Context Navigation

← Previous Change
Next Change →

Changeset 540 for draft-ietf-httpbis

Timestamp:

06/03/09 14:08:48 (12 years ago)

Author:

julian.reschke@…

Message:

reverted back P6's indentation to match the one used in the other parts; also minor xml2rfc syntax cleanup

Location:

draft-ietf-httpbis/latest

Files:

: 8 edited

p1-messaging.html (modified) (1 diff)
p2-semantics.html (modified) (1 diff)
p3-payload.html (modified) (1 diff)
p4-conditional.html (modified) (1 diff)
p5-range.html (modified) (1 diff)
p6-cache.html (modified) (20 diffs)
p6-cache.xml (modified) (10 diffs)
p7-auth.html (modified) (1 diff)

Legend:

: Unmodified
: Added
: Removed

draft-ietf-httpbis/latest/p1-messaging.html

r538	r540
471	471	<tr>
472	472	<td class="header left"></td>
473		<td class="header right">March 5, 2009</td>
	473	<td class="header right">March 6, 2009</td>
474	474	</tr>
475	475	</table>

draft-ietf-httpbis/latest/p2-semantics.html

r538	r540
468	468	<tr>
469	469	<td class="header left"></td>
470		<td class="header right">March 5, 2009</td>
	470	<td class="header right">March 6, 2009</td>
471	471	</tr>
472	472	</table>

draft-ietf-httpbis/latest/p3-payload.html

r538	r540
475	475	<tr>
476	476	<td class="header left"></td>
477		<td class="header right">March 5, 2009</td>
	477	<td class="header right">March 6, 2009</td>
478	478	</tr>
479	479	</table>

draft-ietf-httpbis/latest/p4-conditional.html

r538	r540
464	464	<tr>
465	465	<td class="header left"></td>
466		<td class="header right">March 5, 2009</td>
	466	<td class="header right">March 6, 2009</td>
467	467	</tr>
468	468	</table>

draft-ietf-httpbis/latest/p5-range.html

r532	r540
464	464	<tr>
465	465	<td class="header left"></td>
466		<td class="header right">March 5, 2009</td>
	466	<td class="header right">March 6, 2009</td>
467	467	</tr>
468	468	</table>

