, see "Uniform Resource Identifiers (URI):
953   Generic Syntax and Semantics," RFC 2396 <xref target="RFC2396"/> (which replaces RFCs
954   1738 <xref target="RFC1738"/> and RFC 1808 <xref target="RFC1808"/>). This specification adopts the
955   definitions of "URI-reference", "absoluteURI", "relativeURI", "port",
956   "host","abs_path", "rel_path", and "authority" from that
957   specification.
958</t>
959<t>
960   The HTTP protocol does not place any a priori limit on the length of
961   a URI. Servers MUST be able to handle the URI of any resource they
962   serve, and SHOULD be able to handle URIs of unbounded length if they
963   provide GET-based forms that could generate such URIs. A server
964   SHOULD return 414 (Request-URI Too Long) status if a URI is longer
965   than the server can handle (see Section 9.4.15 of <xref target="Part2"/>).
966</t>
967<t>
968  <list>
969    <t>
970      Note: Servers ought to be cautious about depending on URI lengths
971      above 255 bytes, because some older client or proxy
972      implementations might not properly support these lengths.
973    </t>
974  </list>
975</t>
976</section>
977
978<section title="http URL" anchor="http.url">
979<t>
980   The "http" scheme is used to locate network resources via the HTTP
981   protocol. This section defines the scheme-specific syntax and
982   semantics for http URLs.
983</t>
984<figure><iref primary="true" item="Grammar" subitem="http_URL"/><artwork type="abnf2616"><![CDATA[
985http_URL = "http:" "//" host [ ":" port ] [ abs_path [ "?" query ]]
986]]></artwork></figure>
987<t>
988   If the port is empty or not given, port 80 is assumed. The semantics
989   are that the identified resource is located at the server listening
990   for TCP connections on that port of that host, and the Request-URI
991   for the resource is abs_path (<xref target="request-uri"/>). The use of IP addresses
992   in URLs SHOULD be avoided whenever possible (see RFC 1900 <xref target="RFC1900"/>). If
993   the abs_path is not present in the URL, it MUST be given as "/" when
994   used as a Request-URI for a resource (<xref target="request-uri"/>). If a proxy
995
996   receives a host name which is not a fully qualified domain name, it
997   MAY add its domain to the host name it received. If a proxy receives
998   a fully qualified domain name, the proxy MUST NOT change the host
999   name.
1000</t>
1001</section>
1002
1003<section title="URI Comparison" anchor="uri.comparison">
1004<t>
1005   When comparing two URIs to decide if they match or not, a client
1006   SHOULD use a case-sensitive octet-by-octet comparison of the entire
1007   URIs, with these exceptions:
1008  <list style="symbols">
1009    <t>A port that is empty or not given is equivalent to the default
1010        port for that URI-reference;</t>
1011    <t>Comparisons of host names MUST be case-insensitive;</t>
1012    <t>Comparisons of scheme names MUST be case-insensitive;</t>
1013    <t>An empty abs_path is equivalent to an abs_path of "/".</t>
1014  </list>
1015</t>
1016<t>
1017   Characters other than those in the "reserved" and "unsafe" sets (see
1018   RFC 2396 <xref target="RFC2396"/>) are equivalent to their ""%" HEX HEX" encoding.
1019</t>
1020<t>
1021   For example, the following three URIs are equivalent:
1022</t>
1023<figure><artwork type="example"><![CDATA[
1024   http://abc.com:80/~smith/home.html
1025   http://ABC.com/%7Esmith/home.html
1026   http://ABC.com:/%7esmith/home.html
1027]]></artwork></figure>
1028</section>
1029</section>
1030
1031<section title="Date/Time Formats" anchor="date.time.formats">
1032<section title="Full Date" anchor="full.date">
1033<t>
1034   HTTP applications have historically allowed three different formats
1035   for the representation of date/time stamps:
1036</t>
1037<figure><artwork type="example"><![CDATA[
1038   Sun, 06 Nov 1994 08:49:37 GMT  ; RFC 822, updated by RFC 1123
1039   Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
1040   Sun Nov  6 08:49:37 1994       ; ANSI C's asctime() format
1041]]></artwork></figure>
1042<t>
1043   The first format is preferred as an Internet standard and represents
1044   a fixed-length subset of that defined by RFC 1123 <xref target="RFC1123"/> (an update to
1045   RFC 822 <xref target="RFC822"/>). The second format is in common use, but is based on the
1046   obsolete RFC 850 <xref target="RFC1036"/> date format and lacks a four-digit year.
1047   HTTP/1.1 clients and servers that parse the date value MUST accept
1048   all three formats (for compatibility with HTTP/1.0), though they MUST
1049   only generate the RFC 1123 format for representing HTTP-date values
1050   in header fields. See <xref target="tolerant.applications"/> for further information.
1051</t>
1052<t><list><t>
1053      Note: Recipients of date values are encouraged to be robust in
1054      accepting date values that may have been sent by non-HTTP
1055      applications, as is sometimes the case when retrieving or posting
1056      messages via proxies/gateways to SMTP or NNTP.
1057</t></list></t>
1058<t>
1059   All HTTP date/time stamps MUST be represented in Greenwich Mean Time
1060   (GMT), without exception. For the purposes of HTTP, GMT is exactly
1061   equal to UTC (Coordinated Universal Time). This is indicated in the
1062   first two formats by the inclusion of "GMT" as the three-letter
1063   abbreviation for time zone, and MUST be assumed when reading the
1064   asctime format. HTTP-date is case sensitive and MUST NOT include
1065   additional LWS beyond that specifically included as SP in the
1066   grammar.
1067</t>
1068<figure><iref primary="true" item="Grammar" subitem="HTTP-date"/><iref primary="true" item="Grammar" subitem="rfc1123-date"/><iref primary="true" item="Grammar" subitem="rfc850-date"/><iref primary="true" item="Grammar" subitem="asctime-date"/><iref primary="true" item="Grammar" subitem="date1"/><iref primary="true" item="Grammar" subitem="date2"/><iref primary="true" item="Grammar" subitem="date3"/><iref primary="true" item="Grammar" subitem="time"/><iref primary="true" item="Grammar" subitem="wkday"/><iref primary="true" item="Grammar" subitem="weekday"/><iref primary="true" item="Grammar" subitem="month"/><artwork type="abnf2616"><![CDATA[
1069    HTTP-date    = rfc1123-date | rfc850-date | asctime-date
1070    rfc1123-date = wkday "," SP date1 SP time SP "GMT"
1071    rfc850-date  = weekday "," SP date2 SP time SP "GMT"
1072    asctime-date = wkday SP date3 SP time SP 4DIGIT
1073    date1        = 2DIGIT SP month SP 4DIGIT
1074                   ; day month year (e.g., 02 Jun 1982)
1075    date2        = 2DIGIT "-" month "-" 2DIGIT
1076                   ; day-month-year (e.g., 02-Jun-82)
1077    date3        = month SP ( 2DIGIT | ( SP 1DIGIT ))
1078                   ; month day (e.g., Jun  2)
1079    time         = 2DIGIT ":" 2DIGIT ":" 2DIGIT
1080                   ; 00:00:00 - 23:59:59
1081    wkday        = "Mon" | "Tue" | "Wed"
1082                 | "Thu" | "Fri" | "Sat" | "Sun"
1083    weekday      = "Monday" | "Tuesday" | "Wednesday"
1084                 | "Thursday" | "Friday" | "Saturday" | "Sunday"
1085    month        = "Jan" | "Feb" | "Mar" | "Apr"
1086                 | "May" | "Jun" | "Jul" | "Aug"
1087                 | "Sep" | "Oct" | "Nov" | "Dec"
1088]]></artwork></figure>
1089<t>
1090      Note: HTTP requirements for the date/time stamp format apply only
1091      to their usage within the protocol stream. Clients and servers are
1092      not required to use these formats for user presentation, request
1093      logging, etc.
1094</t>
1095</section>
1096</section>
1097
1098<section title="Transfer Codings" anchor="transfer.codings">
1099<t>
1100   Transfer-coding values are used to indicate an encoding
1101   transformation that has been, can be, or may need to be applied to an
1102   entity-body in order to ensure "safe transport" through the network.
1103   This differs from a content coding in that the transfer-coding is a
1104   property of the message, not of the original entity.
1105</t>
1106<figure><iref primary="true" item="Grammar" subitem="transfer-coding"/><iref primary="true" item="Grammar" subitem="transfer-extension"/><artwork type="abnf2616"><![CDATA[
1107    transfer-coding         = "chunked" | transfer-extension
1108    transfer-extension      = token *( ";" parameter )
1109]]></artwork></figure>
1110<t>
1111   Parameters are in  the form of attribute/value pairs.
1112</t>
1113<figure><iref primary="true" item="Grammar" subitem="parameter"/><iref primary="true" item="Grammar" subitem="attribute"/><iref primary="true" item="Grammar" subitem="value"/><artwork type="abnf2616"><![CDATA[
1114    parameter               = attribute "=" value
1115    attribute               = token
1116    value                   = token | quoted-string
1117]]></artwork></figure>
1118<t>
1119   All transfer-coding values are case-insensitive. HTTP/1.1 uses
1120   transfer-coding values in the TE header field (<xref target="header.te"/>) and in
1121   the Transfer-Encoding header field (<xref target="header.transfer-encoding"/>).
1122</t>
1123<t>
1124   Whenever a transfer-coding is applied to a message-body, the set of
1125   transfer-codings MUST include "chunked", unless the message is
1126   terminated by closing the connection. When the "chunked" transfer-coding
1127   is used, it MUST be the last transfer-coding applied to the
1128   message-body. The "chunked" transfer-coding MUST NOT be applied more
1129   than once to a message-body. These rules allow the recipient to
1130   determine the transfer-length of the message (<xref target="message.length"/>).
1131</t>
1132<t>
1133   Transfer-codings are analogous to the Content-Transfer-Encoding
1134   values of MIME <xref target="RFC2045"/>, which were designed to enable safe transport of
1135   binary data over a 7-bit transport service. However, safe transport
1136   has a different focus for an 8bit-clean transfer protocol. In HTTP,
1137   the only unsafe characteristic of message-bodies is the difficulty in
1138   determining the exact body length (<xref target="message.length"/>), or the desire to
1139   encrypt data over a shared transport.
1140</t>
1141<t>
1142   The Internet Assigned Numbers Authority (IANA) acts as a registry for
1143   transfer-coding value tokens. Initially, the registry contains the
1144   following tokens: "chunked" (<xref target="chunked.transfer.encoding"/>), "identity" (section
1145   3.6.2), "gzip" (<xref target="Part3"/>), "compress" (<xref target="Part3"/>), and "deflate"
1146   (<xref target="Part3"/>).
1147</t>
1148<t>
1149   New transfer-coding value tokens SHOULD be registered in the same way
1150   as new content-coding value tokens (<xref target="Part3"/>).
1151</t>
1152<t>
1153   A server which receives an entity-body with a transfer-coding it does
1154   not understand SHOULD return 501 (Unimplemented), and close the
1155   connection. A server MUST NOT send transfer-codings to an HTTP/1.0
1156   client.
1157</t>
1158
1159<section title="Chunked Transfer Coding" anchor="chunked.transfer.encoding">
1160<t>
1161   The chunked encoding modifies the body of a message in order to
1162   transfer it as a series of chunks, each with its own size indicator,
1163   followed by an OPTIONAL trailer containing entity-header fields. This
1164   allows dynamically produced content to be transferred along with the
1165   information necessary for the recipient to verify that it has
1166   received the full message.
1167</t>
1168<figure><iref primary="true" item="Grammar" subitem="Chunked-Body"/><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="true" item="Grammar" subitem="chunk-extension"/><iref primary="true" item="Grammar" subitem="chunk-ext-name"/><iref primary="true" item="Grammar" subitem="chunk-ext-val"/><iref primary="true" item="Grammar" subitem="chunk-data"/><iref primary="true" item="Grammar" subitem="trailer"/><artwork type="abnf2616"><![CDATA[
1169    Chunked-Body   = *chunk
1170                     last-chunk
1171                     trailer
1172                     CRLF
1173
1174    chunk          = chunk-size [ chunk-extension ] CRLF
1175                     chunk-data CRLF
1176    chunk-size     = 1*HEX
1177    last-chunk     = 1*("0") [ chunk-extension ] CRLF
1178
1179    chunk-extension= *( ";" chunk-ext-name [ "=" chunk-ext-val ] )
1180    chunk-ext-name = token
1181    chunk-ext-val  = token | quoted-string
1182    chunk-data     = chunk-size(OCTET)
1183    trailer        = *(entity-header CRLF)
1184]]></artwork></figure>
1185<t>
1186   The chunk-size field is a string of hex digits indicating the size of
1187   the chunk. The chunked encoding is ended by any chunk whose size is
1188   zero, followed by the trailer, which is terminated by an empty line.
1189</t>
1190<t>
1191   The trailer allows the sender to include additional HTTP header
1192   fields at the end of the message. The Trailer header field can be
1193   used to indicate which header fields are included in a trailer (see
1194   <xref target="header.trailer"/>).
1195</t>
1196<t>
1197   A server using chunked transfer-coding in a response MUST NOT use the
1198   trailer for any header fields unless at least one of the following is
1199   true:
1200  <list style="numbers">
1201    <t>the request included a TE header field that indicates "trailers" is
1202     acceptable in the transfer-coding of the  response, as described in
1203     <xref target="header.te"/>; or,</t>
1204
1205    <t>the server is the origin server for the response, the trailer
1206     fields consist entirely of optional metadata, and the recipient
1207     could use the message (in a manner acceptable to the origin server)
1208     without receiving this metadata.  In other words, the origin server
1209     is willing to accept the possibility that the trailer fields might
1210     be silently discarded along the path to the client.</t>
1211  </list>
1212</t>
1213<t>
1214   This requirement prevents an interoperability failure when the
1215   message is being received by an HTTP/1.1 (or later) proxy and
1216   forwarded to an HTTP/1.0 recipient. It avoids a situation where
1217   compliance with the protocol would have necessitated a possibly
1218   infinite buffer on the proxy.
1219</t>
1220<t>
1221   A process for decoding the "chunked" transfer-coding
1222   can be represented in pseudo-code as:
1223</t>
1224<figure><artwork type="code"><![CDATA[
1225    length := 0
1226    read chunk-size, chunk-extension (if any) and CRLF
1227    while (chunk-size > 0) {
1228       read chunk-data and CRLF
1229       append chunk-data to entity-body
1230       length := length + chunk-size
1231       read chunk-size and CRLF
1232    }
1233    read entity-header
1234    while (entity-header not empty) {
1235       append entity-header to existing header fields
1236       read entity-header
1237    }
1238    Content-Length := length
1239    Remove "chunked" from Transfer-Encoding
1240]]></artwork></figure>
1241<t>
1242   All HTTP/1.1 applications MUST be able to receive and decode the
1243   "chunked" transfer-coding, and MUST ignore chunk-extension extensions
1244   they do not understand.
1245</t>
1246</section>
1247</section>
1248
1249</section>
1250
1251<section title="HTTP Message" anchor="http.message">
1252
1253<section title="Message Types" anchor="message.types">
1254<t>
1255   HTTP messages consist of requests from client to server and responses
1256   from server to client.
1257</t>
1258<figure><iref primary="true" item="Grammar" subitem="HTTP-message"/><artwork type="abnf2616"><![CDATA[
1259    HTTP-message   = Request | Response     ; HTTP/1.1 messages
1260]]></artwork></figure>
1261<t>
1262   Request (<xref target="request"/>) and Response (<xref target="response"/>) messages use the generic
1263   message format of RFC 822 <xref target="RFC822"/> for transferring entities (the payload
1264   of the message). Both types of message consist of a start-line, zero
1265   or more header fields (also known as "headers"), an empty line (i.e.,
1266   a line with nothing preceding the CRLF) indicating the end of the
1267   header fields, and possibly a message-body.
1268</t>
1269<figure><iref primary="true" item="Grammar" subitem="generic-message"/><iref primary="true" item="Grammar" subitem="start-line"/><artwork type="abnf2616"><![CDATA[
1270     generic-message = start-line
1271                       *(message-header CRLF)
1272                       CRLF
1273                       [ message-body ]
1274     start-line      = Request-Line | Status-Line
1275]]></artwork></figure>
1276<t>
1277   In the interest of robustness, servers SHOULD ignore any empty
1278   line(s) received where a Request-Line is expected. In other words, if
1279   the server is reading the protocol stream at the beginning of a
1280   message and receives a CRLF first, it should ignore the CRLF.
1281</t>
1282<t>
1283   Certain buggy HTTP/1.0 client implementations generate extra CRLF's
1284   after a POST request. To restate what is explicitly forbidden by the
1285   BNF, an HTTP/1.1 client MUST NOT preface or follow a request with an
1286   extra CRLF.
1287</t>
1288</section>
1289
1290<section title="Message Headers" anchor="message.headers">
1291<t>
1292   HTTP header fields, which include general-header (<xref target="general.header.fields"/>),
1293   request-header (Section 4 of <xref target="Part2"/>), response-header (Section 6 of <xref target="Part2"/>), and
1294   entity-header (Section 3.1 of <xref target="Part3"/>) fields, follow the same generic format as
1295   that given in Section 3.1 of RFC 822 <xref target="RFC822"/>. Each header field consists
1296   of a name followed by a colon (":") and the field value. Field names
1297   are case-insensitive. The field value MAY be preceded by any amount
1298   of LWS, though a single SP is preferred. Header fields can be
1299   extended over multiple lines by preceding each extra line with at
1300   least one SP or HT. Applications ought to follow "common form", where
1301   one is known or indicated, when generating HTTP constructs, since
1302   there might exist some implementations that fail to accept anything
1303   beyond the common forms.