draft-ietf-httpbis/latest/p6-cache.html

-                      r539
+                      r540
 cite {
   font-style: normal;
+}
+div.note {
+  margin-left: 2em;
+}
 dd {
 …
          <tr>
             <td class="header left"></td>
             <td class="header right">March 5, 2009</td>
+            <td class="header right">March 6, 2009</td>
          </tr>
       </table>
 …
                <li>successfully validated (see <a href="#validation.model" title="Validation Model">Section&nbsp;2.4</a>).
                </li>
             </ul>
+            </ul>
          </li>
       </ul>
+      <p id="rfc.section.2.2.p.2"><span class="comment">[rfc.comment.2: TODO: define method cacheability for GET, HEAD and POST in p2-semantics.]</span></p>
+      <p id="rfc.section.2.2.p.3">When a stored response is used to satisfy a request, caches <em class="bcp14">MUST</em> include a single Age header field <a href="#header.age" id="rfc.xref.header.age.1" title="Age">Section&nbsp;3.1</a> in the response with a value equal to the stored response's current_age; see <a href="#age.calculations" title="Calculating Age">Section&nbsp;2.3.2</a>. <span class="comment">[rfc.comment.3: DISCUSS: this currently includes successfully validated responses.]</span></p>
+      <p id="rfc.section.2.2.p.2"> <span class="comment">[rfc.comment.2: TODO: define method cacheability for GET, HEAD and POST in p2-semantics.]</span>
+      </p>
+      <p id="rfc.section.2.2.p.3">When a stored response is used to satisfy a request, caches <em class="bcp14">MUST</em> include a single Age header field <a href="#header.age" id="rfc.xref.header.age.1" title="Age">Section&nbsp;3.1</a> in the response with a value equal to the stored response's current_age; see <a href="#age.calculations" title="Calculating Age">Section&nbsp;2.3.2</a>. <span class="comment">[rfc.comment.3: DISCUSS: this currently includes successfully validated responses.]</span>
+      </p>
       <p id="rfc.section.2.2.p.4">Requests with methods that are unsafe (<a href="p2-semantics.html#safe.methods" title="Safe Methods">Section 7.1.1</a> of <a href="#Part2" id="rfc.xref.Part2.1"><cite title="HTTP/1.1, part 2: Message Semantics">[Part2]</cite></a>) <em class="bcp14">MUST</em> be written through the cache to the origin server; i.e., A cache must not reply to such a request before having forwarded
          the request and having received a corresponding response.
 …
       </p>
       <p id="rfc.section.2.3.p.3">If an origin server wishes to force a cache to validate every request, it can assign an explicit expiration time in the past.
+         This means that the response is always stale, so that caches should validate it before using it for subsequent requests. <span class="comment">[rfc.comment.5: This wording may cause confusion, because the response may still be served stale.]</span></p>
+         This means that the response is always stale, so that caches should validate it before using it for subsequent requests. <span class="comment">[rfc.comment.5: This wording may cause confusion, because the response may still be served stale.]</span>
+      </p>
       <p id="rfc.section.2.3.p.4">Since origin servers do not always provide explicit expiration times, HTTP caches may also assign heuristic expiration times
          when they are not specified, employing algorithms that use other header values (such as the Last-Modified time) to estimate
 …
          on their results.
       </p>
       <p id="rfc.section.2.3.p.5">The calculation to determine if a response is fresh is:</p>
       <div id="rfc.figure.u.3"></div> <pre class="text">   response_is_fresh = (freshness_lifetime &gt; current_age)
 </pre> <p id="rfc.section.2.3.p.7">The freshness_lifetime is defined in <a href="#calculating.freshness.lifetime" title="Calculating Freshness Lifetime">Section&nbsp;2.3.1</a>; the current_age is defined in <a href="#age.calculations" title="Calculating Age">Section&nbsp;2.3.2</a>.
       </p>
       <p id="rfc.section.2.3.p.8">Additionally, clients may need to influence freshness calculation. They can do this using several request cache directives,
+      <div id="rfc.figure.u.3"></div>
+      <p>The calculation to determine if a response is fresh is:</p>  <pre class="text">   response_is_fresh = (freshness_lifetime &gt; current_age)
+</pre> <p id="rfc.section.2.3.p.6">The freshness_lifetime is defined in <a href="#calculating.freshness.lifetime" title="Calculating Freshness Lifetime">Section&nbsp;2.3.1</a>; the current_age is defined in <a href="#age.calculations" title="Calculating Age">Section&nbsp;2.3.2</a>.
+      </p>
+      <p id="rfc.section.2.3.p.7">Additionally, clients may need to influence freshness calculation. They can do this using several request cache directives,
          with the effect of either increasing or loosening constraints on freshness. See <a href="#cache-request-directive" title="Request Cache-Control Directives">Section&nbsp;3.2.1</a>.
       </p>
       <p id="rfc.section.2.3.p.9"> <span class="comment">[rfc.comment.6: ISSUE: there are not requirements directly applying to cache-request-directives and freshness.]</span>
       </p>
       <p id="rfc.section.2.3.p.10">Note that freshness applies only to cache operation; it cannot be used to force a user agent to refresh its display or reload
+      <p id="rfc.section.2.3.p.8"> <span class="comment">[rfc.comment.6: ISSUE: there are not requirements directly applying to cache-request-directives and freshness.]</span>
+      </p>
+      <p id="rfc.section.2.3.p.9">Note that freshness applies only to cache operation; it cannot be used to force a user agent to refresh its display or reload
          a resource. See <a href="#history.lists" title="History Lists">Section&nbsp;4</a> for an explanation of the difference between caches and history mechanisms.
       </p>
 …
          <li>age_value, if all of the caches along the response path implement HTTP/1.1.</li>
       </ol>
       <p id="rfc.section.2.3.2.p.6">These are combined as</p>
       <div id="rfc.figure.u.4"></div> <pre class="text">    corrected_received_age = max(now - date_value, age_value)
 </pre> <p id="rfc.section.2.3.2.p.8">When an Age value is received, it <em class="bcp14">MUST</em> be interpreted relative to the time the request was initiated, not the time that the response was received.
       </p>
       <div id="rfc.figure.u.5"></div> <pre class="text">   corrected_initial_age = corrected_received_age
+      <div id="rfc.figure.u.4"></div>
+      <p>These are combined as</p>  <pre class="text">    corrected_received_age = max(now - date_value, age_value)
+</pre><p id="rfc.section.2.3.2.p.7">When an Age value is received, it <em class="bcp14">MUST</em> be interpreted relative to the time the request was initiated, not the time that the response was received.
+      </p>
+      <div id="rfc.figure.u.5"></div><pre class="text">   corrected_initial_age = corrected_received_age
                          + (now - request_time)
 </pre> <p id="rfc.section.2.3.2.p.10">where "request_time" is the time (according to the local clock) when the request that elicited this response was sent.</p>
       <p id="rfc.section.2.3.2.p.11">The current_age of a stored response can then be calculated by adding the amount of time (in seconds) since the stored response
+</pre><p id="rfc.section.2.3.2.p.9">where "request_time" is the time (according to the local clock) when the request that elicited this response was sent.</p>
+      <p id="rfc.section.2.3.2.p.10">The current_age of a stored response can then be calculated by adding the amount of time (in seconds) since the stored response
          was last validated by the origin server to the corrected_initial_age.
       </p>
       <p id="rfc.section.2.3.2.p.12">In summary:</p>
       <div id="rfc.figure.u.6"></div> <pre class="text">   age_value     - Age header field-value received with the response
    date_value    - Date header field-value received with the response
    request_time  - local time when the cache made the request
                    resulting in the stored response
    response_time - local time when the cache received the response