1304</t>
1305<figure><iref primary="true" item="Grammar" subitem="message-header"/><iref primary="true" item="Grammar" subitem="field-name"/><iref primary="true" item="Grammar" subitem="field-value"/><iref primary="true" item="Grammar" subitem="field-content"/><artwork type="abnf2616"><![CDATA[
1306    message-header = field-name ":" [ field-value ]
1307    field-name     = token
1308    field-value    = *( field-content | LWS )
1309    field-content  = <the OCTETs making up the field-value
1310                     and consisting of either *TEXT or combinations
1311                     of token, separators, and quoted-string>
1312]]></artwork></figure>
1313<t>
1314   The field-content does not include any leading or trailing LWS:
1315   linear white space occurring before the first non-whitespace
1316   character of the field-value or after the last non-whitespace
1317   character of the field-value. Such leading or trailing LWS MAY be
1318   removed without changing the semantics of the field value. Any LWS
1319   that occurs between field-content MAY be replaced with a single SP
1320   before interpreting the field value or forwarding the message
1321   downstream.
1322</t>
1323<t>
1324   The order in which header fields with differing field names are
1325   received is not significant. However, it is "good practice" to send
1326   general-header fields first, followed by request-header or response-header
1327   fields, and ending with the entity-header fields.
1328</t>
1329<t>
1330   Multiple message-header fields with the same field-name MAY be
1331   present in a message if and only if the entire field-value for that
1332   header field is defined as a comma-separated list [i.e., #(values)].
1333   It MUST be possible to combine the multiple header fields into one
1334   "field-name: field-value" pair, without changing the semantics of the
1335   message, by appending each subsequent field-value to the first, each
1336   separated by a comma. The order in which header fields with the same
1337   field-name are received is therefore significant to the
1338   interpretation of the combined field value, and thus a proxy MUST NOT
1339   change the order of these field values when a message is forwarded.
1340</t>
1341</section>
1342
1343<section title="Message Body" anchor="message.body">
1344<t>
1345   The message-body (if any) of an HTTP message is used to carry the
1346   entity-body associated with the request or response. The message-body
1347   differs from the entity-body only when a transfer-coding has been
1348   applied, as indicated by the Transfer-Encoding header field (<xref target="header.transfer-encoding"/>).
1349</t>
1350<figure><iref primary="true" item="Grammar" subitem="message-body"/><artwork type="abnf2616"><![CDATA[
1351    message-body = entity-body
1352                 | <entity-body encoded as per Transfer-Encoding>
1353]]></artwork></figure>
1354<t>
1355   Transfer-Encoding MUST be used to indicate any transfer-codings
1356   applied by an application to ensure safe and proper transfer of the
1357   message. Transfer-Encoding is a property of the message, not of the
1358   entity, and thus MAY be added or removed by any application along the
1359   request/response chain. (However, <xref target="transfer.codings"/> places restrictions on
1360   when certain transfer-codings may be used.)
1361</t>
1362<t>
1363   The rules for when a message-body is allowed in a message differ for
1364   requests and responses.
1365</t>
1366<t>
1367   The presence of a message-body in a request is signaled by the
1368   inclusion of a Content-Length or Transfer-Encoding header field in
1369   the request's message-headers. A message-body MUST NOT be included in
1370   a request if the specification of the request method (Section 3 of <xref target="Part2"/>)
1371   does not allow sending an entity-body in requests. A server SHOULD
1372   read and forward a message-body on any request; if the request method
1373   does not include defined semantics for an entity-body, then the
1374   message-body SHOULD be ignored when handling the request.
1375</t>
1376<t>
1377   For response messages, whether or not a message-body is included with
1378   a message is dependent on both the request method and the response
1379   status code (<xref target="status.code.and.reason.phrase"/>). All responses to the HEAD request method
1380   MUST NOT include a message-body, even though the presence of entity-header
1381   fields might lead one to believe they do. All 1xx
1382   (informational), 204 (no content), and 304 (not modified) responses
1383   MUST NOT include a message-body. All other responses do include a
1384   message-body, although it MAY be of zero length.
1385</t>
1386</section>
1387
1388<section title="Message Length" anchor="message.length">
1389<t>
1390   The transfer-length of a message is the length of the message-body as
1391   it appears in the message; that is, after any transfer-codings have
1392   been applied. When a message-body is included with a message, the
1393   transfer-length of that body is determined by one of the following
1394   (in order of precedence):
1395</t>
1396<t>
1397  <list style="numbers">
1398    <t>
1399     Any response message which "MUST NOT" include a message-body (such
1400     as the 1xx, 204, and 304 responses and any response to a HEAD
1401     request) is always terminated by the first empty line after the
1402     header fields, regardless of the entity-header fields present in
1403     the message.
1404    </t>
1405    <t>
1406     If a Transfer-Encoding header field (<xref target="header.transfer-encoding"/>) is present and
1407     has any value other than "identity", then the transfer-length is
1408     defined by use of the "chunked" transfer-coding (<xref target="transfer.codings"/>),
1409     unless the message is terminated by closing the connection.
1410    </t>
1411    <t>
1412     If a Content-Length header field (<xref target="header.content-length"/>) is present, its
1413     decimal value in OCTETs represents both the entity-length and the
1414     transfer-length. The Content-Length header field MUST NOT be sent
1415     if these two lengths are different (i.e., if a Transfer-Encoding
1416     header field is present). If a message is received with both a
1417     Transfer-Encoding header field and a Content-Length header field,
1418     the latter MUST be ignored.
1419    </t>
1420    <t>
1421     If the message uses the media type "multipart/byteranges", and the
1422     ransfer-length is not otherwise specified, then this self-elimiting
1423     media type defines the transfer-length. This media type
1424     UST NOT be used unless the sender knows that the recipient can arse
1425     it; the presence in a request of a Range header with ultiple byte-range
1426     specifiers from a 1.1 client implies that the lient can parse
1427     multipart/byteranges responses.
1428    <list style="empty"><t>
1429       A range header might be forwarded by a 1.0 proxy that does not
1430       understand multipart/byteranges; in this case the server MUST
1431       delimit the message using methods defined in items 1, 3 or 5 of
1432       this section.
1433    </t></list>
1434    </t>
1435    <t>
1436     By the server closing the connection. (Closing the connection
1437     cannot be used to indicate the end of a request body, since that
1438     would leave no possibility for the server to send back a response.)
1439    </t>
1440  </list>
1441</t>
1442<t>
1443   For compatibility with HTTP/1.0 applications, HTTP/1.1 requests
1444   containing a message-body MUST include a valid Content-Length header
1445   field unless the server is known to be HTTP/1.1 compliant. If a
1446   request contains a message-body and a Content-Length is not given,
1447   the server SHOULD respond with 400 (bad request) if it cannot
1448   determine the length of the message, or with 411 (length required) if
1449   it wishes to insist on receiving a valid Content-Length.
1450</t>
1451<t>
1452   All HTTP/1.1 applications that receive entities MUST accept the
1453   "chunked" transfer-coding (<xref target="transfer.codings"/>), thus allowing this mechanism
1454   to be used for messages when the message length cannot be determined
1455   in advance.
1456</t>
1457<t>
1458   Messages MUST NOT include both a Content-Length header field and a
1459   non-identity transfer-coding. If the message does include a non-identity
1460   transfer-coding, the Content-Length MUST be ignored.
1461</t>
1462<t>
1463   When a Content-Length is given in a message where a message-body is
1464   allowed, its field value MUST exactly match the number of OCTETs in
1465   the message-body. HTTP/1.1 user agents MUST notify the user when an
1466   invalid length is received and detected.
1467</t>
1468</section>
1469
1470<section title="General Header Fields" anchor="general.header.fields">
1471<t>
1472   There are a few header fields which have general applicability for
1473   both request and response messages, but which do not apply to the
1474   entity being transferred. These header fields apply only to the
1475   message being transmitted.
1476</t>
1477<figure><iref primary="true" item="Grammar" subitem="general-header"/><artwork type="abnf2616"><![CDATA[
1478    general-header = Cache-Control            ; [Part6], Section 3.2
1479                   | Connection               ; Section 8.1
1480                   | Date                     ; Section 8.3
1481                   | Pragma                   ; [Part6], Section 3.4
1482                   | Trailer                  ; Section 8.6
1483                   | Transfer-Encoding        ; Section 8.7
1484                   | Upgrade                  ; Section 8.8
1485                   | Via                      ; Section 8.9
1486                   | Warning                  ; [Part6], Section 3.6
1487]]></artwork></figure>
1488<t>
1489   General-header field names can be extended reliably only in
1490   combination with a change in the protocol version. However, new or
1491   experimental header fields may be given the semantics of general
1492   header fields if all parties in the communication recognize them to
1493   be general-header fields. Unrecognized header fields are treated as
1494   entity-header fields.
1495</t>
1496</section>
1497</section>
1498
1499<section title="Request" anchor="request">
1500<t>
1501   A request message from a client to a server includes, within the
1502   first line of that message, the method to be applied to the resource,
1503   the identifier of the resource, and the protocol version in use.
1504</t>
1505<!--                 Host                      ; should be moved here eventually -->
1506<figure><iref primary="true" item="Grammar" subitem="Request"/><artwork type="abnf2616"><![CDATA[
1507     Request       = Request-Line              ; Section 5.1
1508                     *(( general-header        ; Section 4.5
1509                      | request-header         ; [Part2], Section 4
1510                      | entity-header ) CRLF)  ; [Part3], Section 3.1
1511                     CRLF
1512                     [ message-body ]          ; Section 4.3
1513]]></artwork></figure>
1514
1515<section title="Request-Line" anchor="request-line">
1516<t>
1517   The Request-Line begins with a method token, followed by the
1518   Request-URI and the protocol version, and ending with CRLF. The
1519   elements are separated by SP characters. No CR or LF is allowed
1520   except in the final CRLF sequence.
1521</t>
1522<figure><iref primary="true" item="Grammar" subitem="Request-Line"/><artwork type="abnf2616"><![CDATA[
1523     Request-Line   = Method SP Request-URI SP HTTP-Version CRLF
1524]]></artwork></figure>
1525
1526<section title="Method" anchor="method">
1527<t>
1528   The Method  token indicates the method to be performed on the
1529   resource identified by the Request-URI. The method is case-sensitive.
1530</t>
1531<figure><iref primary="true" item="Grammar" subitem="Method"/><iref primary="true" item="Grammar" subitem="extension-method"/><artwork type="abnf2616"><![CDATA[
1532    Method         = token
1533]]></artwork></figure>
1534</section>
1535
1536<section title="Request-URI" anchor="request-uri">
1537<t>
1538   The Request-URI is a Uniform Resource Identifier (<xref target="uri"/>) and
1539   identifies the resource upon which to apply the request.
1540</t>
1541<figure><iref primary="true" item="Grammar" subitem="Request-URI"/><artwork type="abnf2616"><![CDATA[
1542    Request-URI    = "*" | absoluteURI | abs_path | authority
1543]]></artwork></figure>
1544<t>
1545   The four options for Request-URI are dependent on the nature of the
1546   request. The asterisk "*" means that the request does not apply to a
1547   particular resource, but to the server itself, and is only allowed
1548   when the method used does not necessarily apply to a resource. One
1549   example would be
1550</t>
1551<figure><artwork type="example"><![CDATA[
1552    OPTIONS * HTTP/1.1
1553]]></artwork></figure>
1554<t>
1555   The absoluteURI form is REQUIRED when the request is being made to a
1556   proxy. The proxy is requested to forward the request or service it
1557   from a valid cache, and return the response. Note that the proxy MAY
1558   forward the request on to another proxy or directly to the server
1559   specified by the absoluteURI. In order to avoid request loops, a
1560   proxy MUST be able to recognize all of its server names, including
1561   any aliases, local variations, and the numeric IP address. An example
1562   Request-Line would be:
1563</t>
1564<figure><artwork type="example"><![CDATA[
1565    GET http://www.w3.org/pub/WWW/TheProject.html HTTP/1.1
1566]]></artwork></figure>
1567<t>
1568   To allow for transition to absoluteURIs in all requests in future
1569   versions of HTTP, all HTTP/1.1 servers MUST accept the absoluteURI
1570   form in requests, even though HTTP/1.1 clients will only generate
1571   them in requests to proxies.
1572</t>
1573<t>
1574   The authority form is only used by the CONNECT method (Section 8.9 of <xref target="Part2"/>).
1575</t>
1576<t>
1577   The most common form of Request-URI is that used to identify a
1578   resource on an origin server or gateway. In this case the absolute
1579   path of the URI MUST be transmitted (see <xref target="general.syntax"/>, abs_path) as
1580   the Request-URI, and the network location of the URI (authority) MUST
1581   be transmitted in a Host header field. For example, a client wishing
1582   to retrieve the resource above directly from the origin server would
1583   create a TCP connection to port 80 of the host "www.w3.org" and send
1584   the lines:
1585</t>
1586<figure><artwork type="example"><![CDATA[
1587    GET /pub/WWW/TheProject.html HTTP/1.1
1588    Host: www.w3.org
1589]]></artwork></figure>
1590<t>
1591   followed by the remainder of the Request. Note that the absolute path
1592   cannot be empty; if none is present in the original URI, it MUST be
1593   given as "/" (the server root).
1594</t>
1595<t>
1596   The Request-URI is transmitted in the format specified in
1597   <xref target="general.syntax"/>. If the Request-URI is encoded using the "% HEX HEX" encoding
1598   <xref target="RFC2396"/>, the origin server MUST decode the Request-URI in order to
1599   properly interpret the request. Servers SHOULD respond to invalid
1600   Request-URIs with an appropriate status code.
1601</t>
1602<t>
1603   A transparent proxy MUST NOT rewrite the "abs_path" part of the
1604   received Request-URI when forwarding it to the next inbound server,
1605   except as noted above to replace a null abs_path with "/".
1606</t>
1607<t>
1608  <list><t>
1609      Note: The "no rewrite" rule prevents the proxy from changing the
1610      meaning of the request when the origin server is improperly using
1611      a non-reserved URI character for a reserved purpose.  Implementors
1612      should be aware that some pre-HTTP/1.1 proxies have been known to
1613      rewrite the Request-URI.
1614  </t></list>
1615</t>
1616</section>
1617</section>
1618
1619<section title="The Resource Identified by a Request" anchor="the.resource.identified.by.a.request">
1620<t>
1621   The exact resource identified by an Internet request is determined by
1622   examining both the Request-URI and the Host header field.
1623</t>
1624<t>
1625   An origin server that does not allow resources to differ by the
1626   requested host MAY ignore the Host header field value when
1627   determining the resource identified by an HTTP/1.1 request. (But see
1628   <xref target="changes.to.simplify.multi-homed.web.servers.and.conserve.ip.addresses"/>
1629   for other requirements on Host support in HTTP/1.1.)
1630</t>
1631<t>
1632   An origin server that does differentiate resources based on the host
1633   requested (sometimes referred to as virtual hosts or vanity host
1634   names) MUST use the following rules for determining the requested
1635   resource on an HTTP/1.1 request:
1636  <list style="numbers">
1637    <t>If Request-URI is an absoluteURI, the host is part of the
1638     Request-URI. Any Host header field value in the request MUST be
1639     ignored.</t>
1640    <t>If the Request-URI is not an absoluteURI, and the request includes
1641     a Host header field, the host is determined by the Host header
1642     field value.</t>
1643    <t>If the host as determined by rule 1 or 2 is not a valid host on
1644     the server, the response MUST be a 400 (Bad Request) error message.</t>
1645  </list>
1646</t>
1647<t>
1648   Recipients of an HTTP/1.0 request that lacks a Host header field MAY
1649   attempt to use heuristics (e.g., examination of the URI path for
1650   something unique to a particular host) in order to determine what
1651   exact resource is being requested.
1652</t>
1653</section>
1654
1655</section>
1656
1657
1658<section title="Response" anchor="response">
1659<t>
1660   After receiving and interpreting a request message, a server responds
1661   with an HTTP response message.
1662</t>
1663<figure><iref primary="true" item="Grammar" subitem="Response"/><artwork type="abnf2616"><![CDATA[
1664    Response      = Status-Line               ; Section 6.1
1665                    *(( general-header        ; Section 4.5
1666                     | response-header        ; [Part2], Section 6
1667                     | entity-header ) CRLF)  ; [Part3], Section 3.1
1668                    CRLF
1669                    [ message-body ]          ; Section 4.3
1670]]></artwork></figure>
1671
1672<section title="Status-Line" anchor="status-line">
1673<t>
1674   The first line of a Response message is the Status-Line, consisting
1675   of the protocol version followed by a numeric status code and its
1676   associated textual phrase, with each element separated by SP
1677   characters. No CR or LF is allowed except in the final CRLF sequence.
1678</t>
1679<figure><iref primary="true" item="Grammar" subitem="Status-Line"/><artwork type="abnf2616"><![CDATA[
1680    Status-Line = HTTP-Version SP Status-Code SP Reason-Phrase CRLF
1681]]></artwork></figure>
1682
1683<section title="Status Code and Reason Phrase" anchor="status.code.and.reason.phrase">
1684<t>
1685   The Status-Code element is a 3-digit integer result code of the
1686   attempt to understand and satisfy the request. These codes are fully
1687   defined in Section 9 of <xref target="Part2"/>. The Reason-Phrase is intended to give a short
1688   textual description of the Status-Code. The Status-Code is intended
1689   for use by automata and the Reason-Phrase is intended for the human
1690   user. The client is not required to examine or display the Reason-Phrase.
1691</t>
1692<t>
1693   The first digit of the Status-Code defines the class of response. The
1694   last two digits do not have any categorization role. There are 5
1695   values for the first digit:
1696  <list style="symbols">
1697    <t>
1698      1xx: Informational - Request received, continuing process
1699    </t>
1700    <t>
1701      2xx: Success - The action was successfully received,
1702        understood, and accepted
1703    </t>
1704    <t>
1705      3xx: Redirection - Further action must be taken in order to
1706        complete the request
1707    </t>
1708    <t>
1709      4xx: Client Error - The request contains bad syntax or cannot
1710        be fulfilled
1711    </t>
1712    <t>
1713      5xx: Server Error - The server failed to fulfill an apparently
1714        valid request
1715    </t>
1716  </list>
1717</t>
1718<figure><iref primary="true" item="Grammar" subitem="Status-Code"/><iref primary="true" item="Grammar" subitem="extension-code"/><iref primary="true" item="Grammar" subitem="Reason-Phrase"/><artwork type="abnf2616"><![CDATA[
1719   Status-Code    = 3DIGIT
1720   Reason-Phrase  = *<TEXT, excluding CR, LF>
1721]]></artwork></figure>
1722</section>
1723</section>
1724
1725</section>
1726
1727
1728<section title="Connections" anchor="connections">
1729
1730<section title="Persistent Connections" anchor="persistent.connections">
1731
1732<section title="Purpose" anchor="persistent.purpose">
1733<t>
1734   Prior to persistent connections, a separate TCP connection was
1735   established to fetch each URL, increasing the load on HTTP servers
1736   and causing congestion on the Internet. The use of inline images and
1737   other associated data often require a client to make multiple
1738   requests of the same server in a short amount of time. Analysis of
1739   these performance problems and results from a prototype
1740   implementation are available <xref target="Pad1995"/> <xref target="Spe"/>. Implementation experience and
1741   measurements of actual HTTP/1.1 (RFC 2068) implementations show good
1742   results <xref target="Nie1997"/>. Alternatives have also been explored, for example,
1743   T/TCP <xref target="Tou1998"/>.
1744</t>
1745<t>
1746   Persistent HTTP connections have a number of advantages:
1747  <list style="symbols">
1748      <t>
1749        By opening and closing fewer TCP connections, CPU time is saved
1750        in routers and hosts (clients, servers, proxies, gateways,
1751        tunnels, or caches), and memory used for TCP protocol control
1752        blocks can be saved in hosts.
1753      </t>
1754      <t>
1755        HTTP requests and responses can be pipelined on a connection.
1756        Pipelining allows a client to make multiple requests without
1757        waiting for each response, allowing a single TCP connection to
1758        be used much more efficiently, with much lower elapsed time.
1759      </t>
1760      <t>
1761        Network congestion is reduced by reducing the number of packets
1762        caused by TCP opens, and by allowing TCP sufficient time to
1763        determine the congestion state of the network.
1764      </t>
1765      <t>
1766        Latency on subsequent requests is reduced since there is no time
1767        spent in TCP's connection opening handshake.
1768      </t>
1769      <t>
1770        HTTP can evolve more gracefully, since errors can be reported
1771        without the penalty of closing the TCP connection. Clients using
1772        future versions of HTTP might optimistically try a new feature,
1773        but if communicating with an older server, retry with old
1774        semantics after an error is reported.
1775      </t>
1776    </list>
1777</t>
1778<t>
1779   HTTP implementations SHOULD implement persistent connections.
1780</t>
1781</section>
1782
1783<section title="Overall Operation" anchor="persistent.overall">
1784<t>
1785   A significant difference between HTTP/1.1 and earlier versions of
1786   HTTP is that persistent connections are the default behavior of any
1787   HTTP connection. That is, unless otherwise indicated, the client
1788   SHOULD assume that the server will maintain a persistent connection,
1789   even after error responses from the server.
1790</t>
1791<t>
1792   Persistent connections provide a mechanism by which a client and a
1793   server can signal the close of a TCP connection. This signaling takes
1794   place using the Connection header field (<xref target="header.connection"/>). Once a close
1795   has been signaled, the client MUST NOT send any more requests on that
1796   connection.
1797</t>
1798
1799<section title="Negotiation" anchor="persistent.negotiation">
1800<t>
1801   An HTTP/1.1 server MAY assume that a HTTP/1.1 client intends to
1802   maintain a persistent connection unless a Connection header including
1803   the connection-token "close" was sent in the request. If the server
1804   chooses to close the connection immediately after sending the
1805   response, it SHOULD send a Connection header including the
1806   connection-token close.
1807</t>
1808<t>
1809   An HTTP/1.1 client MAY expect a connection to remain open, but would
1810   decide to keep it open based on whether the response from a server
1811   contains a Connection header with the connection-token close. In case
1812   the client does not want to maintain a connection for more than that
1813   request, it SHOULD send a Connection header including the
1814   connection-token close.
1815</t>
1816<t>
1817   If either the client or the server sends the close token in the
1818   Connection header, that request becomes the last one for the
1819   connection.
1820</t>
1821<t>
1822   Clients and servers SHOULD NOT  assume that a persistent connection is
1823   maintained for HTTP versions less than 1.1 unless it is explicitly
1824   signaled. See <xref target="compatibility.with.http.1.0.persistent.connections"/> for more information on backward
1825   compatibility with HTTP/1.0 clients.
1826</t>
1827<t>
1828   In order to remain persistent, all messages on the connection MUST
1829   have a self-defined message length (i.e., one not defined by closure
1830   of the connection), as described in <xref target="message.length"/>.
1831</t>
1832</section>
1833
1834<section title="Pipelining" anchor="pipelining">
1835<t>
1836   A client that supports persistent connections MAY "pipeline" its
1837   requests (i.e., send multiple requests without waiting for each
1838   response). A server MUST send its responses to those requests in the
1839   same order that the requests were received.
1840</t>
1841<t>
1842   Clients which assume persistent connections and pipeline immediately
1843   after connection establishment SHOULD be prepared to retry their
1844   connection if the first pipelined attempt fails. If a client does
1845   such a retry, it MUST NOT pipeline before it knows the connection is
1846   persistent. Clients MUST also be prepared to resend their requests if
1847   the server closes the connection before sending all of the
1848   corresponding responses.
1849</t>
1850<t>
1851   Clients SHOULD NOT  pipeline requests using non-idempotent methods or
1852   non-idempotent sequences of methods (see Section 8.1.2 of <xref target="Part2"/>). Otherwise, a
1853   premature termination of the transport connection could lead to
1854   indeterminate results. A client wishing to send a non-idempotent
1855   request SHOULD wait to send that request until it has received the
1856   response status for the previous request.
1857</t>
1858</section>
1859</section>
1860
1861<section title="Proxy Servers" anchor="persistent.proxy">
1862<t>
1863   It is especially important that proxies correctly implement the
1864   properties of the Connection header field as specified in <xref target="header.connection"/>.
1865</t>
1866<t>
1867   The proxy server MUST signal persistent connections separately with
1868   its clients and the origin servers (or other proxy servers) that it
1869   connects to. Each persistent connection applies to only one transport
1870   link.
1871</t>
1872<t>
1873   A proxy server MUST NOT establish a HTTP/1.1 persistent connection
1874   with an HTTP/1.0 client (but see RFC 2068 <xref target="RFC2068"/> for information and
1875   discussion of the problems with the Keep-Alive header implemented by
1876   many HTTP/1.0 clients).
1877</t>
1878</section>
1879
1880<section title="Practical Considerations" anchor="persistent.practical">
1881<t>
1882   Servers will usually have some time-out value beyond which they will
1883   no longer maintain an inactive connection. Proxy servers might make
1884   this a higher value since it is likely that the client will be making
1885   more connections through the same server. The use of persistent
1886   connections places no requirements on the length (or existence) of
1887   this time-out for either the client or the server.
1888</t>
1889<t>
1890   When a client or server wishes to time-out it SHOULD issue a graceful
1891   close on the transport connection. Clients and servers SHOULD both
1892   constantly watch for the other side of the transport close, and
1893   respond to it as appropriate. If a client or server does not detect
1894   the other side's close promptly it could cause unnecessary resource
1895   drain on the network.
1896</t>
1897<t>
1898   A client, server, or proxy MAY close the transport connection at any
1899   time. For example, a client might have started to send a new request
1900   at the same time that the server has decided to close the "idle"
1901   connection. From the server's point of view, the connection is being
1902   closed while it was idle, but from the client's point of view, a
1903   request is in progress.
1904</t>
1905<t>
1906   This means that clients, servers, and proxies MUST be able to recover
1907   from asynchronous close events. Client software SHOULD reopen the
1908   transport connection and retransmit the aborted sequence of requests
1909   without user interaction so long as the request sequence is
1910   idempotent (see Section 8.1.2 of <xref target="Part2"/>). Non-idempotent methods or sequences
1911   MUST NOT be automatically retried, although user agents MAY offer a
1912   human operator the choice of retrying the request(s). Confirmation by
1913   user-agent software with semantic understanding of the application
1914   MAY substitute for user confirmation. The automatic retry SHOULD NOT
1915   be repeated if the second sequence of requests fails.
1916</t>
1917<t>
1918   Servers SHOULD always respond to at least one request per connection,
1919   if at all possible. Servers SHOULD NOT  close a connection in the
1920   middle of transmitting a response, unless a network or client failure
1921   is suspected.
1922</t>
1923<t>
1924   Clients that use persistent connections SHOULD limit the number of
1925   simultaneous connections that they maintain to a given server. A
1926   single-user client SHOULD NOT maintain more than 2 connections with
1927   any server or proxy. A proxy SHOULD use up to 2*N connections to
1928   another server or proxy, where N is the number of simultaneously
1929   active users. These guidelines are intended to improve HTTP response
1930   times and avoid congestion.
1931</t>
1932</section>
1933</section>
1934
1935<section title="Message Transmission Requirements" anchor="message.transmission.requirements">
1936
1937<section title="Persistent Connections and Flow Control" anchor="persistent.flow">
1938<t>
1939   HTTP/1.1 servers SHOULD maintain persistent connections and use TCP's
1940   flow control mechanisms to resolve temporary overloads, rather than
1941   terminating connections with the expectation that clients will retry.
1942   The latter technique can exacerbate network congestion.
1943</t>
1944</section>
1945
1946<section title="Monitoring Connections for Error Status Messages" anchor="persistent.monitor">
1947<t>
1948   An HTTP/1.1 (or later) client sending a message-body SHOULD monitor
1949   the network connection for an error status while it is transmitting
1950   the request. If the client sees an error status, it SHOULD
1951   immediately cease transmitting the body. If the body is being sent
1952   using a "chunked" encoding (<xref target="transfer.codings"/>), a zero length chunk and
1953   empty trailer MAY be used to prematurely mark the end of the message.
1954   If the body was preceded by a Content-Length header, the client MUST
1955   close the connection.
1956</t>
1957</section>
1958
1959<section title="Use of the 100 (Continue) Status" anchor="use.of.the.100.status">
1960<t>
1961   The purpose of the 100 (Continue) status (see Section 9.1.1 of <xref target="Part2"/>) is to
1962   allow a client that is sending a request message with a request body
1963   to determine if the origin server is willing to accept the request
1964   (based on the request headers) before the client sends the request
1965   body. In some cases, it might either be inappropriate or highly
1966   inefficient for the client to send the body if the server will reject
1967   the message without looking at the body.
1968</t>
1969<t>
1970   Requirements for HTTP/1.1 clients:
1971  <list style="symbols">
1972    <t>
1973        If a client will wait for a 100 (Continue) response before
1974        sending the request body, it MUST send an Expect request-header
1975        field (Section 10.2 of <xref target="Part2"/>) with the "100-continue" expectation.
1976    </t>
1977    <t>
1978        A client MUST NOT send an Expect request-header field (Section 10.2 of <xref target="Part2"/>)
1979        with the "100-continue" expectation if it does not intend
1980        to send a request body.
1981    </t>
1982  </list>
1983</t>
1984<t>
1985   Because of the presence of older implementations, the protocol allows
1986   ambiguous situations in which a client may send "Expect: 100-continue"
1987   without receiving either a 417 (Expectation Failed) status
1988   or a 100 (Continue) status. Therefore, when a client sends this
1989   header field to an origin server (possibly via a proxy) from which it
1990   has never seen a 100 (Continue) status, the client SHOULD NOT  wait
1991   for an indefinite period before sending the request body.
1992</t>
1993<t>
1994   Requirements for HTTP/1.1 origin servers:
1995  <list style="symbols">
1996    <t> Upon receiving a request which includes an Expect request-header
1997        field with the "100-continue" expectation, an origin server MUST
1998        either respond with 100 (Continue) status and continue to read
1999        from the input stream, or respond with a final status code. The
2000        origin server MUST NOT wait for the request body before sending
2001        the 100 (Continue) response. If it responds with a final status
2002        code, it MAY close the transport connection or it MAY continue
2003        to read and discard the rest of the request.  It MUST NOT
2004        perform the requested method if it returns a final status code.
2005    </t>
2006    <t> An origin server SHOULD NOT  send a 100 (Continue) response if
2007        the request message does not include an Expect request-header
2008        field with the "100-continue" expectation, and MUST NOT send a
2009        100 (Continue) response if such a request comes from an HTTP/1.0
2010        (or earlier) client. There is an exception to this rule: for
2011        compatibility with RFC 2068, a server MAY send a 100 (Continue)
2012        status in response to an HTTP/1.1 PUT or POST request that does
2013        not include an Expect request-header field with the "100-continue"
2014        expectation. This exception, the purpose of which is
2015        to minimize any client processing delays associated with an
2016        undeclared wait for 100 (Continue) status, applies only to
2017        HTTP/1.1 requests, and not to requests with any other HTTP-version
2018        value.
2019    </t>
2020    <t> An origin server MAY omit a 100 (Continue) response if it has
2021        already received some or all of the request body for the
2022        corresponding request.
2023    </t>
2024    <t> An origin server that sends a 100 (Continue) response MUST
2025    ultimately send a final status code, once the request body is
2026        received and processed, unless it terminates the transport
2027        connection prematurely.
2028    </t>
2029    <t> If an origin server receives a request that does not include an
2030        Expect request-header field with the "100-continue" expectation,
2031        the request includes a request body, and the server responds
2032        with a final status code before reading the entire request body
2033        from the transport connection, then the server SHOULD NOT  close
2034        the transport connection until it has read the entire request,
2035        or until the client closes the connection. Otherwise, the client
2036        might not reliably receive the response message. However, this
2037        requirement is not be construed as preventing a server from
2038        defending itself against denial-of-service attacks, or from
2039        badly broken client implementations.
2040      </t>
2041    </list>
2042</t>
2043<t>
2044   Requirements for HTTP/1.1 proxies:
2045  <list style="symbols">
2046    <t> If a proxy receives a request that includes an Expect request-header
2047        field with the "100-continue" expectation, and the proxy
2048        either knows that the next-hop server complies with HTTP/1.1 or
2049        higher, or does not know the HTTP version of the next-hop
2050        server, it MUST forward the request, including the Expect header
2051        field.
2052    </t>
2053    <t> If the proxy knows that the version of the next-hop server is
2054        HTTP/1.0 or lower, it MUST NOT forward the request, and it MUST
2055        respond with a 417 (Expectation Failed) status.
2056    </t>
2057    <t> Proxies SHOULD maintain a cache recording the HTTP version
2058        numbers received from recently-referenced next-hop servers.
2059    </t>
2060    <t> A proxy MUST NOT forward a 100 (Continue) response if the
2061        request message was received from an HTTP/1.0 (or earlier)
2062        client and did not include an Expect request-header field with
2063        the "100-continue" expectation. This requirement overrides the
2064        general rule for forwarding of 1xx responses (see Section 9.1 of <xref target="Part2"/>).
2065    </t>
2066  </list>
2067</t>
2068</section>
2069
2070<section title="Client Behavior if Server Prematurely Closes Connection" anchor="connection.premature">
2071<t>
2072   If an HTTP/1.1 client sends a request which includes a request body,
2073   but which does not include an Expect request-header field with the
2074   "100-continue" expectation, and if the client is not directly
2075   connected to an HTTP/1.1 origin server, and if the client sees the
2076   connection close before receiving any status from the server, the
2077   client SHOULD retry the request.  If the client does retry this
2078   request, it MAY use the following "binary exponential backoff"
2079   algorithm to be assured of obtaining a reliable response:
2080  <list style="numbers">
2081    <t>
2082      Initiate a new connection to the server
2083    </t>
2084    <t>
2085      Transmit the request-headers
2086    </t>
2087    <t>
2088      Initialize a variable R to the estimated round-trip time to the
2089         server (e.g., based on the time it took to establish the
2090         connection), or to a constant value of 5 seconds if the round-trip
2091         time is not available.