    now           - current local time
    apparent_age = max(0, response_time - date_value);
    corrected_received_age = max(apparent_age, age_value);
    response_delay = response_time - request_time;
    corrected_initial_age = corrected_received_age + response_delay;
    resident_time = now - response_time;
    current_age   = corrected_initial_age + resident_time;
 </pre> <h3 id="rfc.section.2.3.3"><a href="#rfc.section.2.3.3">2.3.3</a>&nbsp;<a id="serving.stale.responses" href="#serving.stale.responses">Serving Stale Responses</a></h3>
+      <p id="rfc.section.2.3.2.p.11">In summary:</p>
+      <div id="rfc.figure.u.6"></div><pre class="text">  age_value     - Age header field-value received with the response
+  date_value    - Date header field-value received with the response
+  request_time  - local time when the cache made the request
+                 resulting in the stored response
+  response_time - local time when the cache received the response
+  now           - current local time
+  apparent_age = max(0, response_time - date_value);
+  corrected_received_age = max(apparent_age, age_value);
+  response_delay = response_time - request_time;
+  corrected_initial_age = corrected_received_age + response_delay;
+  resident_time = now - response_time;
+  current_age   = corrected_initial_age + resident_time;
+</pre><h3 id="rfc.section.2.3.3"><a href="#rfc.section.2.3.3">2.3.3</a>&nbsp;<a id="serving.stale.responses" href="#serving.stale.responses">Serving Stale Responses</a></h3>
       <p id="rfc.section.2.3.3.p.1">A "stale" response is one that either has explicit expiry information, or is allowed to have heuristic expiry calculated,
          but is not fresh according to the calculations in <a href="#expiration.model" title="Freshness Model">Section&nbsp;2.3</a>.
 …
       </p>
       <p id="rfc.section.2.3.3.p.3">Caches <em class="bcp14">SHOULD NOT</em> return stale responses unless they are disconnected (i.e., it cannot contact the origin server or otherwise find a forward
          path) or otherwise explicitly allowed (e.g., the max-stale request directive; see <a href="#cache-request-directive" title="Request Cache-Control Directives">Section&nbsp;3.2.1</a>)..
+         path) or otherwise explicitly allowed (e.g., the max-stale request directive; see <a href="#cache-request-directive" title="Request Cache-Control Directives">Section&nbsp;3.2.1</a>).
       </p>
       <p id="rfc.section.2.3.3.p.4">Stale responses <em class="bcp14">SHOULD</em> have a Warning header with the 110 warn-code (see <a href="#header.warning" id="rfc.xref.header.warning.1" title="Warning">Section&nbsp;3.6</a>). Likewise, the 112 warn-code <em class="bcp14">SHOULD</em> be sent on stale responses if the cache is disconnected.
 …
       </p>
       <p id="rfc.section.2.4.p.4">If instead the cache receives a full response (i.e., one with a response body), it is used to satisfy the request and replace
+         the stored response. <span class="comment">[rfc.comment.8: Should there be a requirement here?]</span></p>
+         the stored response. <span class="comment">[rfc.comment.8: Should there be a requirement here?]</span>
+      </p>
       <p id="rfc.section.2.4.p.5">If a cache receives a 5xx response while attempting to validate a response, it <em class="bcp14">MAY</em> either forward this response to the requesting client, or act as if the server failed to respond. In the latter case, it <em class="bcp14">MAY</em> return a previously stored response (which <em class="bcp14">SHOULD</em> include the 111 warn-code; see <a href="#header.warning" id="rfc.xref.header.warning.2" title="Warning">Section&nbsp;3.6</a>) unless the stored response includes the "must-revalidate" cache directive (see <a href="#serving.stale.responses" title="Serving Stale Responses">Section&nbsp;2.3.3</a>).
       </p>
 …
       <p id="rfc.section.2.6.p.3">The selecting request-headers from two requests are defined to match if and only if the selecting request-headers in the first
          request can be transformed to the selecting request-headers in the second request by adding or removing linear white space <span class="comment">[rfc.comment.11: [ref]]</span> at places where this is allowed by the corresponding ABNF, and/or combining multiple message-header fields with the same field
+         name following the rules about message headers in <a href="p1-messaging.html#message.headers" title="Message Headers">Section 4.2</a> of <a href="#Part1" id="rfc.xref.Part1.12"><cite title="HTTP/1.1, part 1: URIs, Connections, and Message Parsing">[Part1]</cite></a>. <span class="comment">[rfc.comment.12: DISCUSS: header-specific canonicalisation]</span></p>
+         name following the rules about message headers in <a href="p1-messaging.html#message.headers" title="Message Headers">Section 4.2</a> of <a href="#Part1" id="rfc.xref.Part1.12"><cite title="HTTP/1.1, part 1: URIs, Connections, and Message Parsing">[Part1]</cite></a>. <span class="comment">[rfc.comment.12: DISCUSS: header-specific canonicalisation]</span>
+      </p>
       <p id="rfc.section.2.6.p.4">A Vary header field-value of "*" always fails to match, and subsequent requests to that resource can only be properly interpreted
          by the origin server.
 …
       <p id="rfc.section.2.6.p.7">If a cache receives a successful response whose Content-Location field matches that of an existing stored response for the
          same Request-URI, whose entity-tag differs from that of the existing stored response, and whose Date is more recent than that
+         of the existing response, the existing response <em class="bcp14">SHOULD NOT</em> be returned in response to future requests and <em class="bcp14">SHOULD</em> be deleted from the cache.<span class="comment">[rfc.comment.13: DISCUSS: Not sure if this is necessary.]</span></p>
+         of the existing response, the existing response <em class="bcp14">SHOULD NOT</em> be returned in response to future requests and <em class="bcp14">SHOULD</em> be deleted from the cache.<span class="comment">[rfc.comment.13: DISCUSS: Not sure if this is necessary.]</span>
+      </p>
       <h2 id="rfc.section.2.7"><a href="#rfc.section.2.7">2.7</a>&nbsp;<a id="combining.headers" href="#combining.headers">Combining Responses</a></h2>
       <p id="rfc.section.2.7.p.1">When a cache receives a 304 (Not Modified) response or a 206 (Partial Content) response, it needs to update the stored response
 …
          all such old headers <em class="bcp14">MUST</em> be replaced. It <em class="bcp14">MAY</em> store the combined entity-body.
       </p>
+      <p id="rfc.section.2.7.p.5"><span class="comment">[rfc.comment.14: ISSUE: discuss how to handle HEAD updates]</span></p>
+      <p id="rfc.section.2.7.p.5"> <span class="comment">[rfc.comment.14: ISSUE: discuss how to handle HEAD updates]</span>
+      </p>
       <h1 id="rfc.section.3"><a href="#rfc.section.3">3.</a>&nbsp;<a id="header.fields" href="#header.fields">Header Field Definitions</a></h1>
       <p id="rfc.section.3.p.1">This section defines the syntax and semantics of HTTP/1.1 header fields related to caching.</p>
 …
          <p id="rfc.section.3.1.p.3">  Age field-values are non-negative decimal integers, representing time in seconds.</p>
       </div>
       <div id="rfc.figure.u.8"></div> <pre class="inline"><span id="rfc.iref.g.3"></span>  <a href="#rule.delta-seconds" class="smpl">delta-seconds</a>  = 1*<a href="#notation" class="smpl">DIGIT</a>
 </pre> <p id="rfc.section.3.1.p.5">If a cache receives a value larger than the largest positive integer it can represent, or if any of its age calculations overflows,
+      <div id="rfc.figure.u.8"></div><pre class="inline"><span id="rfc.iref.g.3"></span>  <a href="#rule.delta-seconds" class="smpl">delta-seconds</a>  = 1*<a href="#notation" class="smpl">DIGIT</a>
+</pre><p id="rfc.section.3.1.p.5">If a cache receives a value larger than the largest positive integer it can represent, or if any of its age calculations overflows,
          it <em class="bcp14">MUST</em> transmit an Age header with a field-value of 2147483648 (2<sup>31</sup>). Caches <em class="bcp14">SHOULD</em> use an arithmetic type of at least 31 bits of range.