2092    </t>
2093    <t>
2094       Compute T = R * (2**N), where N is the number of previous
2095         retries of this request.
2096    </t>
2097    <t>
2098       Wait either for an error response from the server, or for T
2099         seconds (whichever comes first)
2100    </t>
2101    <t>
2102       If no error response is received, after T seconds transmit the
2103         body of the request.
2104    </t>
2105    <t>
2106       If client sees that the connection is closed prematurely,
2107         repeat from step 1 until the request is accepted, an error
2108         response is received, or the user becomes impatient and
2109         terminates the retry process.
2110    </t>
2111  </list>
2112</t>
2113<t>
2114   If at any point an error status is received, the client
2115  <list style="symbols">
2116      <t>SHOULD NOT  continue and</t>
2117
2118      <t>SHOULD close the connection if it has not completed sending the
2119        request message.</t>
2120    </list>
2121</t>
2122</section>
2123</section>
2124</section>
2125
2126
2127<section title="Header Field Definitions" anchor="header.fields">
2128<t>
2129   This section defines the syntax and semantics of all standard
2130   HTTP/1.1 header fields. For entity-header fields, both sender and
2131   recipient refer to either the client or the server, depending on who
2132   sends and who receives the entity.
2133</t>
2134
2135<section title="Connection" anchor="header.connection">
2136  <iref primary="true" item="Connection header"/>
2137  <iref primary="true" item="Headers" subitem="Connection"/>
2138<t>
2139   The Connection general-header field allows the sender to specify
2140   options that are desired for that particular connection and MUST NOT
2141   be communicated by proxies over further connections.
2142</t>
2143<t>
2144   The Connection header has the following grammar:
2145</t>
2146<figure><iref primary="true" item="Grammar" subitem="Connection"/><iref primary="true" item="Grammar" subitem="connection-token"/><artwork type="abnf2616"><![CDATA[
2147    Connection = "Connection" ":" 1#(connection-token)
2148    connection-token  = token
2149]]></artwork></figure>
2150<t>
2151   HTTP/1.1 proxies MUST parse the Connection header field before a
2152   message is forwarded and, for each connection-token in this field,
2153   remove any header field(s) from the message with the same name as the
2154   connection-token. Connection options are signaled by the presence of
2155   a connection-token in the Connection header field, not by any
2156   corresponding additional header field(s), since the additional header
2157   field may not be sent if there are no parameters associated with that
2158   connection option.
2159</t>
2160<t>
2161   Message headers listed in the Connection header MUST NOT include
2162   end-to-end headers, such as Cache-Control.
2163</t>
2164<t>
2165   HTTP/1.1 defines the "close" connection option for the sender to
2166   signal that the connection will be closed after completion of the
2167   response. For example,
2168</t>
2169<figure><artwork type="example"><![CDATA[
2170    Connection: close
2171]]></artwork></figure>
2172<t>
2173   in either the request or the response header fields indicates that
2174   the connection SHOULD NOT  be considered `persistent' (<xref target="persistent.connections"/>)
2175   after the current request/response is complete.
2176</t>
2177<t>
2178   HTTP/1.1 applications that do not support persistent connections MUST
2179   include the "close" connection option in every message.
2180</t>
2181<t>
2182   A system receiving an HTTP/1.0 (or lower-version) message that
2183   includes a Connection header MUST, for each connection-token in this
2184   field, remove and ignore any header field(s) from the message with
2185   the same name as the connection-token. This protects against mistaken
2186   forwarding of such header fields by pre-HTTP/1.1 proxies. See <xref target="compatibility.with.http.1.0.persistent.connections"/>.
2187</t>
2188</section>
2189
2190<section title="Content-Length" anchor="header.content-length">
2191  <iref primary="true" item="Content-Length header"/>
2192  <iref primary="true" item="Headers" subitem="Content-Length"/>
2193<t>
2194   The Content-Length entity-header field indicates the size of the
2195   entity-body, in decimal number of OCTETs, sent to the recipient or,
2196   in the case of the HEAD method, the size of the entity-body that
2197   would have been sent had the request been a GET.
2198</t>
2199<figure><iref primary="true" item="Grammar" subitem="Content-Length"/><artwork type="abnf2616"><![CDATA[
2200    Content-Length    = "Content-Length" ":" 1*DIGIT
2201]]></artwork></figure>
2202<t>
2203   An example is
2204</t>
2205<figure><artwork type="example"><![CDATA[
2206    Content-Length: 3495
2207]]></artwork></figure>
2208<t>
2209   Applications SHOULD use this field to indicate the transfer-length of
2210   the message-body, unless this is prohibited by the rules in <xref target="message.length"/>.
2211</t>
2212<t>
2213   Any Content-Length greater than or equal to zero is a valid value.
2214   <xref target="message.length"/> describes how to determine the length of a message-body
2215   if a Content-Length is not given.
2216</t>
2217<t>
2218   Note that the meaning of this field is significantly different from
2219   the corresponding definition in MIME, where it is an optional field
2220   used within the "message/external-body" content-type. In HTTP, it
2221   SHOULD be sent whenever the message's length can be determined prior
2222   to being transferred, unless this is prohibited by the rules in
2223   <xref target="message.length"/>.
2224</t>
2225</section>
2226
2227<section title="Date" anchor="header.date">
2228  <iref primary="true" item="Date header"/>
2229  <iref primary="true" item="Headers" subitem="Date"/>
2230<t>
2231   The Date general-header field represents the date and time at which
2232   the message was originated, having the same semantics as orig-date in
2233   RFC 822. The field value is an HTTP-date, as described in <xref target="full.date"/>;
2234   it MUST be sent in RFC 1123 <xref target="RFC1123"/>-date format.
2235</t>
2236<figure><iref primary="true" item="Grammar" subitem="Date"/><artwork type="abnf2616"><![CDATA[
2237    Date  = "Date" ":" HTTP-date
2238]]></artwork></figure>
2239<t>
2240   An example is
2241</t>
2242<figure><artwork type="example"><![CDATA[
2243    Date: Tue, 15 Nov 1994 08:12:31 GMT
2244]]></artwork></figure>
2245<t>
2246   Origin servers MUST include a Date header field in all responses,
2247   except in these cases:
2248  <list style="numbers">
2249      <t>If the response status code is 100 (Continue) or 101 (Switching
2250         Protocols), the response MAY include a Date header field, at
2251         the server's option.</t>
2252
2253      <t>If the response status code conveys a server error, e.g. 500
2254         (Internal Server Error) or 503 (Service Unavailable), and it is
2255         inconvenient or impossible to generate a valid Date.</t>
2256
2257      <t>If the server does not have a clock that can provide a
2258         reasonable approximation of the current time, its responses
2259         MUST NOT include a Date header field. In this case, the rules
2260         in <xref target="clockless.origin.server.operation"/> MUST be followed.</t>
2261  </list>
2262</t>
2263<t>
2264   A received message that does not have a Date header field MUST be
2265   assigned one by the recipient if the message will be cached by that
2266   recipient or gatewayed via a protocol which requires a Date. An HTTP
2267   implementation without a clock MUST NOT cache responses without
2268   revalidating them on every use. An HTTP cache, especially a shared
2269   cache, SHOULD use a mechanism, such as NTP <xref target="RFC1305"/>, to synchronize its
2270   clock with a reliable external standard.
2271</t>
2272<t>
2273   Clients SHOULD only send a Date header field in messages that include
2274   an entity-body, as in the case of the PUT and POST requests, and even
2275   then it is optional. A client without a clock MUST NOT send a Date
2276   header field in a request.
2277</t>
2278<t>
2279   The HTTP-date sent in a Date header SHOULD NOT  represent a date and
2280   time subsequent to the generation of the message. It SHOULD represent
2281   the best available approximation of the date and time of message
2282   generation, unless the implementation has no means of generating a
2283   reasonably accurate date and time. In theory, the date ought to
2284   represent the moment just before the entity is generated. In
2285   practice, the date can be generated at any time during the message
2286   origination without affecting its semantic value.
2287</t>
2288
2289<section title="Clockless Origin Server Operation" anchor="clockless.origin.server.operation">
2290<t>
2291   Some origin server implementations might not have a clock available.
2292   An origin server without a clock MUST NOT assign Expires or Last-Modified
2293   values to a response, unless these values were associated
2294   with the resource by a system or user with a reliable clock. It MAY
2295   assign an Expires value that is known, at or before server
2296   configuration time, to be in the past (this allows "pre-expiration"
2297   of responses without storing separate Expires values for each
2298   resource).
2299</t>
2300</section>
2301</section>
2302
2303<section title="Host" anchor="header.host">
2304  <iref primary="true" item="Host header"/>
2305  <iref primary="true" item="Headers" subitem="Host"/>
2306<t>
2307   The Host request-header field specifies the Internet host and port
2308   number of the resource being requested, as obtained from the original
2309   URI given by the user or referring resource (generally an HTTP URL,
2310   as described in <xref target="http.url"/>). The Host field value MUST represent
2311   the naming authority of the origin server or gateway given by the
2312   original URL. This allows the origin server or gateway to
2313   differentiate between internally-ambiguous URLs, such as the root "/"
2314   URL of a server for multiple host names on a single IP address.
2315</t>
2316<figure><iref primary="true" item="Grammar" subitem="Host"/><artwork type="abnf2616"><![CDATA[
2317    Host = "Host" ":" host [ ":" port ] ; Section 3.2.2
2318]]></artwork></figure>
2319<t>
2320   A "host" without any trailing port information implies the default
2321   port for the service requested (e.g., "80" for an HTTP URL). For
2322   example, a request on the origin server for
2323   &lt;http://www.w3.org/pub/WWW/&gt; would properly include:
2324</t>
2325<figure><artwork type="example"><![CDATA[
2326    GET /pub/WWW/ HTTP/1.1
2327    Host: www.w3.org
2328]]></artwork></figure>
2329<t>
2330   A client MUST include a Host header field in all HTTP/1.1 request
2331   messages . If the requested URI does not include an Internet host
2332   name for the service being requested, then the Host header field MUST
2333   be given with an empty value. An HTTP/1.1 proxy MUST ensure that any
2334   request message it forwards does contain an appropriate Host header
2335   field that identifies the service being requested by the proxy. All
2336   Internet-based HTTP/1.1 servers MUST respond with a 400 (Bad Request)
2337   status code to any HTTP/1.1 request message which lacks a Host header
2338   field.
2339</t>
2340<t>
2341   See sections <xref target="the.resource.identified.by.a.request" format="counter"/>
2342   and <xref target="changes.to.simplify.multi-homed.web.servers.and.conserve.ip.addresses" format="counter"/>
2343   for other requirements relating to Host.
2344</t>
2345</section>
2346
2347<section title="TE" anchor="header.te">
2348  <iref primary="true" item="TE header"/>
2349  <iref primary="true" item="Headers" subitem="TE"/>
2350<t>
2351   The TE request-header field indicates what extension transfer-codings
2352   it is willing to accept in the response and whether or not it is
2353   willing to accept trailer fields in a chunked transfer-coding. Its
2354   value may consist of the keyword "trailers" and/or a comma-separated
2355   list of extension transfer-coding names with optional accept
2356   parameters (as described in <xref target="transfer.codings"/>).
2357</t>
2358<figure><iref primary="true" item="Grammar" subitem="TE"/><iref primary="true" item="Grammar" subitem="t-codings"/><artwork type="abnf2616"><![CDATA[
2359    TE        = "TE" ":" #( t-codings )
2360    t-codings = "trailers" | ( transfer-extension [ accept-params ] )
2361]]></artwork></figure>
2362<t>
2363   The presence of the keyword "trailers" indicates that the client is
2364   willing to accept trailer fields in a chunked transfer-coding, as
2365   defined in <xref target="chunked.transfer.encoding"/>. This keyword is reserved for use with
2366   transfer-coding values even though it does not itself represent a
2367   transfer-coding.
2368</t>
2369<t>
2370   Examples of its use are:
2371</t>
2372<figure><artwork type="example"><![CDATA[
2373    TE: deflate
2374    TE:
2375    TE: trailers, deflate;q=0.5
2376]]></artwork></figure>
2377<t>
2378   The TE header field only applies to the immediate connection.
2379   Therefore, the keyword MUST be supplied within a Connection header
2380   field (<xref target="header.connection"/>) whenever TE is present in an HTTP/1.1 message.
2381</t>
2382<t>
2383   A server tests whether a transfer-coding is acceptable, according to
2384   a TE field, using these rules:
2385  <list style="numbers">
2386    <t>The "chunked" transfer-coding is always acceptable. If the
2387         keyword "trailers" is listed, the client indicates that it is
2388         willing to accept trailer fields in the chunked response on
2389         behalf of itself and any downstream clients. The implication is
2390         that, if given, the client is stating that either all
2391         downstream clients are willing to accept trailer fields in the
2392         forwarded response, or that it will attempt to buffer the
2393         response on behalf of downstream recipients.
2394      <vspace blankLines="1"/>
2395         Note: HTTP/1.1 does not define any means to limit the size of a
2396         chunked response such that a client can be assured of buffering
2397         the entire response.</t>
2398    <t>If the transfer-coding being tested is one of the transfer-codings
2399         listed in the TE field, then it is acceptable unless it
2400         is accompanied by a qvalue of 0. (As defined in Section 2.4 of <xref target="Part3"/>, a
2401         qvalue of 0 means "not acceptable.")</t>
2402    <t>If multiple transfer-codings are acceptable, then the
2403         acceptable transfer-coding with the highest non-zero qvalue is
2404         preferred.  The "chunked" transfer-coding always has a qvalue
2405         of 1.</t>
2406  </list>
2407</t>
2408<t>
2409   If the TE field-value is empty or if no TE field is present, the only
2410   transfer-coding  is "chunked". A message with no transfer-coding is
2411   always acceptable.
2412</t>
2413</section>
2414
2415<section title="Trailer" anchor="header.trailer">
2416  <iref primary="true" item="Trailer header"/>
2417  <iref primary="true" item="Headers" subitem="Trailer"/>
2418<t>
2419   The Trailer general field value indicates that the given set of
2420   header fields is present in the trailer of a message encoded with
2421   chunked transfer-coding.
2422</t>
2423<figure><iref primary="true" item="Grammar" subitem="Trailer"/><artwork type="abnf2616"><![CDATA[
2424    Trailer  = "Trailer" ":" 1#field-name
2425]]></artwork></figure>
2426<t>
2427   An HTTP/1.1 message SHOULD include a Trailer header field in a
2428   message using chunked transfer-coding with a non-empty trailer. Doing
2429   so allows the recipient to know which header fields to expect in the
2430   trailer.
2431</t>
2432<t>
2433   If no Trailer header field is present, the trailer SHOULD NOT  include
2434   any header fields. See <xref target="chunked.transfer.encoding"/> for restrictions on the use of
2435   trailer fields in a "chunked" transfer-coding.
2436</t>
2437<t>
2438   Message header fields listed in the Trailer header field MUST NOT
2439   include the following header fields:
2440  <list style="symbols">
2441    <t>Transfer-Encoding</t>
2442    <t>Content-Length</t>
2443    <t>Trailer</t>
2444  </list>
2445</t>
2446</section>
2447
2448<section title="Transfer-Encoding" anchor="header.transfer-encoding">
2449  <iref primary="true" item="Transfer-Encoding header"/>
2450  <iref primary="true" item="Headers" subitem="Transfer-Encoding"/>
2451<t>
2452   The Transfer-Encoding general-header field indicates what (if any)
2453   type of transformation has been applied to the message body in order
2454   to safely transfer it between the sender and the recipient. This
2455   differs from the content-coding in that the transfer-coding is a
2456   property of the message, not of the entity.
2457</t>
2458<figure><iref primary="true" item="Grammar" subitem="Transfer-Encoding"/><artwork type="abnf2616"><![CDATA[
2459  Transfer-Encoding       = "Transfer-Encoding" ":" 1#transfer-coding
2460]]></artwork></figure>
2461<t>
2462   Transfer-codings are defined in <xref target="transfer.codings"/>. An example is:
2463</t>
2464<figure><artwork type="example"><![CDATA[
2465  Transfer-Encoding: chunked
2466]]></artwork></figure>
2467<t>
2468   If multiple encodings have been applied to an entity, the transfer-codings
2469   MUST be listed in the order in which they were applied.
2470   Additional information about the encoding parameters MAY be provided
2471   by other entity-header fields not defined by this specification.
2472</t>
2473<t>
2474   Many older HTTP/1.0 applications do not understand the Transfer-Encoding
2475   header.
2476</t>
2477</section>
2478
2479<section title="Upgrade" anchor="header.upgrade">
2480  <iref primary="true" item="Upgrade header"/>
2481  <iref primary="true" item="Headers" subitem="Upgrade"/>
2482<t>
2483   The Upgrade general-header allows the client to specify what
2484   additional communication protocols it supports and would like to use
2485   if the server finds it appropriate to switch protocols. The server
2486   MUST use the Upgrade header field within a 101 (Switching Protocols)
2487   response to indicate which protocol(s) are being switched.
2488</t>
2489<figure><iref primary="true" item="Grammar" subitem="Upgrade"/><artwork type="abnf2616"><![CDATA[
2490    Upgrade        = "Upgrade" ":" 1#product
2491]]></artwork></figure>
2492<t>
2493   For example,
2494</t>
2495<figure><artwork type="example"><![CDATA[
2496    Upgrade: HTTP/2.0, SHTTP/1.3, IRC/6.9, RTA/x11
2497]]></artwork></figure>
2498<t>
2499   The Upgrade header field is intended to provide a simple mechanism
2500   for transition from HTTP/1.1 to some other, incompatible protocol. It
2501   does so by allowing the client to advertise its desire to use another
2502   protocol, such as a later version of HTTP with a higher major version
2503   number, even though the current request has been made using HTTP/1.1.
2504   This eases the difficult transition between incompatible protocols by
2505   allowing the client to initiate a request in the more commonly
2506   supported protocol while indicating to the server that it would like
2507   to use a "better" protocol if available (where "better" is determined
2508   by the server, possibly according to the nature of the method and/or
2509   resource being requested).
2510</t>
2511<t>
2512   The Upgrade header field only applies to switching application-layer
2513   protocols upon the existing transport-layer connection. Upgrade
2514   cannot be used to insist on a protocol change; its acceptance and use
2515   by the server is optional. The capabilities and nature of the
2516   application-layer communication after the protocol change is entirely
2517   dependent upon the new protocol chosen, although the first action
2518   after changing the protocol MUST be a response to the initial HTTP
2519   request containing the Upgrade header field.
2520</t>
2521<t>
2522   The Upgrade header field only applies to the immediate connection.
2523   Therefore, the upgrade keyword MUST be supplied within a Connection
2524   header field (<xref target="header.connection"/>) whenever Upgrade is present in an
2525   HTTP/1.1 message.
2526</t>
2527<t>
2528   The Upgrade header field cannot be used to indicate a switch to a
2529   protocol on a different connection. For that purpose, it is more
2530   appropriate to use a 301, 302, 303, or 305 redirection response.
2531</t>
2532<t>
2533   This specification only defines the protocol name "HTTP" for use by
2534   the family of Hypertext Transfer Protocols, as defined by the HTTP
2535   version rules of <xref target="http.version"/> and future updates to this
2536   specification. Any token can be used as a protocol name; however, it
2537   will only be useful if both the client and server associate the name
2538   with the same protocol.
2539</t>
2540</section>
2541
2542<section title="Via" anchor="header.via">
2543  <iref primary="true" item="Via header"/>
2544  <iref primary="true" item="Headers" subitem="Via"/>
2545<t>
2546   The Via general-header field MUST be used by gateways and proxies to
2547   indicate the intermediate protocols and recipients between the user
2548   agent and the server on requests, and between the origin server and
2549   the client on responses. It is analogous to the "Received" field of
2550   RFC 822 <xref target="RFC822"/> and is intended to be used for tracking message forwards,
2551   avoiding request loops, and identifying the protocol capabilities of
2552   all senders along the request/response chain.
2553</t>
2554<figure><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"/><artwork type="abnf2616"><![CDATA[
2555   Via =  "Via" ":" 1#( received-protocol received-by [ comment ] )
2556   received-protocol = [ protocol-name "/" ] protocol-version
2557   protocol-name     = token
2558   protocol-version  = token
2559   received-by       = ( host [ ":" port ] ) | pseudonym
2560   pseudonym         = token
2561]]></artwork></figure>
2562<t>
2563   The received-protocol indicates the protocol version of the message
2564   received by the server or client along each segment of the
2565   request/response chain. The received-protocol version is appended to
2566   the Via field value when the message is forwarded so that information
2567   about the protocol capabilities of upstream applications remains
2568   visible to all recipients.
2569</t>
2570<t>
2571   The protocol-name is optional if and only if it would be "HTTP". The
2572   received-by field is normally the host and optional port number of a
2573   recipient server or client that subsequently forwarded the message.
2574   However, if the real host is considered to be sensitive information,
2575   it MAY be replaced by a pseudonym. If the port is not given, it MAY
2576   be assumed to be the default port of the received-protocol.
2577</t>
2578<t>
2579   Multiple Via field values represents each proxy or gateway that has
2580   forwarded the message. Each recipient MUST append its information
2581   such that the end result is ordered according to the sequence of
2582   forwarding applications.
2583</t>
2584<t>
2585   Comments MAY be used in the Via header field to identify the software
2586   of the recipient proxy or gateway, analogous to the User-Agent and
2587   Server header fields. However, all comments in the Via field are
2588   optional and MAY be removed by any recipient prior to forwarding the
2589   message.
2590</t>
2591<t>
2592   For example, a request message could be sent from an HTTP/1.0 user
2593   agent to an internal proxy code-named "fred", which uses HTTP/1.1 to
2594   forward the request to a public proxy at nowhere.com, which completes
2595   the request by forwarding it to the origin server at www.ics.uci.edu.
2596   The request received by www.ics.uci.edu would then have the following
2597   Via header field:
2598</t>
2599<figure><artwork type="example"><![CDATA[
2600    Via: 1.0 fred, 1.1 nowhere.com (Apache/1.1)
2601]]></artwork></figure>
2602<t>
2603   Proxies and gateways used as a portal through a network firewall
2604   SHOULD NOT, by default, forward the names and ports of hosts within
2605   the firewall region. This information SHOULD only be propagated if
2606   explicitly enabled. If not enabled, the received-by host of any host
2607   behind the firewall SHOULD be replaced by an appropriate pseudonym
2608   for that host.
2609</t>
2610<t>
2611   For organizations that have strong privacy requirements for hiding
2612   internal structures, a proxy MAY combine an ordered subsequence of
2613   Via header field entries with identical received-protocol values into
2614   a single such entry. For example,
2615</t>
2616<figure><artwork type="example"><![CDATA[
2617    Via: 1.0 ricky, 1.1 ethel, 1.1 fred, 1.0 lucy
2618]]></artwork></figure>
2619<t>
2620        could be collapsed to
2621</t>
2622<figure><artwork type="example"><![CDATA[
2623    Via: 1.0 ricky, 1.1 mertz, 1.0 lucy
2624]]></artwork></figure>
2625<t>
2626   Applications SHOULD NOT  combine multiple entries unless they are all
2627   under the same organizational control and the hosts have already been
2628   replaced by pseudonyms. Applications MUST NOT combine entries which
2629   have different received-protocol values.
2630</t>
2631</section>
2632
2633</section>
2634
2635<section title="IANA Considerations" anchor="IANA.considerations">
2636<t>
2637   TBD.
2638</t>
2639</section>
2640
2641<section title="Security Considerations" anchor="security.considerations">
2642<t>
2643   This section is meant to inform application developers, information
2644   providers, and users of the security limitations in HTTP/1.1 as
2645   described by this document. The discussion does not include
2646   definitive solutions to the problems revealed, though it does make
2647   some suggestions for reducing security risks.
2648</t>
2649
2650<section title="Personal Information" anchor="personal.information">
2651<t>
2652   HTTP clients are often privy to large amounts of personal information
2653   (e.g. the user's name, location, mail address, passwords, encryption
2654   keys, etc.), and SHOULD be very careful to prevent unintentional
2655   leakage of this information via the HTTP protocol to other sources.
2656   We very strongly recommend that a convenient interface be provided
2657   for the user to control dissemination of such information, and that
2658   designers and implementors be particularly careful in this area.
2659   History shows that errors in this area often create serious security
2660   and/or privacy problems and generate highly adverse publicity for the
2661   implementor's company.
2662</t>
2663</section>
2664
2665<section title="Abuse of Server Log Information" anchor="abuse.of.server.log.information">
2666<t>
2667   A server is in the position to save personal data about a user's
2668   requests which might identify their reading patterns or subjects of
2669   interest. This information is clearly confidential in nature and its
2670   handling can be constrained by law in certain countries. People using
2671   the HTTP protocol to provide data are responsible for ensuring that
2672   such material is not distributed without the permission of any
2673   individuals that are identifiable by the published results.
2674</t>
2675</section>
2676
2677<section title="Attacks Based On File and Path Names" anchor="attack.pathname">
2678<t>
2679   Implementations of HTTP origin servers SHOULD be careful to restrict
2680   the documents returned by HTTP requests to be only those that were
2681   intended by the server administrators. If an HTTP server translates
2682   HTTP URIs directly into file system calls, the server MUST take
2683   special care not to serve files that were not intended to be
2684   delivered to HTTP clients. For example, UNIX, Microsoft Windows, and
2685   other operating systems use ".." as a path component to indicate a
2686   directory level above the current one. On such a system, an HTTP
2687   server MUST disallow any such construct in the Request-URI if it
2688   would otherwise allow access to a resource outside those intended to
2689   be accessible via the HTTP server. Similarly, files intended for
2690   reference only internally to the server (such as access control
2691   files, configuration files, and script code) MUST be protected from
2692   inappropriate retrieval, since they might contain sensitive
2693   information. Experience has shown that minor bugs in such HTTP server
2694   implementations have turned into security risks.
2695</t>
2696</section>
2697
2698<section title="DNS Spoofing" anchor="dns.spoofing">
2699<t>
2700   Clients using HTTP rely heavily on the Domain Name Service, and are
2701   thus generally prone to security attacks based on the deliberate
2702   mis-association of IP addresses and DNS names. Clients need to be
2703   cautious in assuming the continuing validity of an IP number/DNS name
2704   association.
2705</t>
2706<t>
2707   In particular, HTTP clients SHOULD rely on their name resolver for
2708   confirmation of an IP number/DNS name association, rather than
2709   caching the result of previous host name lookups. Many platforms
2710   already can cache host name lookups locally when appropriate, and
2711   they SHOULD be configured to do so. It is proper for these lookups to
2712   be cached, however, only when the TTL (Time To Live) information
2713   reported by the name server makes it likely that the cached
2714   information will remain useful.
2715</t>
2716<t>
2717   If HTTP clients cache the results of host name lookups in order to
2718   achieve a performance improvement, they MUST observe the TTL
2719   information reported by DNS.
2720</t>
2721<t>
2722   If HTTP clients do not observe this rule, they could be spoofed when
2723   a previously-accessed server's IP address changes. As network
2724   renumbering is expected to become increasingly common <xref target="RFC1900"/>, the
2725   possibility of this form of attack will grow. Observing this
2726   requirement thus reduces this potential security vulnerability.
2727</t>
2728<t>
2729   This requirement also improves the load-balancing behavior of clients
2730   for replicated servers using the same DNS name and reduces the
2731   likelihood of a user's experiencing failure in accessing sites which
2732   use that strategy.
2733</t>
2734</section>
2735
2736<section title="Proxies and Caching" anchor="attack.proxies">
2737<t>
2738   By their very nature, HTTP proxies are men-in-the-middle, and
2739   represent an opportunity for man-in-the-middle attacks. Compromise of
2740   the systems on which the proxies run can result in serious security
2741   and privacy problems. Proxies have access to security-related
2742   information, personal information about individual users and
2743   organizations, and proprietary information belonging to users and
2744   content providers. A compromised proxy, or a proxy implemented or
2745   configured without regard to security and privacy considerations,
2746   might be used in the commission of a wide range of potential attacks.
2747</t>
2748<t>
2749   Proxy operators should protect the systems on which proxies run as
2750   they would protect any system that contains or transports sensitive
2751   information. In particular, log information gathered at proxies often
2752   contains highly sensitive personal information, and/or information
2753   about organizations. Log information should be carefully guarded, and
2754   appropriate guidelines for use developed and followed. (<xref target="abuse.of.server.log.information"/>).
2755</t>
2756<t>
2757   Proxy implementors should consider the privacy and security
2758   implications of their design and coding decisions, and of the
2759   configuration options they provide to proxy operators (especially the
2760   default configuration).
2761</t>
2762<t>
2763   Users of a proxy need to be aware that they are no trustworthier than
2764   the people who run the proxy; HTTP itself cannot solve this problem.
2765</t>
2766<t>
2767   The judicious use of cryptography, when appropriate, may suffice to
2768   protect against a broad range of security and privacy attacks. Such
2769   cryptography is beyond the scope of the HTTP/1.1 specification.
2770</t>
2771</section>
2772
2773<section title="Denial of Service Attacks on Proxies" anchor="attack.DoS">
2774<t>
2775   They exist. They are hard to defend against. Research continues.
2776   Beware.
2777</t>
2778</section>
2779</section>
2780
2781<section title="Acknowledgments" anchor="ack">
2782<t>
2783   This specification makes heavy use of the augmented BNF and generic
2784   constructs defined by David H. Crocker for RFC 822 <xref target="RFC822"/>. Similarly, it
2785   reuses many of the definitions provided by Nathaniel Borenstein and
2786   Ned Freed for MIME <xref target="RFC2045"/>. We hope that their inclusion in this
2787   specification will help reduce past confusion over the relationship
2788   between HTTP and Internet mail message formats.
2789</t>
2790<t>
2791   The HTTP protocol has evolved considerably over the years. It has
2792   benefited from a large and active developer community--the many
2793   people who have participated on the www-talk mailing list--and it is
2794   that community which has been most responsible for the success of
2795   HTTP and of the World-Wide Web in general. Marc Andreessen, Robert
2796   Cailliau, Daniel W. Connolly, Bob Denny, John Franks, Jean-Francois
2797   Groff, Phillip M. Hallam-Baker, Hakon W. Lie, Ari Luotonen, Rob
2798   McCool, Lou Montulli, Dave Raggett, Tony Sanders, and Marc
2799   VanHeyningen deserve special recognition for their efforts in
2800   defining early aspects of the protocol.
2801</t>
2802<t>
2803   This document has benefited greatly from the comments of all those
2804   participating in the HTTP-WG. In addition to those already mentioned,
2805   the following individuals have contributed to this specification:
2806</t>
2807<figure><artwork type="text/plain"><![CDATA[
2808    Gary Adams                  Ross Patterson
2809    Harald Tveit Alvestrand     Albert Lunde
2810    Keith Ball                  John C. Mallery
2811    Brian Behlendorf            Jean-Philippe Martin-Flatin
2812    Paul Burchard               Mitra
2813    Maurizio Codogno            David Morris
2814    Mike Cowlishaw              Gavin Nicol
2815    Roman Czyborra              Bill Perry
2816    Michael A. Dolan            Jeffrey Perry
2817    David J. Fiander            Scott Powers
2818    Alan Freier                 Owen Rees
2819    Marc Hedlund                Luigi Rizzo
2820    Greg Herlihy                David Robinson
2821    Koen Holtman                Marc Salomon
2822    Alex Hopmann                Rich Salz
2823    Bob Jernigan                Allan M. Schiffman
2824    Shel Kaphan                 Jim Seidman
2825    Rohit Khare                 Chuck Shotton
2826    John Klensin                Eric W. Sink
2827    Martijn Koster              Simon E. Spero
2828    Alexei Kosut                Richard N. Taylor
2829    David M. Kristol            Robert S. Thau
2830    Daniel LaLiberte            Bill (BearHeart) Weinman
2831    Ben Laurie                  Francois Yergeau
2832    Paul J. Leach               Mary Ellen Zurko
2833    Daniel DuBois               Josh Cohen
2834]]></artwork></figure>
2835<t>
2836   Thanks to the "cave men" of Palo Alto. You know who you are.
2837</t>
2838<t>
2839   Jim Gettys (the current editor of this document) wishes particularly
2840   to thank Roy Fielding, the previous editor of this document, along
2841   with John Klensin, Jeff Mogul, Paul Leach, Dave Kristol, Koen
2842   Holtman, John Franks, Josh Cohen, Alex Hopmann, Scott Lawrence, and
2843   Larry Masinter for their help. And thanks go particularly to Jeff
2844   Mogul and Scott Lawrence for performing the "MUST/MAY/SHOULD" audit.
2845</t>
2846<t>
2847   The Apache Group, Anselm Baird-Smith, author of Jigsaw, and Henrik
2848   Frystyk implemented RFC 2068 early, and we wish to thank them for the
2849   discovery of many of the problems that this document attempts to
2850   rectify.