       </p>
 …
       <p id="rfc.section.3.2.p.1">The general-header field "Cache-Control" is used to specify directives that <em class="bcp14">MUST</em> be obeyed by all caches along the request/response chain. The directives specify behavior intended to prevent caches from
          adversely interfering with the request or response. Cache directives are unidirectional in that the presence of a directive
          in a request does not imply that the same directive is to be given in the response.
       </p>
       <dl class="empty">
          <dd>Note that HTTP/1.0 caches might not implement Cache-Control and might only implement Pragma: no-cache (see <a href="#header.pragma" id="rfc.xref.header.pragma.2" title="Pragma">Section&nbsp;3.4</a>).
          </dd>
       </dl>
       <p id="rfc.section.3.2.p.2">Cache directives <em class="bcp14">MUST</em> be passed through by a proxy or gateway application, regardless of their significance to that application, since the directives
+         in a request does not imply that the same directive is to be given in the response.
+      </p>
+      <div class="note">
+         <p>Note that HTTP/1.0 caches might not implement Cache-Control and might only implement Pragma: no-cache (see <a href="#header.pragma" id="rfc.xref.header.pragma.2" title="Pragma">Section&nbsp;3.4</a>).
+         </p>
+      </div>
+      <p id="rfc.section.3.2.p.3">Cache directives <em class="bcp14">MUST</em> be passed through by a proxy or gateway application, regardless of their significance to that application, since the directives
          might be applicable to all recipients along the request/response chain. It is not possible to target a directive to a specific
          cache.
 …
   <a href="#header.cache-control" class="smpl">cache-extension</a> = <a href="#core.rules" class="smpl">token</a> [ "=" ( <a href="#core.rules" class="smpl">token</a> / <a href="#core.rules" class="smpl">quoted-string</a> ) ]
 </pre><h3 id="rfc.section.3.2.1"><a href="#rfc.section.3.2.1">3.2.1</a>&nbsp;<a id="cache-request-directive" href="#cache-request-directive">Request Cache-Control Directives</a></h3>
       <div id="rfc.figure.u.10"></div> <pre class="inline"><span id="rfc.iref.g.7"></span>  <a href="#header.cache-control" class="smpl">cache-request-directive</a> =
+      <div id="rfc.figure.u.10"></div><pre class="inline"><span id="rfc.iref.g.7"></span>  <a href="#header.cache-control" class="smpl">cache-request-directive</a> =
        "no-cache"
      / "no-store"
 …
      / "only-if-cached"
      / <a href="#header.cache-control" class="smpl">cache-extension</a>
 </pre> <p id="rfc.section.3.2.1.p.2"> <span id="rfc.iref.c.4"></span>  <span id="rfc.iref.n.1"></span> no-cache
+</pre><p id="rfc.section.3.2.1.p.2"> <span id="rfc.iref.c.4"></span>  <span id="rfc.iref.n.1"></span> no-cache
       </p>
       <dl class="empty">
 …
       </dl>
       <h3 id="rfc.section.3.2.2"><a href="#rfc.section.3.2.2">3.2.2</a>&nbsp;<a id="cache-response-directive" href="#cache-response-directive">Response Cache-Control Directives</a></h3>
       <div id="rfc.figure.u.11"></div> <pre class="inline"><span id="rfc.iref.g.8"></span>  <a href="#header.cache-control" class="smpl">cache-response-directive</a> =
+      <div id="rfc.figure.u.11"></div><pre class="inline"><span id="rfc.iref.g.8"></span>  <a href="#header.cache-control" class="smpl">cache-response-directive</a> =
        "public"
      / "private" [ "=" <a href="#notation" class="smpl">DQUOTE</a> 1#<a href="#abnf.dependencies" class="smpl">field-name</a> <a href="#notation" class="smpl">DQUOTE</a> ]
 …
      / "s-maxage" "=" <a href="#rule.delta-seconds" class="smpl">delta-seconds</a>
      / <a href="#header.cache-control" class="smpl">cache-extension</a>
 </pre> <p id="rfc.section.3.2.2.p.2"> <span id="rfc.iref.c.11"></span>  <span id="rfc.iref.p.1"></span> public
+</pre><p id="rfc.section.3.2.2.p.2"> <span id="rfc.iref.c.11"></span>  <span id="rfc.iref.p.1"></span> public
       </p>
       <dl class="empty">
 …
          otherwise private response in their shared cache(s) could do so by including
       </p>
       <div id="rfc.figure.u.12"></div> <pre class="text">  Cache-Control: private, community="UCI"
 </pre> <p id="rfc.section.3.2.3.p.5">A cache seeing this header field will act correctly even if the cache does not understand the community cache-extension, since
+      <div id="rfc.figure.u.12"></div><pre class="text">  Cache-Control: private, community="UCI"
+</pre><p id="rfc.section.3.2.3.p.5">A cache seeing this header field will act correctly even if the cache does not understand the community cache-extension, since
          it will also see and understand the private directive and thus default to the safe behavior.
       </p>
 …
       <div id="rfc.figure.u.13"></div><pre class="inline"><span id="rfc.iref.g.9"></span><span id="rfc.iref.g.10"></span>  <a href="#header.expires" class="smpl">Expires</a>   = "Expires" ":" <a href="#core.rules" class="smpl">OWS</a> <a href="#header.expires" class="smpl">Expires-v</a>
   <a href="#header.expires" class="smpl">Expires-v</a> = <a href="#abnf.dependencies" class="smpl">HTTP-date</a>
+</pre><p id="rfc.section.3.3.p.5">For example</p>
+      <div id="rfc.figure.u.14"></div> <pre class="text">  Expires: Thu, 01 Dec 1994 16:00:00 GMT
+</pre> <p id="rfc.section.3.3.p.7"> </p>
+      <dl class="empty">
+         <dd> <b>Note:</b> if a response includes a Cache-Control field with the max-age directive (see <a href="#cache-response-directive" title="Response Cache-Control Directives">Section&nbsp;3.2.2</a>), that directive overrides the Expires field. Likewise, the s-maxage directive overrides Expires in shared caches.
+         </dd>
+      </dl>
+      <p id="rfc.section.3.3.p.8">HTTP/1.1 servers <em class="bcp14">SHOULD NOT</em> send Expires dates more than one year in the future.
+      </p>
+      <p id="rfc.section.3.3.p.9">HTTP/1.1 clients and caches <em class="bcp14">MUST</em> treat other invalid date formats, especially including the value "0", as in the past (i.e., "already expired").
+</pre><div id="rfc.figure.u.14"></div>
+      <p>For example</p>  <pre class="text">  Expires: Thu, 01 Dec 1994 16:00:00 GMT
+</pre><div class="note">
+         <p> <b>Note:</b> if a response includes a Cache-Control field with the max-age directive (see <a href="#cache-response-directive" title="Response Cache-Control Directives">Section&nbsp;3.2.2</a>), that directive overrides the Expires field. Likewise, the s-maxage directive overrides Expires in shared caches.
+         </p>
+      </div>
+      <p id="rfc.section.3.3.p.7">HTTP/1.1 servers <em class="bcp14">SHOULD NOT</em> send Expires dates more than one year in the future.
+      </p>
+      <p id="rfc.section.3.3.p.8">HTTP/1.1 clients and caches <em class="bcp14">MUST</em> treat other invalid date formats, especially including the value "0", as in the past (i.e., "already expired").
       </p>
       <div id="rfc.iref.p.4"></div>
 …
          has the same semantics as the no-cache response directive (see <a href="#cache-response-directive" title="Response Cache-Control Directives">Section&nbsp;3.2.2</a>) and is defined here for backward compatibility with HTTP/1.0. Clients <em class="bcp14">SHOULD</em> include both header fields when a no-cache request is sent to a server not known to be HTTP/1.1 compliant. HTTP/1.1 caches <em class="bcp14">SHOULD</em> treat "Pragma: no-cache" as if the client had sent "Cache-Control: no-cache".
       </p>
+      <p id="rfc.section.3.4.p.4"> </p>
+      <dl class="empty">
+         <dd> <b>Note:</b> because the meaning of "Pragma: no-cache" as a response-header field is not actually specified, it does not provide a reliable
+      <div class="note">
+         <p> <b>Note:</b> because the meaning of "Pragma: no-cache" as a response-header field is not actually specified, it does not provide a reliable
             replacement for "Cache-Control: no-cache" in a response.
          </dd>
       </dl>
+         </p>
+      </div>
       <p id="rfc.section.3.4.p.5">This mechanism is deprecated; no new Pragma directives will be defined in HTTP.</p>
       <div id="rfc.iref.v.3"></div>

draft-ietf-httpbis/latest/p6-cache.xml

-                      r539
+                      r540
 <?xml version="1.0" encoding="utf-8"?>
 <?xml-stylesheet type='text/xsl' href='../myxml2rfc.xslt'?>
+<?xml-stylesheet type='text/xsl' href='rfc2629.xslt'?>
 <!DOCTYPE rfc [
   <!ENTITY MAY "<bcp14 xmlns='http://purl.org/net/xml2rfc/ext'>MAY</bcp14>">
 …
 <?rfc-ext allow-markup-in-artwork="yes" ?>
 <?rfc-ext include-references-in-index="yes" ?>
-<?oxygen RNGSchema="../../rfc2629xslt/rfc2629-ext.rnc" type="compact"?>
 <rfc category="std" docName="draft-ietf-httpbis-p6-cache-&ID-VERSION;" ipr="pre5378Trust200902"
   obsoletes="2616" x:maturity-level="draft" xmlns:x="http://purl.org/net/xml2rfc/ext">
 …
     <workgroup>HTTPbis Working Group</workgroup>
+    <abstract>
+      <t>The Hypertext Transfer Protocol (HTTP) is an application-level protocol for distributed,
+        collaborative, hypermedia information systems. This document is Part 6 of the seven-part
+        specification that defines the protocol referred to as "HTTP/1.1" and, taken together,
+        obsoletes RFC 2616. Part 6 defines requirements on HTTP caches and the associated header
+        fields that control cache behavior or indicate cacheable response messages.</t>
+    </abstract>
+    <note title="Editorial Note (To be removed by RFC Editor)">
+      <t>Discussion of this draft should take place on the HTTPBIS working group mailing list
+        (ietf-http-wg@w3.org). The current issues list is at <eref
+          target="http://tools.ietf.org/wg/httpbis/trac/report/11" /> and related documents
+        (including fancy diffs) can be found at <eref target="http://tools.ietf.org/wg/httpbis/" />.</t>
+      <t>The changes in this draft are summarized in <xref target="changes.since.05" />.</t>
+    </note>
+<abstract>
+<t>
+  The Hypertext Transfer Protocol (HTTP) is an application-level protocol for distributed,
+  collaborative, hypermedia information systems. This document is Part 6 of the seven-part
+  specification that defines the protocol referred to as "HTTP/1.1" and, taken together,
+  obsoletes RFC 2616. Part 6 defines requirements on HTTP caches and the associated header
+  fields that control cache behavior or indicate cacheable response messages.
+</t>
+</abstract>
+<note title="Editorial Note (To be removed by RFC Editor)">
+<t>
+  Discussion of this draft should take place on the HTTPBIS working group mailing list
+  (ietf-http-wg@w3.org). The current issues list is at <eref
+  target="http://tools.ietf.org/wg/httpbis/trac/report/11" /> and related documents
+  (including fancy diffs) can be found at <eref target="http://tools.ietf.org/wg/httpbis/" />.
+</t>
+<t>
+  The changes in this draft are summarized in <xref target="changes.since.05" />.
+</t>
+</note>
   </front>
   <middle>
+    <section anchor="caching" title="Introduction">
+      <t>HTTP is typically used for distributed information systems, where performance can be
+        improved by the use of response caches. This document defines aspects of HTTP/1.1 related to
+        caching and reusing response messages.</t>
+      <section anchor="intro.purpose" title="Purpose">
+        <iref item="cache" />
+        <t>An HTTP <x:dfn>cache</x:dfn> is a local store of response messages and the subsystem that
+          controls its message storage, retrieval, and deletion. A cache stores cacheable responses
+          in order to reduce the response time and network bandwidth consumption on future,
+          equivalent requests. Any client or server may include a cache, though a cache cannot be
+          used by a server that is acting as a tunnel.</t>
+        <t>Caching would be useless if it did not significantly improve performance. The goal of
+          caching in HTTP/1.1 is to reuse a prior response message to satisfy a current request. In
+          some cases, a stored response can be reused without the need for a network request,
+          reducing latency and network round-trips; a "freshness" mechanism is used for this purpose
+          (see <xref target="expiration.model" />). Even when a new request is required, it is often
+          possible to reuse all or parts of the payload of a prior response to satisfy the request,
+          thereby reducing network bandwidth usage; a "validation" mechanism is used for this
+          purpose (see <xref target="validation.model" />).</t>
+      </section>
+      <section anchor="intro.terminology" title="Terminology">
+        <t>This specification uses a number of terms to refer to the roles played by participants
+          in, and objects of, HTTP caching.</t>
+        <t>
+          <iref item="cacheable" />
+          <x:dfn>cacheable</x:dfn>
+          <list>
+            <t>A response is cacheable if a cache is allowed to store a copy of the response message
+              for use in answering subsequent requests. Even when a response is cacheable, there may
+              be additional constraints on whether a cache can use the cached copy to satisfy a
+              particular request.</t>
+          </list>
+        </t>
+        <t>
+          <iref item="explicit expiration time" />
+          <x:dfn>explicit expiration time</x:dfn>
+          <list>
+            <t>The time at which the origin server intends that an entity should no longer be
+              returned by a cache without further validation.</t>
+          </list>
+        </t>
+        <t>
+          <iref item="heuristic expiration time" />
+          <x:dfn>heuristic expiration time</x:dfn>
+          <list>
+            <t>An expiration time assigned by a cache when no explicit expiration time is
+            available.</t>
+          </list>
+        </t>
+        <t>
+          <iref item="age" />
+          <x:dfn>age</x:dfn>
+          <list>
+            <t>The age of a response is the time since it was sent by, or successfully validated
+              with, the origin server.</t>
+          </list>
+        </t>
+        <t>
+          <iref item="first-hand" />
+          <x:dfn>first-hand</x:dfn>
+          <list>
+            <t>A response is first-hand if the freshness model is not in use; i.e., its age is
+.</t>
+          </list>
+        </t>
+        <t>
+          <iref item="freshness lifetime" />
+          <x:dfn>freshness lifetime</x:dfn>
+          <list>
+            <t>The length of time between the generation of a response and its expiration time. </t>
+          </list>
+        </t>
+        <t>
+          <iref item="fresh" />
+          <x:dfn>fresh</x:dfn>
+          <list>
+            <t>A response is fresh if its age has not yet exceeded its freshness lifetime.</t>
+          </list>
+        </t>
+        <t>
+          <iref item="stale" />
+          <x:dfn>stale</x:dfn>
+          <list>
+            <t>A response is stale if its age has passed its freshness lifetime (either explicit or heuristic).</t>
+          </list>
+        </t>
+        <t>
+          <iref item="validator" />
+          <x:dfn>validator</x:dfn>
+          <list>
+            <t>A protocol element (e.g., an entity tag or a Last-Modified time) that is used to find
+              out whether a stored response is an equivalent copy of an entity.</t>
+          </list>
+        </t>
+        <t anchor="shared.and.non-shared.caches">
+          <iref item="validator" />
+          <x:dfn>shared cache</x:dfn>
+          <list>
+            <t>A cache that is accessible to more than one user. A non-shared cache is
+              dedicated to a single user.</t>
+          </list>
+        </t>
+      </section>
+      <section anchor="intro.requirements" title="Requirements">
+        <t>The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD
+          NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as
+          described in <xref target="RFC2119" />.</t>
+        <t>An implementation is not compliant if it fails to satisfy one or more of the &MUST;
+          or &REQUIRED; level requirements for the protocols it implements. An implementation
+          that satisfies all the &MUST; or &REQUIRED; level and all the &SHOULD; level
+          requirements for its protocols is said to be "unconditionally compliant"; one that
+          satisfies all the &MUST; level requirements but not all the &SHOULD; level
+          requirements for its protocols is said to be "conditionally compliant."</t>
+      </section>
+<section anchor="caching" title="Introduction">
+<t>
+  HTTP is typically used for distributed information systems, where performance can be
+  improved by the use of response caches. This document defines aspects of HTTP/1.1 related to
+  caching and reusing response messages.
+</t>
+<section anchor="intro.purpose" title="Purpose">
+<iref item="cache" />
+<t>
+  An HTTP <x:dfn>cache</x:dfn> is a local store of response messages and the subsystem that
+  controls its message storage, retrieval, and deletion. A cache stores cacheable responses
+  in order to reduce the response time and network bandwidth consumption on future,
+  equivalent requests. Any client or server may include a cache, though a cache cannot be
+  used by a server that is acting as a tunnel.
+</t>
+<t>
+  Caching would be useless if it did not significantly improve performance. The goal of
+  caching in HTTP/1.1 is to reuse a prior response message to satisfy a current request. In
+  some cases, a stored response can be reused without the need for a network request,
+  reducing latency and network round-trips; a "freshness" mechanism is used for this purpose
+  (see <xref target="expiration.model" />). Even when a new request is required, it is often
+  possible to reuse all or parts of the payload of a prior response to satisfy the request,
+  thereby reducing network bandwidth usage; a "validation" mechanism is used for this
+  purpose (see <xref target="validation.model" />).
+</t>
+</section>
+<section anchor="intro.terminology" title="Terminology">
+<t
+  >This specification uses a number of terms to refer to the roles played by participants
+  in, and objects of, HTTP caching.
+</t>
+<t>
+  <iref item="cacheable" />
+  <x:dfn>cacheable</x:dfn>
+  <list>
+    <t>A response is cacheable if a cache is allowed to store a copy of the response message
+      for use in answering subsequent requests. Even when a response is cacheable, there may
+      be additional constraints on whether a cache can use the cached copy to satisfy a
+      particular request.</t>
+  </list>
+</t>
+<t>
+  <iref item="explicit expiration time" />
+  <x:dfn>explicit expiration time</x:dfn>
+  <list>
+    <t>The time at which the origin server intends that an entity should no longer be
+      returned by a cache without further validation.</t>
+  </list>
+</t>
+<t>
+  <iref item="heuristic expiration time" />
+  <x:dfn>heuristic expiration time</x:dfn>
+  <list>
+    <t>An expiration time assigned by a cache when no explicit expiration time is
+    available.</t>
+  </list>
+</t>
+<t>
+  <iref item="age" />
+  <x:dfn>age</x:dfn>
+  <list>
+    <t>The age of a response is the time since it was sent by, or successfully validated
+      with, the origin server.</t>
+  </list>
+</t>
+<t>
+  <iref item="first-hand" />
+  <x:dfn>first-hand</x:dfn>
+  <list>
+    <t>A response is first-hand if the freshness model is not in use; i.e., its age is
+.</t>
+  </list>
+</t>
+<t>
+  <iref item="freshness lifetime" />
+  <x:dfn>freshness lifetime</x:dfn>
+  <list>
+    <t>The length of time between the generation of a response and its expiration time. </t>
+  </list>
+</t>
+<t>
+  <iref item="fresh" />
+  <x:dfn>fresh</x:dfn>