2851</t>
2852<t>
2853   Based on an XML translation of RFC 2616 by Julian Reschke.
2854</t>
2855</section>
2856
2857</middle>
2858<back>
2859<references>
2860
2861<reference anchor="Part2">
2862   <front>
2863      <title abbrev="HTTP/1.1">HTTP/1.1, part 2: Message Semantics</title>
2864      <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
2865         <organization abbrev="Day Software">Day Software</organization>
2866         <address><email>fielding@gbiv.com</email></address>
2867      </author>
2868      <author initials="J." surname="Gettys" fullname="Jim Gettys">
2869         <organization>One Laptop per Child</organization>
2870         <address><email>jg@laptop.org</email></address>
2871      </author>
2872      <author initials="J." surname="Mogul" fullname="Jeffrey C. Mogul">
2873         <organization abbrev="HP">Hewlett-Packard Company</organization>
2874         <address><email>JeffMogul@acm.org</email></address>
2875      </author>
2876      <author initials="H." surname="Frystyk" fullname="Henrik Frystyk Nielsen">
2877         <organization abbrev="Microsoft">Microsoft Corporation</organization>
2878         <address><email>henrikn@microsoft.com</email></address>
2879      </author>
2880      <author initials="L." surname="Masinter" fullname="Larry Masinter">
2881         <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
2882         <address><email>LMM@acm.org</email></address>
2883      </author>
2884      <author initials="P." surname="Leach" fullname="Paul J. Leach">
2885         <organization abbrev="Microsoft">Microsoft Corporation</organization>
2886         <address><email>paulle@microsoft.com</email></address>
2887      </author>
2888      <author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
2889         <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
2890         <address><email>timbl@w3.org</email></address>
2891      </author>
2892      <date month="December" year="2007"/>
2893   </front>
2894   <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p2-semantics-00"/>
2895   
2896</reference>
2897
2898<reference anchor="Part3">
2899   <front>
2900      <title abbrev="HTTP/1.1">HTTP/1.1, part 3: Message Payload and Content Negotiation</title>
2901      <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
2902         <organization abbrev="Day Software">Day Software</organization>
2903         <address><email>fielding@gbiv.com</email></address>
2904      </author>
2905      <author initials="J." surname="Gettys" fullname="Jim Gettys">
2906         <organization>One Laptop per Child</organization>
2907         <address><email>jg@laptop.org</email></address>
2908      </author>
2909      <author initials="J." surname="Mogul" fullname="Jeffrey C. Mogul">
2910         <organization abbrev="HP">Hewlett-Packard Company</organization>
2911         <address><email>JeffMogul@acm.org</email></address>
2912      </author>
2913      <author initials="H." surname="Frystyk" fullname="Henrik Frystyk Nielsen">
2914         <organization abbrev="Microsoft">Microsoft Corporation</organization>
2915         <address><email>henrikn@microsoft.com</email></address>
2916      </author>
2917      <author initials="L." surname="Masinter" fullname="Larry Masinter">
2918         <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
2919         <address><email>LMM@acm.org</email></address>
2920      </author>
2921      <author initials="P." surname="Leach" fullname="Paul J. Leach">
2922         <organization abbrev="Microsoft">Microsoft Corporation</organization>
2923         <address><email>paulle@microsoft.com</email></address>
2924      </author>
2925      <author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
2926         <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
2927         <address><email>timbl@w3.org</email></address>
2928      </author>
2929      <date month="December" year="2007"/>
2930   </front>
2931   <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p3-payload-00"/>
2932   
2933</reference>
2934
2935<reference anchor="Part6">
2936   <front>
2937      <title abbrev="HTTP/1.1">HTTP/1.1, part 6: Caching</title>
2938      <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
2939         <organization abbrev="Day Software">Day Software</organization>
2940         <address><email>fielding@gbiv.com</email></address>
2941      </author>
2942      <author initials="J." surname="Gettys" fullname="Jim Gettys">
2943         <organization>One Laptop per Child</organization>
2944         <address><email>jg@laptop.org</email></address>
2945      </author>
2946      <author initials="J." surname="Mogul" fullname="Jeffrey C. Mogul">
2947         <organization abbrev="HP">Hewlett-Packard Company</organization>
2948         <address><email>JeffMogul@acm.org</email></address>
2949      </author>
2950      <author initials="H." surname="Frystyk" fullname="Henrik Frystyk Nielsen">
2951         <organization abbrev="Microsoft">Microsoft Corporation</organization>
2952         <address><email>henrikn@microsoft.com</email></address>
2953      </author>
2954      <author initials="L." surname="Masinter" fullname="Larry Masinter">
2955         <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
2956         <address><email>LMM@acm.org</email></address>
2957      </author>
2958      <author initials="P." surname="Leach" fullname="Paul J. Leach">
2959         <organization abbrev="Microsoft">Microsoft Corporation</organization>
2960         <address><email>paulle@microsoft.com</email></address>
2961      </author>
2962      <author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
2963         <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
2964         <address><email>timbl@w3.org</email></address>
2965      </author>
2966      <date month="December" year="2007"/>
2967   </front>
2968   <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p6-cache-00"/>
2969   
2970</reference>
2971
2972<reference anchor="RFC2616">
2973   <front>
2974      <title>Hypertext Transfer Protocol -- HTTP/1.1</title>
2975      <author initials="R." surname="Fielding" fullname="R. Fielding">
2976         <organization>University of California, Irvine</organization>
2977         <address><email>fielding@ics.uci.edu</email></address>
2978      </author>
2979      <author initials="J." surname="Gettys" fullname="J. Gettys">
2980         <organization>W3C</organization>
2981         <address><email>jg@w3.org</email></address>
2982      </author>
2983      <author initials="J." surname="Mogul" fullname="J. Mogul">
2984         <organization>Compaq Computer Corporation</organization>
2985         <address><email>mogul@wrl.dec.com</email></address>
2986      </author>
2987      <author initials="H." surname="Frystyk" fullname="H. Frystyk">
2988         <organization>MIT Laboratory for Computer Science</organization>
2989         <address><email>frystyk@w3.org</email></address>
2990      </author>
2991      <author initials="L." surname="Masinter" fullname="L. Masinter">
2992         <organization>Xerox Corporation</organization>
2993         <address><email>masinter@parc.xerox.com</email></address>
2994      </author>
2995      <author initials="P." surname="Leach" fullname="P. Leach">
2996         <organization>Microsoft Corporation</organization>
2997         <address><email>paulle@microsoft.com</email></address>
2998      </author>
2999      <author initials="T." surname="Berners-Lee" fullname="T. Berners-Lee">
3000         <organization>W3C</organization>
3001         <address><email>timbl@w3.org</email></address>
3002      </author>
3003      <date month="June" year="1999"/>
3004   </front>
3005   <seriesInfo name="RFC" value="2616"/>
3006</reference>
3007
3008<reference anchor="RFC1436">
3009<front>
3010<title abbrev="Gopher">The Internet Gopher Protocol (a distributed document search and retrieval protocol)</title>
3011<author initials="F." surname="Anklesaria" fullname="Farhad Anklesaria">
3012<organization>University of Minnesota, Computer and Information Services</organization>
3013<address>
3014<postal>
3015<street>100 Union Street SE</street>
3016<street>Room 152 Shepherd Labs</street>
3017<city>Minneapolis</city>
3018<region>MN</region>
3019<code>55455</code>
3020<country>US</country></postal>
3021<phone>+1 612 625 1300</phone>
3022<email>fxa@boombox.micro.umn.edu</email></address></author>
3023<author initials="M." surname="McCahill" fullname="Mark McCahill">
3024<organization>University of Minnesota, Computer and Information Services</organization>
3025<address>
3026<postal>
3027<street>100 Union Street SE</street>
3028<street>Room 152 Shepherd Labs</street>
3029<city>Minneapolis</city>
3030<region>MN</region>
3031<code>55455</code>
3032<country>US</country></postal>
3033<phone>+1 612 625 1300</phone>
3034<email>mpm@boombox.micro.umn.edu</email></address></author>
3035<author initials="P." surname="Lindner" fullname="Paul Lindner">
3036<organization>University of Minnesota, Computer and Information Services</organization>
3037<address>
3038<postal>
3039<street>100 Union Street SE</street>
3040<street>Room 152 Shepherd Labs</street>
3041<city>Minneapolis</city>
3042<region>MN</region>
3043<code>55455</code>
3044<country>US</country></postal>
3045<phone>+1 612 625 1300</phone>
3046<email>lindner@boombox.micro.umn.edu</email></address></author>
3047<author initials="D." surname="Johnson" fullname="David Johnson">
3048<organization>University of Minnesota, Computer and Information Services</organization>
3049<address>
3050<postal>
3051<street>100 Union Street SE</street>
3052<street>Room 152 Shepherd Labs</street>
3053<city>Minneapolis</city>
3054<region>MN</region>
3055<code>55455</code>
3056<country>US</country></postal>
3057<phone>+1 612 625 1300</phone>
3058<email>dmj@boombox.micro.umn.edu</email></address></author>
3059<author initials="D." surname="Torrey" fullname="Daniel Torrey">
3060<organization>University of Minnesota, Computer and Information Services</organization>
3061<address>
3062<postal>
3063<street>100 Union Street SE</street>
3064<street>Room 152 Shepherd Labs</street>
3065<city>Minneapolis</city>
3066<region>MN</region>
3067<code>55455</code>
3068<country>US</country></postal>
3069<phone>+1 612 625 1300</phone>
3070<email>daniel@boombox.micro.umn.edu</email></address></author>
3071<author initials="B." surname="Alberti" fullname="Bob Alberti">
3072<organization>University of Minnesota, Computer and Information Services</organization>
3073<address>
3074<postal>
3075<street>100 Union Street SE</street>
3076<street>Room 152 Shepherd Labs</street>
3077<city>Minneapolis</city>
3078<region>MN</region>
3079<code>55455</code>
3080<country>US</country></postal>
3081<phone>+1 612 625 1300</phone>
3082<email>alberti@boombox.micro.umn.edu</email></address></author>
3083<date month="March" year="1993"/>
3084</front>
3085<seriesInfo name="RFC" value="1436"/>
3086</reference>
3087
3088<reference anchor="RFC1630">
3089<front>
3090<title abbrev="URIs in WWW">Universal Resource Identifiers in WWW: A Unifying Syntax for the Expression of Names and Addresses of Objects on the Network as used in the World-Wide Web</title>
3091<author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
3092<organization>CERN, World-Wide Web project</organization>
3093<address>
3094<postal>
3095<street>1211 Geneva 23</street>
3096<city/>
3097<region/>
3098<code/>
3099<country>CH</country></postal>
3100<phone>+41 22 7673755</phone>
3101<facsimile>+41 22 7677155</facsimile>
3102<email>timbl@info.cern.ch</email></address></author>
3103<date month="June" year="1994"/></front>
3104<seriesInfo name="RFC" value="1630"/>
3105</reference>
3106
3107<reference anchor="RFC1738">
3108<front>
3109<title>Uniform Resource Locators (URL)</title>
3110<author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
3111<organization>CERN, World-Wide Web project</organization>
3112<address>
3113<postal>
3114<street>1211 Geneva 23</street>
3115<city/>
3116<region/>
3117<code/>
3118<country>CH</country></postal>
3119<phone>+41 22 7673755</phone>
3120<facsimile>+41 22 7677155</facsimile>
3121<email>timbl@info.cern.ch</email></address></author>
3122<author initials="L." surname="Masinter" fullname="Larry Masinter">
3123<organization>Xerox PARC</organization>
3124<address>
3125<postal>
3126<street>3333 Coyote Hill Road</street>
3127<city>Palo Alto</city>
3128<region>CA</region>
3129<code>94034</code>
3130<country>US</country></postal>
3131<phone>+1 415 812 4365</phone>
3132<facsimile>+1 415 812 4333</facsimile>
3133<email>masinter@parc.xerox.com</email></address></author>
3134<author initials="M." surname="McCahill" fullname="Mark McCahill">
3135<organization>University of Minnesota, Computer and Information Services</organization>
3136<address>
3137<postal>
3138<street>100 Union Street SE, Shepherd Labs</street>
3139<street>Room 152</street>
3140<city>Minneapolis</city>
3141<region>MN</region>
3142<code>55455</code>
3143<country>US</country></postal>
3144<phone>+1 612 625 1300</phone>
3145<email>mpm@boombox.micro.umn.edu</email></address></author>
3146<date month="December" year="1994"/>
3147</front>
3148<seriesInfo name="RFC" value="1738"/>
3149</reference>
3150
3151<reference anchor="RFC1945">
3152<front>
3153<title abbrev="HTTP/1.0">Hypertext Transfer Protocol -- HTTP/1.0</title>
3154<author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
3155<organization>MIT, Laboratory for Computer Science</organization>
3156<address>
3157<postal>
3158<street>545 Technology Square</street>
3159<city>Cambridge</city>
3160<region>MA</region>
3161<code>02139</code>
3162<country>US</country></postal>
3163<phone/>
3164<facsimile>+1 617 258 8682</facsimile>
3165<email>timbl@w3.org</email></address></author>
3166<author initials="R.T." surname="Fielding" fullname="Roy T. Fielding">
3167<organization>University of California, Irvine, Department of Information and Computer Science</organization>
3168<address>
3169<postal>
3170<street/>
3171<city>Irvine</city>
3172<region>CA</region>
3173<code>92717-3425</code>
3174<country>US</country></postal>
3175<phone/>
3176<facsimile>+1 714 824 4056</facsimile>
3177<email>fielding@ics.uci.edu</email></address></author>
3178<author initials="H.F." surname="Nielsen" fullname="Henrik Frystyk Nielsen">
3179<organization>W3 Consortium, MIT Laboratory for Computer Science</organization>
3180<address>
3181<postal>
3182<street>545 Technology Square</street>
3183<city>Cambridge</city>
3184<region>MA</region>
3185<code>02139</code>
3186<country>US</country></postal>
3187<phone/>
3188<facsimile>+1 617 258 8682</facsimile>
3189<email>frystyk@w3.org</email></address></author>
3190<date month="May" year="1996"/>
3191</front>
3192<seriesInfo name="RFC" value="1945"/>
3193</reference>
3194
3195<reference anchor="RFC2045">
3196<front>
3197<title abbrev="Internet Message Bodies">Multipurpose Internet Mail Extensions (MIME) Part One: Format of Internet Message Bodies</title>
3198<author initials="N." surname="Freed" fullname="Ned Freed">
3199<organization>Innosoft International, Inc.</organization>
3200<address>
3201<postal>
3202<street>1050 East Garvey Avenue South</street>
3203<city>West Covina</city>
3204<region>CA</region>
3205<code>91790</code>
3206<country>US</country></postal>
3207<phone>+1 818 919 3600</phone>
3208<facsimile>+1 818 919 3614</facsimile>
3209<email>ned@innosoft.com</email></address></author>
3210<author initials="N.S." surname="Borenstein" fullname="Nathaniel S. Borenstein">
3211<organization>First Virtual Holdings</organization>
3212<address>
3213<postal>
3214<street>25 Washington Avenue</street>
3215<city>Morristown</city>
3216<region>NJ</region>
3217<code>07960</code>
3218<country>US</country></postal>
3219<phone>+1 201 540 8967</phone>
3220<facsimile>+1 201 993 3032</facsimile>
3221<email>nsb@nsb.fv.com</email></address></author>
3222<date month="November" year="1996"/>
3223</front>
3224<seriesInfo name="RFC" value="2045"/>
3225</reference>
3226
3227<reference anchor="RFC1123">
3228<front>
3229<title>Requirements for Internet Hosts - Application and Support</title>
3230<author initials="R." surname="Braden" fullname="Robert Braden">
3231<organization>University of Southern California (USC), Information Sciences Institute</organization>
3232<address>
3233<postal>
3234<street>4676 Admiralty Way</street>
3235<city>Marina del Rey</city>
3236<region>CA</region>
3237<code>90292-6695</code>
3238<country>US</country></postal>
3239<phone>+1 213 822 1511</phone>
3240<email>Braden@ISI.EDU</email></address></author>
3241<date month="October" year="1989"/></front>
3242<seriesInfo name="STD" value="3"/>
3243<seriesInfo name="RFC" value="1123"/>
3244</reference>
3245
3246<reference anchor="RFC822">
3247<front>
3248<title abbrev="Standard for ARPA Internet Text Messages">Standard for the format of ARPA Internet text messages</title>
3249<author initials="D.H." surname="Crocker" fullname="David H. Crocker">
3250<organization>University of Delaware, Dept. of Electrical Engineering</organization>
3251<address>
3252<postal>
3253<street/>
3254<city>Newark</city>
3255<region>DE</region>
3256<code>19711</code>
3257<country>US</country></postal>
3258<email>DCrocker@UDel-Relay</email></address></author>
3259<date month="August" day="13" year="1982"/></front>
3260<seriesInfo name="STD" value="11"/>
3261<seriesInfo name="RFC" value="822"/>
3262</reference>
3263
3264<reference anchor="WAIS">
3265<front>
3266<title>WAIS Interface Protocol Prototype Functional Specification (v1.5)</title>
3267<author initials="F." surname="Davis" fullname="F. Davis">
3268<organization>Thinking Machines Corporation</organization>
3269</author>
3270<author initials="B." surname="Kahle" fullname="B. Kahle">
3271<organization>Thinking Machines Corporation</organization>
3272</author>
3273<author initials="H." surname="Morris" fullname="H. Morris">
3274<organization>Thinking Machines Corporation</organization>
3275</author>
3276<author initials="J." surname="Salem" fullname="J. Salem">
3277<organization>Thinking Machines Corporation</organization>
3278</author>
3279<author initials="T." surname="Shen" fullname="T. Shen">
3280<organization>Thinking Machines Corporation</organization>
3281</author>
3282<author initials="R." surname="Wang" fullname="R. Wang">
3283<organization>Thinking Machines Corporation</organization>
3284</author>
3285<author initials="J." surname="Sui" fullname="J. Sui">
3286<organization>Thinking Machines Corporation</organization>
3287</author>
3288<author initials="M." surname="Grinbaum" fullname="M. Grinbaum">
3289<organization>Thinking Machines Corporation</organization>
3290</author>