+  <list>
+    <t>A response is fresh if its age has not yet exceeded its freshness lifetime.</t>
+  </list>
+</t>
+<t>
+  <iref item="stale" />
+  <x:dfn>stale</x:dfn>
+  <list>
+    <t>A response is stale if its age has passed its freshness lifetime (either explicit or heuristic).</t>
+  </list>
+</t>
+<t>
+  <iref item="validator" />
+  <x:dfn>validator</x:dfn>
+  <list>
+    <t>A protocol element (e.g., an entity tag or a Last-Modified time) that is used to find
+      out whether a stored response is an equivalent copy of an entity.</t>
+  </list>
+</t>
+<t anchor="shared.and.non-shared.caches">
+  <iref item="validator" />
+  <x:dfn>shared cache</x:dfn>
+  <list>
+    <t>A cache that is accessible to more than one user. A non-shared cache is
+      dedicated to a single user.</t>
+  </list>
+</t>
+</section>
+<section anchor="intro.requirements" title="Requirements">
+<t>
+  The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD
+  NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as
+  described in <xref target="RFC2119" />.
+</t>
+<t>
+  An implementation is not compliant if it fails to satisfy one or more of the &MUST;
+  or &REQUIRED; level requirements for the protocols it implements. An implementation
+  that satisfies all the &MUST; or &REQUIRED; level and all the &SHOULD; level
+  requirements for its protocols is said to be "unconditionally compliant"; one that
+  satisfies all the &MUST; level requirements but not all the &SHOULD; level
+  requirements for its protocols is said to be "conditionally compliant."
+</t>
+</section>
 <section title="Syntax Notation" anchor="notation">
 …
 </section>
+    <section anchor="caching.overview" title="Cache Operation">
+      <section anchor="response.cacheability" title="Response Cacheability">
+        <t>A cache &MUST-NOT; store a response to any request, unless: <list style="symbols">
+                <t>The request method is defined as being cacheable, and</t>
+            <t>the "no-store" cache directive (see <xref target="header.cache-control" />) does not
+              appear in request or response headers, and</t>
+            <t>the "private" cache response directive (see <xref target="header.cache-control" />
+            does not appear in the response, if the cache is shared, and</t>
+            <t>the "Authorization" header (see &header-authorization;) does not appear in the request, if
+             the cache is shared (unless the "public" directive is present; see <xref
+                target="header.cache-control" />), and</t>
+            <t>the cache understands partial responses, if the response is partial or incomplete
+              (see <xref target="errors.or.incomplete.response.cache.behavior" />).</t>
+          </list>
+        </t>
+        <t>Note that in normal operation, most caches will not store a response that has neither a
+          cache validator nor an explicit expiration time, as such responses are not usually
+          useful to store. However, caches are not prohibited from storing such responses.</t>
+        <section anchor="errors.or.incomplete.response.cache.behavior"
+          title="Storing Partial and Incomplete Responses">
+          <t>A cache that receives an incomplete response (for example, with fewer bytes of data
+            than specified in a Content-Length header) can store the response, but &MUST;
+            treat it as a partial response &partial;. Partial responses
+            can be combined as described in &combining-byte-ranges;; the result might be a
+            full response or might still be partial. A cache &MUST-NOT; return a partial
+            response to a client without explicitly marking it as such using the 206 (Partial
+            Content) status code.</t>
+          <t>A cache that does not support the Range and Content-Range headers &MUST-NOT; store
+            incomplete or partial responses.</t>
+        </section>
+      </section>
+      <section anchor="constructing.responses.from.caches"
+        title="Constructing Responses from Caches">
+        <t>For a presented request, a cache &MUST-NOT; return a stored response, unless:
+          <list style="symbols">
+            <t>The presented Request-URI and that of the stored response match (see
+              <cref>TBD</cref>), and</t>
+            <t>the request method associated with the stored response allows it to be
+              used for the presented request, and</t>
+            <t>selecting request-headers nominated by the stored response (if any) match those presented (see <xref
+                target="caching.negotiated.responses" />), and</t>
+            <t>the presented request and stored response are free from directives that would prevent
+              its use (see <xref target="header.cache-control" /> and <xref target="header.pragma"
+             />), and</t>
+            <t>the stored response is either:
+            <list style="symbols">
+                <t>fresh (see <xref target="expiration.model" />), or</t>
+                <t>allowed to be served stale (see <xref target="serving.stale.responses" />), or</t>
+                <t>successfully validated (see <xref target="validation.model" />).</t>
+            </list></t>
+          </list>
+        </t>
+        <t><cref>TODO: define method cacheability for GET, HEAD and POST in p2-semantics.</cref></t>
+        <t>When a stored response is used to satisfy a request, caches &MUST; include a
+          single Age header field <xref target="header.age" /> in the response with a value equal to the stored response's
+          current_age; see <xref target="age.calculations" />.
+                <cref>DISCUSS: this currently includes successfully validated responses.</cref></t>
+        <t>Requests with methods that are unsafe (&safe-methods;) &MUST; be written through the cache to
+          the origin server; i.e., A cache must not reply to such a request before having forwarded the request and having received a
+   corresponding response.</t>
+                <t>Also, note that unsafe requests might invalidate already stored responses; see
+            <xref target="invalidation.after.updates.or.deletions" />.</t>
+        <t>Caches &MUST; use the most recent response (as determined by the Date header) when
+          more than one suitable response is stored. They can also forward a request with
+          "Cache-Control: max-age=0" or "Cache-Control: no-cache" to disambiguate which response to
+          use.</t>
+        <t>
+          <cref>TODO: end-to-end and hop-by-hop headers, non-modifiable headers removed; re-spec in
+            p1</cref>
+        </t>
+      </section>
+      <section anchor="expiration.model" title="Freshness Model">
+        <t>When a response is "fresh" in the cache, it can be used to satisfy subsequent
+          requests without contacting the origin server, thereby improving efficiency.</t>
+        <t>The primary mechanism for determining freshness is for an origin server to provide an
+          explicit expiration time in the future, using either the Expires header (<xref
+            target="header.expires" />) or the max-age response cache directive (<xref
+            target="cache-response-directive" />). Generally, origin servers will assign future
+          explicit expiration times to responses in the belief that the entity is not likely to
+          change in a semantically significant way before the expiration time is reached.</t>
+        <t>If an origin server wishes to force a cache to validate every request, it can
+          assign an explicit expiration time in the past. This means that the response is always
+          stale, so that caches should validate it before using it for subsequent requests. <cref>This wording may cause confusion, because the response may still be served stale.</cref></t>
+        <t>Since origin servers do not always provide explicit expiration times, HTTP caches may
+          also assign heuristic expiration times when they are not specified, employing algorithms that
+          use other header values (such as the Last-Modified time) to estimate a plausible
+          expiration time. The HTTP/1.1 specification does not provide specific algorithms, but does
+          impose worst-case constraints on their results.</t>
+        <t>The calculation to determine if a response is fresh is:</t>
+        <figure>
+          <artwork type="code">
+<section anchor="caching.overview" title="Cache Operation">
+<section anchor="response.cacheability" title="Response Cacheability">
+<t>
+  A cache &MUST-NOT; store a response to any request, unless:
+  <list style="symbols">
+    <t>The request method is defined as being cacheable, and</t>
+    <t>the "no-store" cache directive (see <xref target="header.cache-control" />) does not
+       appear in request or response headers, and</t>
+    <t>the "private" cache response directive (see <xref target="header.cache-control" />
+       does not appear in the response, if the cache is shared, and</t>
+    <t>the "Authorization" header (see &header-authorization;) does not appear in the request, if
+       the cache is shared (unless the "public" directive is present; see <xref
+       target="header.cache-control" />), and</t>
+    <t>the cache understands partial responses, if the response is partial or incomplete
+       (see <xref target="errors.or.incomplete.response.cache.behavior" />).</t>
+  </list>
+</t>
+<t>
+  Note that in normal operation, most caches will not store a response that has neither a
+  cache validator nor an explicit expiration time, as such responses are not usually
+  useful to store. However, caches are not prohibited from storing such responses.
+</t>
+<section anchor="errors.or.incomplete.response.cache.behavior" title="Storing Partial and Incomplete Responses">
+<t>
+  A cache that receives an incomplete response (for example, with fewer bytes of data
+  than specified in a Content-Length header) can store the response, but &MUST;
+  treat it as a partial response &partial;. Partial responses
+  can be combined as described in &combining-byte-ranges;; the result might be a
+  full response or might still be partial. A cache &MUST-NOT; return a partial
+  response to a client without explicitly marking it as such using the 206 (Partial
+  Content) status code.
+</t>
+<t>
+  A cache that does not support the Range and Content-Range headers &MUST-NOT; store
+  incomplete or partial responses.
+</t>
+</section>
+</section>
+<section anchor="constructing.responses.from.caches" title="Constructing Responses from Caches">
+<t>
+For a presented request, a cache &MUST-NOT; return a stored response, unless:
+  <list style="symbols">
+    <t>The presented Request-URI and that of the stored response match (see
+      <cref>TBD</cref>), and</t>
+    <t>the request method associated with the stored response allows it to be
+      used for the presented request, and</t>
+    <t>selecting request-headers nominated by the stored response (if any) match those presented (see <xref
+      target="caching.negotiated.responses" />), and</t>
+    <t>the presented request and stored response are free from directives that would prevent
+      its use (see <xref target="header.cache-control" /> and <xref target="header.pragma"/>),
+      and</t>
+    <t>the stored response is either:
+      <list style="symbols">
+        <t>fresh (see <xref target="expiration.model" />), or</t>
+        <t>allowed to be served stale (see <xref target="serving.stale.responses" />), or</t>
+        <t>successfully validated (see <xref target="validation.model" />).</t>
+      </list>
+    </t>
+  </list>
+</t>
+<t>
+  <cref>TODO: define method cacheability for GET, HEAD and POST in p2-semantics.</cref>
+</t>
+<t>
+  When a stored response is used to satisfy a request, caches &MUST; include a
+  single Age header field <xref target="header.age" /> in the response with a value equal to the stored response's
+  current_age; see <xref target="age.calculations" />.
+  <cref>DISCUSS: this currently includes successfully validated responses.</cref>
+</t>
+<t>
+  Requests with methods that are unsafe (&safe-methods;) &MUST; be written through the cache to
+  the origin server; i.e., A cache must not reply to such a request before having forwarded the request and having received a
+  corresponding response.
+</t>
+<t>
+  Also, note that unsafe requests might invalidate already stored responses; see
+  <xref target="invalidation.after.updates.or.deletions" />.
+</t>
+<t>
+  Caches &MUST; use the most recent response (as determined by the Date header) when
+  more than one suitable response is stored. They can also forward a request with
+  "Cache-Control: max-age=0" or "Cache-Control: no-cache" to disambiguate which response to
+  use.
+</t>
+<t>
+  <cref>TODO: end-to-end and hop-by-hop headers, non-modifiable headers removed; re-spec in p1</cref>
+</t>
+</section>
+<section anchor="expiration.model" title="Freshness Model">
+<t>
+  When a response is "fresh" in the cache, it can be used to satisfy subsequent
+  requests without contacting the origin server, thereby improving efficiency.
+</t>
+<t>
+  The primary mechanism for determining freshness is for an origin server to provide an
+  explicit expiration time in the future, using either the Expires header (<xref
+  target="header.expires" />) or the max-age response cache directive (<xref
+  target="cache-response-directive" />). Generally, origin servers will assign future
+  explicit expiration times to responses in the belief that the entity is not likely to
+  change in a semantically significant way before the expiration time is reached.
+</t>
+<t>
+  If an origin server wishes to force a cache to validate every request, it can
+  assign an explicit expiration time in the past. This means that the response is always
+  stale, so that caches should validate it before using it for subsequent requests.
+  <cref>This wording may cause confusion, because the response may still be served stale.</cref>
+</t>
+<t>
+  Since origin servers do not always provide explicit expiration times, HTTP caches may
+  also assign heuristic expiration times when they are not specified, employing algorithms that
+  use other header values (such as the Last-Modified time) to estimate a plausible
+  expiration time. The HTTP/1.1 specification does not provide specific algorithms, but does
+  impose worst-case constraints on their results.
+</t>
+<figure>
+<preamble>
+The calculation to determine if a response is fresh is:
+</preamble>
+<artwork type="code">
    response_is_fresh = (freshness_lifetime &gt; current_age)
 </artwork>
+        </figure>
+        <t>The freshness_lifetime is defined in <xref target="calculating.freshness.lifetime" />;
+          the current_age is defined in <xref target="age.calculations" />.</t>
+        <t>Additionally, clients may need to influence freshness calculation. They can do this using
+          several request cache directives, with the effect of either increasing or loosening
+          constraints on freshness. See <xref target="cache-request-directive" />.</t>
+        <t>
+          <cref>ISSUE: there are not requirements directly applying to cache-request-directives and
+            freshness.</cref>
+        </t>
+        <t>Note that freshness applies only to cache operation; it cannot be used to force a user agent
+        to refresh its display or reload a resource. See <xref target="history.lists" /> for an explanation of
+          the difference between caches and history mechanisms.</t>
+        <section anchor="calculating.freshness.lifetime" title="Calculating Freshness Lifetime">
+          <t>A cache can calculate the freshness lifetime (denoted as freshness_lifetime) of a
+            response by using the first match of: <list style="symbols">
+              <t>If the cache is shared and the s-maxage response cache directive (<xref
+                  target="cache-response-directive" />) is present, use its value, or</t>
+              <t>If the max-age response cache directive (<xref target="cache-response-directive"
+                 />) is present, use its value, or</t>
+              <t>If the Expires response header (<xref target="header.expires" />) is present, use
+                its value minus the value of the Date response header, or</t>
+              <t>Otherwise, no explicit expiration time is present in the response, but a heuristic
+                may be used; see <xref target="heuristic.freshness" />.</t>
+            </list>
+          </t>
+          <t>Note that this calculation is not vulnerable to clock skew, since all of the
+            information comes from the origin server.</t>
+          <section anchor="heuristic.freshness" title="Calculating Heuristic Freshness">
+            <t>If no explicit expiration time is present in a stored response that has a status code
+              of 200, 203, 206, 300, 301 or 410, a heuristic expiration time can be
+              calculated. Heuristics &MUST-NOT; be used for other response status codes. </t>
+            <t> When a heuristic is used to calculate freshness lifetime, the cache &SHOULD;
+              attach a Warning header with a 113 warn-code to the response if its current_age is
+              more than 24 hours and such a warning is not already present.</t>
+            <t>Also, if the response has a Last-Modified header (&header-last-modified;), the
+              heuristic expiration value &SHOULD; be no more than some fraction of the interval
+              since that time. A typical setting of this fraction might be 10%.</t>
+            <t>
+              <cref>REVIEW: took away HTTP/1.0 query string heuristic uncacheability.</cref>
+            </t>
+          </section>
+        </section>
+        <section anchor="age.calculations" title="Calculating Age">
+          <t>HTTP/1.1 uses the Age response-header to convey the estimated age of the response
+            message when obtained from a cache. The Age field value is the cache's estimate of the
+            amount of time since the response was generated or validated by the origin server. In
+            essence, the Age value is the sum of the time that the response has been resident in
+            each of the caches along the path from the origin server, plus the amount of time it has
+            been in transit along network paths.</t>
+          <t>The term "age_value" denotes the value of the Age header, in a form appropriate for
+            arithmetic operations.</t>
+          <t>HTTP/1.1 requires origin servers to send a Date header, if possible, with every
+            response, giving the time at which the response was generated (see &header-date;).
+            The term "date_value" denotes the value of the Date header, in a form appropriate for
+            arithmetic operations.</t>
+          <t>The term "now" means "the current value of the clock at the host performing the
+            calculation." Hosts that use HTTP, but especially hosts running origin servers and
+            caches, &SHOULD; use NTP <xref target="RFC1305" /> or some similar protocol to
+            synchronize their clocks to a globally accurate time standard.</t>
+          <t>A response's age can be calculated in two entirely independent ways: <list
+              style="numbers">
+              <t>now minus date_value, if the local clock is reasonably well synchronized to the
+                origin server's clock. If the result is negative, the result is replaced by zero.</t>
+              <t>age_value, if all of the caches along the response path implement HTTP/1.1.</t>
+            </list>
+          </t>
+          <t>These are combined as</t>
+          <figure>
+            <artwork type="code">
+</figure>
+<t>
+  The freshness_lifetime is defined in <xref target="calculating.freshness.lifetime" />;
+  the current_age is defined in <xref target="age.calculations" />.
+</t>
+<t>
+  Additionally, clients may need to influence freshness calculation. They can do this using
+  several request cache directives, with the effect of either increasing or loosening
+  constraints on freshness. See <xref target="cache-request-directive" />.
+</t>
+<t>
+  <cref>ISSUE: there are not requirements directly applying to cache-request-directives and
+  freshness.</cref>
+</t>
+<t>
+  Note that freshness applies only to cache operation; it cannot be used to force a user agent
+  to refresh its display or reload a resource. See <xref target="history.lists" /> for an explanation of
+  the difference between caches and history mechanisms.
+</t>
+<section anchor="calculating.freshness.lifetime" title="Calculating Freshness Lifetime">
+<t>
+  A cache can calculate the freshness lifetime (denoted as freshness_lifetime) of a
+  response by using the first match of:
+  <list style="symbols">
+    <t>If the cache is shared and the s-maxage response cache directive (<xref
+      target="cache-response-directive" />) is present, use its value, or</t>
+    <t>If the max-age response cache directive (<xref target="cache-response-directive"
+      />) is present, use its value, or</t>
+    <t>If the Expires response header (<xref target="header.expires" />) is present, use
+      its value minus the value of the Date response header, or</t>
+    <t>Otherwise, no explicit expiration time is present in the response, but a heuristic
+      may be used; see <xref target="heuristic.freshness" />.</t>
+  </list>
+</t>
+<t>
+  Note that this calculation is not vulnerable to clock skew, since all of the
+  information comes from the origin server.
+</t>
+<section anchor="heuristic.freshness" title="Calculating Heuristic Freshness">
+<t>
+  If no explicit expiration time is present in a stored response that has a status code
+  of 200, 203, 206, 300, 301 or 410, a heuristic expiration time can be
+  calculated. Heuristics &MUST-NOT; be used for other response status codes.
+</t>
+<t>
+  When a heuristic is used to calculate freshness lifetime, the cache &SHOULD;
+  attach a Warning header with a 113 warn-code to the response if its current_age is
+  more than 24 hours and such a warning is not already present.
+</t>
+<t>
+  Also, if the response has a Last-Modified header (&header-last-modified;), the
+  heuristic expiration value &SHOULD; be no more than some fraction of the interval
+  since that time. A typical setting of this fraction might be 10%.
+</t>
+<t>
+  <cref>REVIEW: took away HTTP/1.0 query string heuristic uncacheability.</cref>
+</t>
+</section>
+</section>
+<section anchor="age.calculations" title="Calculating Age">
+<t>
+  HTTP/1.1 uses the Age response-header to convey the estimated age of the response
+  message when obtained from a cache. The Age field value is the cache's estimate of the
+  amount of time since the response was generated or validated by the origin server. In
+  essence, the Age value is the sum of the time that the response has been resident in
+  each of the caches along the path from the origin server, plus the amount of time it has
+  been in transit along network paths.
+</t>
+<t>
+  The term "age_value" denotes the value of the Age header, in a form appropriate for