3291<date month="April" year="1990"/></front>
3292<seriesInfo name="Thinking Machines Corporation" value=""/>
3293</reference>
3294
3295<reference anchor="RFC1808">
3296<front>
3297<title>Relative Uniform Resource Locators</title>
3298<author initials="R." surname="Fielding" fullname="Roy T. Fielding">
3299<organization>University of California Irvine, Department of Information and Computer Science</organization>
3300<address>
3301<postal>
3302<street/>
3303<city>Irvine</city>
3304<region>CA</region>
3305<code>92717-3425</code>
3306<country>US</country></postal>
3307<phone>+1 714 824 4049</phone>
3308<facsimile>+1 714 824 4056</facsimile>
3309<email>fielding@ics.uci.edu</email></address></author>
3310<date month="June" year="1995"/>
3311</front>
3312<seriesInfo name="RFC" value="1808"/>
3313</reference>
3314
3315<reference anchor="RFC1036">
3316<front>
3317<title abbrev="Standard for USENET Messages">Standard for interchange of USENET messages</title>
3318<author initials="M." surname="Horton" fullname="M. Horton">
3319<organization>AT&amp;amp;T Bell Laboratories</organization></author>
3320<author initials="R." surname="Adams" fullname="R. Adams">
3321<organization>Center for Seismic Studies</organization></author>
3322<date month="December" year="1987"/></front>
3323<seriesInfo name="RFC" value="1036"/>
3324</reference>
3325
3326<reference anchor="RFC977">
3327<front>
3328<title abbrev="Network News Transfer Protocol">Network News Transfer Protocol</title>
3329<author initials="B." surname="Kantor" fullname="Brian Kantor">
3330<organization>University of California, San Diego (UCSD)</organization></author>
3331<author initials="P." surname="Lapsley" fullname="Phil Lapsley">
3332<organization>University of California, Berkeley (UCB)</organization></author>
3333<date month="February" year="1986"/></front>
3334<seriesInfo name="RFC" value="977"/>
3335</reference>
3336
3337<reference anchor="RFC2047">
3338<front>
3339<title abbrev="Message Header Extensions">MIME (Multipurpose Internet Mail Extensions) Part Three: Message Header Extensions for Non-ASCII Text</title>
3340<author initials="K." surname="Moore" fullname="Keith Moore">
3341<organization>University of Tennessee</organization>
3342<address>
3343<postal>
3344<street>107 Ayres Hall</street>
3345<street>Knoxville TN 37996-1301</street></postal>
3346<email>moore@cs.utk.edu</email></address></author>
3347<date month="November" year="1996"/>
3348<area>Applications</area>
3349<keyword>Amercian Standard Code for Information Interchange</keyword>
3350<keyword>mail</keyword>
3351<keyword>multipurpose internet mail extensions</keyword>
3352</front>
3353<seriesInfo name="RFC" value="2047"/>
3354</reference>
3355
3356<reference anchor="RFC821">
3357<front>
3358<title>Simple Mail Transfer Protocol</title>
3359<author initials="J.B." surname="Postel" fullname="Jonathan B. Postel">
3360<organization>University of Southern California (USC)/Information Sciences Institute</organization>
3361<address>
3362<postal>
3363<street>4676 Admiralty Way</street>
3364<city>Marina del Rey</city>
3365<region>CA</region>
3366<code>90291</code>
3367<country>US</country></postal>
3368<phone>+1 213 822 1511</phone></address></author>
3369<date month="August" year="1982"/></front>
3370<seriesInfo name="STD" value="10"/>
3371<seriesInfo name="RFC" value="821"/>
3372</reference>
3373
3374<reference anchor="RFC1590">
3375<front>
3376<title>Media Type Registration Procedure</title>
3377<author initials="J." surname="Postel" fullname="Jon Postel">
3378<organization>USC/Information Sciences Institute</organization>
3379<address>
3380<postal>
3381<street>4676 Admiralty Way</street>
3382<city>Marina del Rey</city>
3383<region>CA</region>
3384<code>90292</code>
3385<country>US</country></postal>
3386<phone>+1 310 822 1511</phone>
3387<facsimile>+1 310 823 6714</facsimile>
3388<email>Postel@ISI.EDU</email></address></author>
3389<date month="November" year="1996"/>
3390</front>
3391<seriesInfo name="RFC" value="1590"/>
3392</reference>
3393
3394<reference anchor="RFC959">
3395<front>
3396<title abbrev="File Transfer Protocol">File Transfer Protocol</title>
3397<author initials="J." surname="Postel" fullname="J. Postel">
3398<organization>Information Sciences Institute (ISI)</organization></author>
3399<author initials="J." surname="Reynolds" fullname="J. Reynolds">
3400<organization/></author>
3401<date month="October" year="1985"/></front>
3402<seriesInfo name="STD" value="9"/>
3403<seriesInfo name="RFC" value="959"/>
3404</reference>
3405
3406<reference anchor="RFC1700">
3407<front>
3408<title abbrev="Assigned Numbers">Assigned Numbers</title>
3409<author initials="J." surname="Reynolds" fullname="Joyce K. Reynolds">
3410<organization>USC/Information Sciences Institute</organization>
3411<address>
3412<postal>
3413<street>4676 Admiralty Way</street>
3414<city>Marina del Rey</city>
3415<region>CA</region>
3416<code>90292-6695</code>
3417<country>US</country></postal>
3418<phone>+1 310 822 1511</phone>
3419<email>jkrey@isi.edu</email></address></author>
3420<author initials="J." surname="Postel" fullname="Jon Postel">
3421<organization>USC/Information Sciences Institute</organization>
3422<address>
3423<postal>
3424<street>4676 Admiralty Way</street>
3425<city>Marina del Rey</city>
3426<region>CA</region>
3427<code>90292-6695</code>
3428<country>US</country></postal>
3429<phone>+1 310 822 1511</phone>
3430<email>postel@isi.edu</email></address></author>
3431<date month="October" year="1994"/>
3432<abstract>
3433<t/></abstract></front>
3434<seriesInfo name="STD" value="2"/>
3435<seriesInfo name="RFC" value="1700"/>
3436</reference>
3437
3438<reference anchor="RFC1737">
3439<front>
3440<title abbrev="Requirements for Uniform Resource Names">Functional Requirements for Uniform Resource Names</title>
3441<author initials="L." surname="Masinter" fullname="Larry Masinter">
3442<organization>Xerox Palo Alto Research Center</organization>
3443<address>
3444<postal>
3445<street>3333 Coyote Hill Road</street>
3446<city>Palo Alto</city>
3447<region>CA</region>
3448<code>94304</code>
3449<country>US</country></postal>
3450<phone>+1 415 812 4365</phone>
3451<facsimile>+1 415 812 4333</facsimile>
3452<email>masinter@parc.xerox.com</email></address></author>
3453<author initials="K." surname="Sollins" fullname="Karen Sollins">
3454<organization>MIT Laboratory for Computer Science</organization>
3455<address>
3456<postal>
3457<street>545 Technology Square</street>
3458<city>Cambridge</city>
3459<region>MA</region>
3460<code>02139</code>
3461<country>US</country></postal>
3462<phone>+1 617 253 2673</phone>
3463<email>sollins@lcs.mit.edu</email></address></author>
3464<date month="December" year="1994"/></front>
3465<seriesInfo name="RFC" value="1737"/>
3466</reference>
3467
3468<reference anchor="USASCII">
3469  <front>
3470    <title>Coded Character Set -- 7-bit American Standard Code for Information Interchange</title>
3471    <author>
3472      <organization>American National Standards Institute</organization>
3473    </author>
3474    <date year="1986"/>
3475  </front>
3476  <seriesInfo name="ANSI" value="X3.4"/>
3477</reference>
3478
3479<reference anchor="ISO-8859">
3480<front>
3481<title>Information technology - 8-bit single byte coded graphic - character sets</title>
3482<author>
3483<organization>International Organization for Standardization</organization>
3484</author>
3485<date year="1987-1990"/>
3486</front>
3487<annotation>
3488  Part 1: Latin alphabet No. 1, ISO-8859-1:1987.
3489  Part 2: Latin alphabet No. 2, ISO-8859-2, 1987.
3490  Part 3: Latin alphabet No. 3, ISO-8859-3, 1988.
3491  Part 4: Latin alphabet No. 4, ISO-8859-4, 1988.
3492  Part 5: Latin/Cyrillic alphabet, ISO-8859-5, 1988.
3493  Part 6: Latin/Arabic alphabet, ISO-8859-6, 1987.
3494  Part 7: Latin/Greek alphabet, ISO-8859-7, 1987.
3495  Part 8: Latin/Hebrew alphabet, ISO-8859-8, 1988.
3496  Part 9: Latin alphabet No. 5, ISO-8859-9, 1990.
3497</annotation>
3498</reference>
3499
3500<reference anchor="RFC1900">
3501<front>
3502<title>Renumbering Needs Work</title>
3503<author initials="B." surname="Carpenter" fullname="Brian E. Carpenter">
3504<organization>CERN, Computing and Networks Division</organization>
3505<address>
3506<postal>
3507<street>1211 Geneva 23</street>
3508<country>CH</country></postal>
3509<phone>+41 22 7674967</phone>
3510<facsimile>+41 22 7677155</facsimile>
3511<email>brian@dxcoms.cern.ch</email></address></author>
3512<author initials="Y." surname="Rekhter" fullname="Yakov Rekhter">
3513<organization>cisco Systems</organization>
3514<address>
3515<postal>
3516<street>170 West Tasman Drive</street>
3517<city>San Jose</city>
3518<region>CA</region>
3519<code>95134</code>
3520<country>US</country></postal>
3521<phone>+1 914 528 0090</phone>
3522<email>yakov@cisco.com</email></address></author>
3523<date month="February" year="1996"/>
3524</front>
3525<seriesInfo name="RFC" value="1900"/>
3526</reference>
3527
3528<reference anchor="Pad1995">
3529<front>
3530<title>Improving HTTP Latency</title>
3531<author initials="V.N." surname="Padmanabhan" fullname="Venkata N. Padmanabhan"><organization/></author>
3532<author initials="J.C." surname="Mogul" fullname="Jeffrey C. Mogul"><organization/></author>
3533<date year="1995" month="Dec"/>
3534</front>
3535<seriesInfo name="Computer Networks and ISDN Systems" value="v. 28, pp. 25-35"/>
3536<annotation>
3537  Slightly revised version of paper in Proc. 2nd International WWW Conference '94: Mosaic and the Web, Oct. 1994,
3538  which is available at <eref target="http://www.ncsa.uiuc.edu/SDG/IT94/Proceedings/DDay/mogul/HTTPLatency.html"/>.
3539</annotation>
3540</reference>
3541
3542<reference anchor="Tou1998" target="http://www.isi.edu/touch/pubs/http-perf96/">
3543<front>
3544<title>Analysis of HTTP Performance</title>
3545<author initials="J." surname="Touch" fullname="Joe Touch">
3546  <organization>USC/Information Sciences Institute</organization>
3547  <address><email>touch@isi.edu</email></address>
3548</author>
3549<author initials="J." surname="Heidemann" fullname="John Heidemann">
3550  <organization>USC/Information Sciences Institute</organization>
3551  <address><email>johnh@isi.edu</email></address>
3552</author>
3553<author initials="K." surname="Obraczka" fullname="Katia Obraczka">
3554  <organization>USC/Information Sciences Institute</organization>
3555  <address><email>katia@isi.edu</email></address>
3556</author>
3557<date year="1998" month="Aug"/>
3558</front>
3559<seriesInfo name="ISI Research Report" value="ISI/RR-98-463 (original report dated Aug.1996)"/>
3560</reference>
3561
3562<reference anchor="RFC1305">
3563<front>
3564<title>Network Time Protocol (Version 3) Specification, Implementation</title>
3565<author initials="D." surname="Mills" fullname="David L. Mills">
3566<organization>University of Delaware, Electrical Engineering Department</organization>
3567<address>
3568<postal>
3569<street/>
3570<city>Newark</city>
3571<region>DE</region>
3572<code>19716</code>
3573<country>US</country></postal>
3574<phone>+1 302 451 8247</phone>
3575<email>mills@udel.edu</email></address></author>
3576<date month="March" year="1992"/>
3577<abstract>
3578<t>This document describes the Network Time Protocol (NTP), specifies its normal structure and summarizes information useful for its implementation. NTP provides the mechanisms to synchronize time and coordinate time distribution in a large, diverse internet operating at rates from mundane to lightwave. It uses a returnable-time design in which a distributed subnet of time servers operating in a self-organizing, hierarchical-master-slave configuration synchronizes local clocks within the subnet and to national time standards via wire or radio. The servers can also redistribute reference time via local routing algorithms and time daemons.</t></abstract></front>
3579<seriesInfo name="RFC" value="1305"/>
3580</reference>
3581
3582<reference anchor="Spe" target="http://sunsite.unc.edu/mdma-release/http-prob.html">
3583<front>
3584<title>Analysis of HTTP Performance Problems</title>
3585<author initials="S." surname="Spero" fullname="Simon E. Spero">
3586  <organization/>
3587</author>
3588<date/>
3589</front>
3590</reference>
3591
3592<reference anchor="RFC2068">
3593<front>
3594<title abbrev="HTTP/1.1">Hypertext Transfer Protocol -- HTTP/1.1</title>
3595<author initials="R." surname="Fielding" fullname="Roy T. Fielding">
3596<organization>University of California, Irvine, Department of Information and Computer Science</organization>
3597<address>
3598<postal>
3599<street/>
3600<city>Irvine</city>
3601<region>CA</region>
3602<code>92717-3425</code>
3603<country>US</country></postal>
3604<facsimile>+1 714 824 4056</facsimile>
3605<email>fielding@ics.uci.edu</email></address></author>
3606<author initials="J." surname="Gettys" fullname="Jim Gettys">
3607<organization>MIT Laboratory for Computer Science</organization>
3608<address>
3609<postal>
3610<street>545 Technology Square</street>
3611<city>Cambridge</city>
3612<region>MA</region>
3613<code>02139</code>
3614<country>US</country></postal>
3615<facsimile>+1 617 258 8682</facsimile>
3616<email>jg@w3.org</email></address></author>
3617<author initials="J." surname="Mogul" fullname="Jeffrey C. Mogul">
3618<organization>Digital Equipment Corporation, Western Research Laboratory</organization>
3619<address>
3620<postal>
3621<street>250 University Avenue</street>
3622<city>Palo Alto</city>
3623<region>CA</region>
3624<code>94301</code>
3625<country>US</country></postal>
3626<email>mogul@wrl.dec.com</email></address></author>
3627<author initials="H." surname="Nielsen" fullname="Henrik Frystyk Nielsen">
3628<organization>MIT Laboratory for Computer Science</organization>
3629<address>
3630<postal>
3631<street>545 Technology Square</street>
3632<city>Cambridge</city>
3633<region>MA</region>
3634<code>02139</code>
3635<country>US</country></postal>
3636<facsimile>+1 617 258 8682</facsimile>
3637<email>frystyk@w3.org</email></address></author>
3638<author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
3639<organization>MIT Laboratory for Computer Science</organization>
3640<address>
3641<postal>
3642<street>545 Technology Square</street>
3643<city>Cambridge</city>
3644<region>MA</region>
3645<code>02139</code>
3646<country>US</country></postal>
3647<facsimile>+1 617 258 8682</facsimile>
3648<email>timbl@w3.org</email></address></author>
3649<date month="January" year="1997"/>
3650<abstract>
3651<t>The Hypertext Transfer Protocol (HTTP) is an application-level protocol for distributed, collaborative, hypermedia information systems. It is a generic, stateless, object-oriented protocol which can be used for many tasks, such as name servers and distributed object management systems, through extension of its request methods. A feature of HTTP is the typing and negotiation of data representation, allowing systems to be built independently of the data being transferred.</t>
3652<t>HTTP has been in use by the World-Wide Web global information initiative since 1990. This specification defines the protocol referred to as "HTTP/1.1".</t></abstract></front>
3653<seriesInfo name="RFC" value="2068"/>
3654</reference>
3655
3656<reference anchor="RFC2119">
3657<front>
3658<title abbrev="RFC Key Words">Key words for use in RFCs to Indicate Requirement Levels</title>
3659<author initials="S." surname="Bradner" fullname="Scott Bradner">
3660<organization>Harvard University</organization>
3661<address>
3662<postal>
3663<street>1350 Mass. Ave.</street>
3664<street>Cambridge</street>
3665<street>MA 02138</street></postal>
3666<phone>- +1 617 495 3864</phone>
3667<email>-</email></address></author>
3668<date month="March" year="1997"/>
3669<area>General</area>
3670<keyword>keyword</keyword>
3671</front>
3672
3673<seriesInfo name="BCP" value="14"/>
3674<seriesInfo name="RFC" value="2119"/>
3675</reference>
3676
3677<reference anchor="RFC2145">
3678<front>
3679<title abbrev="HTTP Version Numbers">Use and Interpretation of HTTP Version Numbers</title>
3680<author initials="J.C." surname="Mogul" fullname="Jeffrey C. Mogul">
3681<organization>Western Research Laboratory</organization>
3682<address>
3683<postal>
3684<street>Digital Equipment Corporation</street>
3685<street>250 University Avenue</street>
3686<street>Palo Alto</street>
3687<street>California</street>
3688<street>94305</street>
3689<country>USA</country></postal>
3690<email>mogul@wrl.dec.com</email></address></author>
3691<author initials="R.T." surname="Fielding" fullname="Roy T. Fielding">
3692<organization>Department of Information and Computer Science</organization>
3693<address>
3694<postal>
3695<street>University of California</street>
3696<street>Irvine</street>
3697<street>CA 92717-3425</street>
3698<country>USA</country></postal>
3699<facsimile>+1 (714) 824-4056</facsimile>
3700<email>fielding@ics.uci.edu</email></address></author>
3701<author initials="J." surname="Gettys" fullname="Jim Gettys">
3702<organization>MIT Laboratory for Computer Science</organization>
3703<address>
3704<postal>
3705<street>545 Technology Square</street>
3706<street>Cambridge</street>
3707<street>MA 02139</street>
3708<country>USA</country></postal>
3709<facsimile>+1 (617) 258 8682</facsimile>
3710<email>jg@w3.org</email></address></author>
3711<author initials="H.F." surname="Nielsen" fullname="Henrik Frystyk Nielsen">
3712<organization>W3 Consortium</organization>
3713<address>
3714<postal>
3715<street>MIT Laboratory for Computer Science</street>
3716<street>545 Technology Square</street>
3717<street>Cambridge</street>
3718<street>MA 02139</street>
3719<country>USA</country></postal>
3720<facsimile>+1 (617) 258 8682</facsimile>
3721<email>frystyk@w3.org</email></address></author>
3722<date month="May" year="1997"/>
3723<area>Applications</area>
3724<keyword>HTTP</keyword>
3725<keyword>hypertext transfer protocol</keyword>
3726<abstract>
3727<t>
3728   HTTP request and response messages include an HTTP protocol version
3729   number.  Some confusion exists concerning the proper use and
3730   interpretation of HTTP version numbers, and concerning
3731   interoperability of HTTP implementations of different protocol
3732   versions.  This document is an attempt to clarify the situation.  It
3733   is not a modification of the intended meaning of the existing
3734   HTTP/1.0 and HTTP/1.1 documents, but it does describe the intention
3735   of the authors of those documents, and can be considered definitive
3736   when there is any ambiguity in those documents concerning HTTP
3737   version numbers, for all versions of HTTP.
3738</t></abstract></front>
3739<seriesInfo name="RFC" value="2145"/>
3740</reference>
3741
3742<reference anchor="Nie1997">
3743<front>
3744<title>Network Performance Effects of HTTP/1.1, CSS1, and PNG</title>
3745<author initials="H.F.." surname="Nielsen" fullname="H.F. Nielsen">
3746  <organization/>
3747</author>
3748<author initials="J." surname="Gettys" fullname="J. Gettys">
3749  <organization/>
3750</author>
3751<author initials="E." surname="Prud'hommeaux" fullname="E. Prud'hommeaux">
3752  <organization/>
3753</author>
3754<author initials="H." surname="Lie" fullname="H. Lie">
3755  <organization/>
3756</author>
3757<author initials="C." surname="Lilley" fullname="C. Lilley">
3758  <organization/>
3759</author>
3760<date year="1997" month="Sep"/>
3761</front>
3762<seriesInfo name="Proceedings of ACM SIGCOMM '97, Cannes France" value=""/>
3763</reference>
3764
3765<reference anchor="RFC2396">
3766<front>
3767<title abbrev="URI Generic Syntax">Uniform Resource Identifiers (URI): Generic Syntax</title>
3768<author initials="T." surname="Berners-Lee" fullname="Tim Berners-Lee">
3769<organization abbrev="MIT/LCS">World Wide Web Consortium</organization>
3770<address>
3771<postal>
3772<street>MIT Laboratory for Computer Science, NE43-356</street>
3773<street>545 Technology Square</street>
3774<city>Cambridge</city>
3775<region>MA</region>
3776<code>02139</code></postal>
3777<facsimile>+1(617)258-8682</facsimile>
3778<email>timbl@w3.org</email></address></author>
3779<author initials="R.T." surname="Fielding" fullname="Roy T. Fielding">
3780<organization abbrev="U.C. Irvine">Department of Information and Computer Science</organization>
3781<address>
3782<postal>
3783<street>University of California, Irvine</street>
3784<city>Irvine</city>
3785<region>CA</region>
3786<code>92697-3425</code></postal>
3787<facsimile>+1(949)824-1715</facsimile>
3788<email>fielding@ics.uci.edu</email></address></author>
3789<author initials="L." surname="Masinter" fullname="Larry Masinter">
3790<organization abbrev="Xerox Corporation">Xerox PARC</organization>
3791<address>
3792<postal>
3793<street>3333 Coyote Hill Road</street>
3794<city>Palo Alto</city>
3795<region>CA</region>
3796<code>94034</code></postal>
3797<facsimile>+1(415)812-4333</facsimile>
3798<email>masinter@parc.xerox.com</email></address></author>
3799<date month="August" year="1998"/>
3800<area>Applications</area>
3801<keyword>uniform resource</keyword>
3802<keyword>URI</keyword>
3803</front>
3804<seriesInfo name="RFC" value="2396"/>
3805</reference>
3806
3807<reference anchor="RFC2324">
3808<front>
3809<title abbrev="HTCPCP/1.0">Hyper Text Coffee Pot Control Protocol (HTCPCP/1.0)</title>
3810<author initials="L." surname="Masinter" fullname="Larry Masinter">
3811<organization>Xerox Palo Alto Research Center</organization>
3812<address>
3813<postal>
3814<street>3333 Coyote Hill Road</street>
3815<city>Palo Alto</city>
3816<region>CA</region>
3817<code>94304</code></postal>
3818<email>masinter@parc.xerox.com</email></address></author>
3819<date month="April" day="1" year="1998"/>
3820<area>General</area>
3821<keyword>control protocol</keyword>
3822<keyword>coffee</keyword>
3823</front>
3824<seriesInfo name="RFC" value="2324"/>
3825</reference>
3826
3827</references>
3828
3829<section title="Internet Media Type message/http and application/http" anchor="internet.media.type.http">
3830<iref item="Media Type" subitem="message/http" primary="true"/>
3831<iref item="message/http Media Type" primary="true"/>
3832<iref item="Media Type" subitem="application/http" primary="true"/>
3833<iref item="application/http Media Type" primary="true"/>
3834<t>
3835   In addition to defining the HTTP/1.1 protocol, this document serves
3836   as the specification for the Internet media type "message/http" and
3837   "application/http". The message/http type can be used to enclose a
3838   single HTTP request or response message, provided that it obeys the
3839   MIME restrictions for all "message" types regarding line length and
3840   encodings. The application/http type can be used to enclose a
3841   pipeline of one or more HTTP request or response messages (not
3842   intermixed). The following is to be registered with IANA <xref target="RFC1590"/>.
3843</t>
3844<t>
3845  <list style="hanging">
3846    <t hangText="Media Type name:">
3847      message
3848    </t>
3849    <t hangText="Media subtype name:">
3850      http
3851    </t>
3852    <t hangText="Required parameters:">
3853      none
3854    </t>
3855    <t hangText="Optional parameters:">
3856      version, msgtype
3857      <list style="hanging">
3858        <t hangText="version:">
3859          The HTTP-Version number of the enclosed message
3860          (e.g., "1.1"). If not present, the version can be
3861          determined from the first line of the body.
3862        </t>
3863        <t hangText="msgtype:">
3864          The message type -- "request" or "response". If not
3865          present, the type can be determined from the first
3866          line of the body.
3867        </t>
3868      </list>
3869    </t>
3870    <t hangText="Encoding considerations:">
3871      only "7bit", "8bit", or "binary" are permitted
3872    </t>
3873    <t hangText="Security considerations:">
3874      none
3875    </t>
3876  </list>
3877</t>
3878<t>
3879  <list style="hanging">
3880    <t hangText="Media Type name:">
3881      application
3882    </t>
3883    <t hangText="Media subtype name:">
3884      http
3885    </t>
3886    <t hangText="Required parameters:">
3887      none
3888    </t>
3889    <t hangText="Optional parameters:">
3890      version, msgtype
3891      <list style="hanging">
3892        <t hangText="version:">
3893          The HTTP-Version number of the enclosed messages
3894          (e.g., "1.1"). If not present, the version can be
3895          determined from the first line of the body.
3896        </t>
3897        <t hangText="msgtype:">
3898          The message type -- "request" or "response". If not
3899          present, the type can be determined from the first
3900          line of the body.
3901        </t>
3902      </list>
3903    </t>
3904    <t hangText="Encoding considerations:">
3905      HTTP messages enclosed by this type
3906      are in "binary" format; use of an appropriate
3907      Content-Transfer-Encoding is required when
3908      transmitted via E-mail.
3909    </t>
3910    <t hangText="Security considerations:">
3911      none
3912    </t>
3913  </list>
3914</t>
3915</section>
3916
3917<section title="Tolerant Applications" anchor="tolerant.applications">
3918<t>
3919   Although this document specifies the requirements for the generation
3920   of HTTP/1.1 messages, not all applications will be correct in their
3921   implementation. We therefore recommend that operational applications
3922   be tolerant of deviations whenever those deviations can be
3923   interpreted unambiguously.
3924</t>
3925<t>
3926   Clients SHOULD be tolerant in parsing the Status-Line and servers
3927   tolerant when parsing the Request-Line. In particular, they SHOULD
3928   accept any amount of SP or HT characters between fields, even though
3929   only a single SP is required.
3930</t>
3931<t>
3932   The line terminator for message-header fields is the sequence CRLF.
3933   However, we recommend that applications, when parsing such headers,
3934   recognize a single LF as a line terminator and ignore the leading CR.
3935</t>
3936<t>
3937   The character set of an entity-body SHOULD be labeled as the lowest
3938   common denominator of the character codes used within that body, with
3939   the exception that not labeling the entity is preferred over labeling
3940   the entity with the labels US-ASCII or ISO-8859-1. See <xref target="Part3"/>.
3941</t>
3942<t>
3943   Additional rules for requirements on parsing and encoding of dates
3944   and other potential problems with date encodings include:
3945</t>
3946<t>
3947  <list style="symbols">
3948     <t>HTTP/1.1 clients and caches SHOULD assume that an RFC-850 date
3949        which appears to be more than 50 years in the future is in fact
3950        in the past (this helps solve the "year 2000" problem).</t>
3951
3952     <t>An HTTP/1.1 implementation MAY internally represent a parsed
3953        Expires date as earlier than the proper value, but MUST NOT
3954        internally represent a parsed Expires date as later than the
3955        proper value.</t>
3956
3957     <t>All expiration-related calculations MUST be done in GMT. The
3958        local time zone MUST NOT influence the calculation or comparison
3959        of an age or expiration time.</t>
3960
3961     <t>If an HTTP header incorrectly carries a date value with a time
3962        zone other than GMT, it MUST be converted into GMT using the
3963        most conservative possible conversion.</t>
3964  </list>
3965</t>
3966</section>
3967
3968<section title="Conversion of Date Formats" anchor="conversion.of.date.formats">
3969<t>
3970   HTTP/1.1 uses a restricted set of date formats (<xref target="full.date"/>) to
3971   simplify the process of date comparison. Proxies and gateways from
3972   other protocols SHOULD ensure that any Date header field present in a
3973   message conforms to one of the HTTP/1.1 formats and rewrite the date
3974   if necessary.
3975</t>
3976</section>
3977
3978<section title="Compatibility with Previous Versions" anchor="compatibility">
3979<t>
3980   It is beyond the scope of a protocol specification to mandate
3981   compliance with previous versions. HTTP/1.1 was deliberately
3982   designed, however, to make supporting previous versions easy. It is
3983   worth noting that, at the time of composing this specification
3984   (1996), we would expect commercial HTTP/1.1 servers to:
3985  <list style="symbols">
3986     <t>recognize the format of the Request-Line for HTTP/0.9, 1.0, and
3987        1.1 requests;</t>
3988
3989     <t>understand any valid request in the format of HTTP/0.9, 1.0, or
3990        1.1;</t>
3991
3992     <t>respond appropriately with a message in the same major version
3993        used by the client.</t>
3994  </list>
3995</t>
3996<t>
3997   And we would expect HTTP/1.1 clients to:
3998  <list style="symbols">
3999     <t>recognize the format of the Status-Line for HTTP/1.0 and 1.1
4000        responses;</t>
4001
4002     <t>understand any valid response in the format of HTTP/0.9, 1.0, or
4003        1.1.</t>
4004  </list>
4005</t>
4006<t>
4007   For most implementations of HTTP/1.0, each connection is established
4008   by the client prior to the request and closed by the server after
4009   sending the response. Some implementations implement the Keep-Alive
4010   version of persistent connections described in Section 19.7.1 of RFC
4011   2068 <xref target="RFC2068"/>.
4012</t>
4013
4014<section title="Changes from HTTP/1.0" anchor="changes.from.1.0">
4015<t>
4016   This section summarizes major differences between versions HTTP/1.0
4017   and HTTP/1.1.
4018</t>
4019
4020<section title="Changes to Simplify Multi-homed Web Servers and Conserve IP Addresses" anchor="changes.to.simplify.multi-homed.web.servers.and.conserve.ip.addresses">
4021<t>
4022   The requirements that clients and servers support the Host request-header,
4023   report an error if the Host request-header (<xref target="header.host"/>) is
4024   missing from an HTTP/1.1 request, and accept absolute URIs (<xref target="request-uri"/>)
4025   are among the most important changes defined by this
4026   specification.
4027</t>
4028<t>
4029   Older HTTP/1.0 clients assumed a one-to-one relationship of IP
4030   addresses and servers; there was no other established mechanism for
4031   distinguishing the intended server of a request than the IP address
4032   to which that request was directed. The changes outlined above will
4033   allow the Internet, once older HTTP clients are no longer common, to
4034   support multiple Web sites from a single IP address, greatly
4035   simplifying large operational Web servers, where allocation of many
4036   IP addresses to a single host has created serious problems. The
4037   Internet will also be able to recover the IP addresses that have been
4038   allocated for the sole purpose of allowing special-purpose domain
4039   names to be used in root-level HTTP URLs. Given the rate of growth of
4040   the Web, and the number of servers already deployed, it is extremely
4041   important that all implementations of HTTP (including updates to
4042   existing HTTP/1.0 applications) correctly implement these
4043   requirements:
4044  <list style="symbols">
4045     <t>Both clients and servers MUST support the Host request-header.</t>
4046
4047     <t>A client that sends an HTTP/1.1 request MUST send a Host header.</t>
4048
4049     <t>Servers MUST report a 400 (Bad Request) error if an HTTP/1.1
4050        request does not include a Host request-header.</t>
4051
4052     <t>Servers MUST accept absolute URIs.</t>
4053  </list>
4054</t>
4055</section>
4056</section>
4057
4058<section title="Compatibility with HTTP/1.0 Persistent Connections" anchor="compatibility.with.http.1.0.persistent.connections">
4059<t>
4060   Some clients and servers might wish to be compatible with some
4061   previous implementations of persistent connections in HTTP/1.0
4062   clients and servers. Persistent connections in HTTP/1.0 are
4063   explicitly negotiated as they are not the default behavior. HTTP/1.0
4064   experimental implementations of persistent connections are faulty,
4065   and the new facilities in HTTP/1.1 are designed to rectify these
4066   problems. The problem was that some existing 1.0 clients may be
4067   sending Keep-Alive to a proxy server that doesn't understand
4068   Connection, which would then erroneously forward it to the next
4069   inbound server, which would establish the Keep-Alive connection and
4070   result in a hung HTTP/1.0 proxy waiting for the close on the
4071   response. The result is that HTTP/1.0 clients must be prevented from
4072   using Keep-Alive when talking to proxies.
4073</t>
4074<t>
4075   However, talking to proxies is the most important use of persistent
4076   connections, so that prohibition is clearly unacceptable. Therefore,
4077   we need some other mechanism for indicating a persistent connection
4078   is desired, which is safe to use even when talking to an old proxy
4079   that ignores Connection. Persistent connections are the default for
4080   HTTP/1.1 messages; we introduce a new keyword (Connection: close) for
4081   declaring non-persistence. See <xref target="header.connection"/>.
4082</t>
4083<t>
4084   The original HTTP/1.0 form of persistent connections (the Connection:
4085   Keep-Alive and Keep-Alive header) is documented in RFC 2068. <xref target="RFC2068"/>
4086</t>
4087</section>
4088
4089<section title="Changes from RFC 2068" anchor="changes.from.rfc.2068">
4090<t>
4091   This specification has been carefully audited to correct and
4092   disambiguate key word usage; RFC 2068 had many problems in respect to
4093   the conventions laid out in RFC 2119 <xref target="RFC2119"/>.
4094</t>
4095<t>
4096   Transfer-coding and message lengths all interact in ways that
4097   required fixing exactly when chunked encoding is used (to allow for
4098   transfer encoding that may not be self delimiting); it was important
4099   to straighten out exactly how message lengths are computed.
4100</t>
4101<t>
4102   The use and interpretation of HTTP version numbers has been clarified
4103   by RFC 2145. Require proxies to upgrade requests to highest protocol
4104   version they support to deal with problems discovered in HTTP/1.0
4105   implementations (<xref target="http.version"/>)
4106</t>
4107<t>
4108   Proxies should be able to add Content-Length when appropriate.
4109</t>
4110<t>
4111   Transfer-coding had significant problems, particularly with
4112   interactions with chunked encoding. The solution is that transfer-codings
4113   become as full fledged as content-codings. This involves
4114   adding an IANA registry for transfer-codings (separate from content
4115   codings), a new header field (TE) and enabling trailer headers in the
4116   future. Transfer encoding is a major performance benefit, so it was
4117   worth fixing <xref target="Nie1997"/>. TE also solves another, obscure, downward
4118   interoperability problem that could have occurred due to interactions
4119   between authentication trailers, chunked encoding and HTTP/1.0
4120   clients.(Section <xref target="transfer.codings" format="counter"/>, <xref target="chunked.transfer.encoding" format="counter"/>,
4121   and <xref target="header.te" format="counter"/>)
4122</t>
4123</section>
4124</section>
4125</back>
4126</rfc>
Note: See TracBrowser for help on using the repository browser.