+  arithmetic operations.
+</t>
+<t>
+  HTTP/1.1 requires origin servers to send a Date header, if possible, with every
+  response, giving the time at which the response was generated (see &header-date;).
+  The term "date_value" denotes the value of the Date header, in a form appropriate for
+  arithmetic operations.
+</t>
+<t>
+  The term "now" means "the current value of the clock at the host performing the
+  calculation." Hosts that use HTTP, but especially hosts running origin servers and
+  caches, &SHOULD; use NTP <xref target="RFC1305" /> or some similar protocol to
+  synchronize their clocks to a globally accurate time standard.
+</t>
+<t>A response's age can be calculated in two entirely independent ways:
+  <list style="numbers">
+    <t>now minus date_value, if the local clock is reasonably well synchronized to the
+      origin server's clock. If the result is negative, the result is replaced by zero.</t>
+    <t>age_value, if all of the caches along the response path implement HTTP/1.1.</t>
+  </list>
+</t>
+<figure>
+<preamble>These are combined as</preamble>
+<artwork type="code">
     corrected_received_age = max(now - date_value, age_value)
 </artwork>
           </figure>
           <t>When an Age value is received, it &MUST; be interpreted relative to the time the
             request was initiated, not the time that the response was received.</t>
           <figure>
             <artwork type="code">
+</artwork></figure>
+<t>
+  When an Age value is received, it &MUST; be interpreted relative to the time the
+  request was initiated, not the time that the response was received.
+</t>
+<figure><artwork type="code">
    corrected_initial_age = corrected_received_age
                          + (now - request_time)
+</artwork>
+          </figure>
+          <t>where "request_time" is the time (according to the local clock) when the request that
+            elicited this response was sent.</t>
+          <t>The current_age of a stored response can then be calculated by adding the amount of
+            time (in seconds) since the stored response was last validated by the origin server to
+            the corrected_initial_age.</t>
+          <t>In summary:</t>
+          <figure>
+            <artwork type="code">
+   age_value     - Age header field-value received with the response
+   date_value    - Date header field-value received with the response
+   request_time  - local time when the cache made the request
+                   resulting in the stored response
+   response_time - local time when the cache received the response
+   now           - current local time
+   apparent_age = max(0, response_time - date_value);
+   corrected_received_age = max(apparent_age, age_value);
+   response_delay = response_time - request_time;
+   corrected_initial_age = corrected_received_age + response_delay;
+   resident_time = now - response_time;
+   current_age   = corrected_initial_age + resident_time;
+</artwork>
+          </figure>
+        </section>
+        <section anchor="serving.stale.responses" title="Serving Stale Responses">
+          <t>A "stale" response is one that either has explicit expiry information, or is allowed to
+            have heuristic expiry calculated, but is not fresh according to the calculations in
+              <xref target="expiration.model" />.</t>
+          <t>Caches &MUST-NOT; return a stale response if it is prohibited by an explicit
+            in-protocol directive (e.g., by a "no-store" or "no-cache" cache directive, a
+            "must-revalidate" cache-response-directive, or an applicable "s-maxage" or
+            "proxy-revalidate" cache-response-directive; see <xref target="cache-response-directive"
+             />). </t>
+          <t>Caches &SHOULD-NOT; return stale responses unless they are
+          disconnected (i.e., it cannot contact the origin server or otherwise find a forward path)
+          or otherwise explicitly allowed (e.g., the max-stale request directive; see <xref target="cache-request-directive" />)..</t>
+          <t>Stale responses &SHOULD; have a Warning header with the 110 warn-code (see <xref
+              target="header.warning" />). Likewise, the 112 warn-code &SHOULD; be sent on stale responses if
+              the cache is disconnected.</t>
+          <t>If a cache receives a first-hand response (either an entire response, or a 304 (Not
+            Modified) response) that it would normally forward to the requesting client, and the
+            received response is no longer fresh, the cache &SHOULD; forward it to the
+            requesting client without adding a new Warning (but without removing any existing
+            Warning headers). A cache &SHOULD-NOT; attempt to validate a response simply because
+            that response became stale in transit.</t>
+        </section>
+      </section>
+      <section anchor="validation.model" title="Validation Model">
+        <t>Checking with the origin server to see if a stale or otherwise unusable cached response
+        can be reused is called "validating" or "revalidating." Doing so potentially avoids
+        the overhead of retransmitting the response body when the stored response is valid.</t>
+        <t>HTTP's conditional request mechanism &conditional; is used for this purpose. When a stored
+          response includes one or more validators, such as the field values of an ETag or
+          Last-Modified header field, then a validating request &SHOULD; be made conditional
+          to those field values.</t>
+         <t>A 304 (Not Modified) response status code indicates that the stored
+          response can be updated and reused; see <xref target="combining.headers"/>.</t>
+        <t>If instead the cache receives a full response (i.e., one with a response body), it is used to satisfy the
+        request and replace the stored response. <cref>Should there be a requirement here?</cref></t>
+        <t>If a cache receives a 5xx response while attempting to validate a response, it &MAY;
+          either forward this response to the requesting client, or act as if the server failed to
+          respond. In the latter case, it &MAY; return a previously stored response (which &SHOULD; include the
+warn-code; see <xref target="header.warning"/>) unless the
+          stored response includes the "must-revalidate" cache directive (see <xref
+            target="serving.stale.responses" />).</t>
+      </section>
+      <section anchor="invalidation.after.updates.or.deletions"
+        title="Request Methods that Invalidate">
+        <t>Because unsafe methods (&safe-methods;) have the potential for changing state on the
+          origin server, intervening caches can use them to keep their contents
+          up-to-date.</t>
+        <t>The following HTTP methods &MUST; cause a cache to invalidate the Request-URI as well
+          as the Location and Content-Location headers (if present): <list style="symbols">
+            <t>PUT</t>
+            <t>DELETE</t>
+            <t>POST</t>
+          </list>
+        </t>
+        <t>An invalidation based on the URI in a Location or Content-Location header &MUST-NOT;
+          be performed if the host part of that URI differs from the host part in the Request-URI.
+          This helps prevent denial of service attacks.</t>
+        <t>
+          <cref>TODO: "host part" needs to be specified better.</cref>
+        </t>
+        <t>A cache that passes through requests for methods it does not understand &SHOULD;
+          invalidate the Request-URI.</t>
+        <t>Here, "invalidate" means that the cache will either remove all stored responses related
+          to the Request-URI, or will mark these as "invalid" and in need of a mandatory validation
+          before they can be returned in response to a subsequent request.</t>
+        <t>Note that this does not guarantee that all appropriate responses are invalidated. For
+          example, the request that caused the change at the origin server might not have gone
+          through the cache where a response is stored.</t>
+        <t>
+          <cref>TODO: specify that only successful (2xx, 3xx?) responses invalidate.</cref>
+        </t>
+      </section>
+      <section anchor="caching.negotiated.responses" title="Caching Negotiated Responses">
+        <t>Use of server-driven content negotiation (&server-driven-negotiation;) alters
+          the conditions under which a cache can use the response for subsequent
+          requests.</t>
+        <t>When a cache receives a request that can be satisfied by a stored response
+          that includes a Vary header field (<xref target="header.vary"/>), it &MUST-NOT; use that response unless
+          all of the selecting request-headers in the presented request match the corresponding
+          stored request-headers from the original request.</t>
+        <t>The selecting request-headers from two requests are defined to match if and only if the
+          selecting request-headers in the first request can be transformed to the selecting
+          request-headers in the second request by adding or removing linear white space
+          <cref>[ref]</cref> at places where this is allowed by the corresponding ABNF, and/or
+          combining multiple message-header fields with the same field name following the rules
+          about message headers in &message-headers;. <cref>DISCUSS: header-specific canonicalisation</cref></t>
+        <t>A Vary header field-value of "*" always fails to match, and subsequent requests to that
+          resource can only be properly interpreted by the origin server.</t>
+        <t>If no stored response matches, the cache &MAY; forward the presented request to the origin
+          server in a conditional request, and &SHOULD; include all ETags stored with
+          potentially suitable responses in an If-None-Match request header. If the server responds with 304 (Not Modified) and
+          includes an entity tag or Content-Location that indicates the entity to be used, that
+          cached response &MUST; be used to satisfy the presented request, and &SHOULD;
+          be used to update the corresponding stored response; see <xref target="combining.headers"/>.</t>
+        <t>If any of the stored responses contains only partial content, its entity-tag &SHOULD-NOT;
+          be included in the If-None-Match header field unless the request is for a range that would
+          be fully satisfied by that stored response.</t>
+        <t>If a cache receives a successful response whose Content-Location field matches that of an
+          existing stored response for the same Request-URI, whose entity-tag differs from that of
+          the existing stored response, and whose Date is more recent than that of the existing
+          response, the existing response &SHOULD-NOT; be returned in response to future
+          requests and &SHOULD; be deleted from the cache.<cref>DISCUSS: Not sure if this is necessary.</cref></t>
+      </section>
+      <section anchor="combining.headers" title="Combining Responses">
+        <t>When a cache receives a 304 (Not Modified) response or a 206 (Partial Content) response,
+          it needs to update the stored response with the new one, so that the updated response can
+          be sent to the client.</t>
+        <t>If the status code is 304 (Not Modified), the cache &SHOULD; use the stored entity-body as
+          the updated entity-body. If the status code is 206 (Partial Content) and the ETag or
+          Last-Modified headers match exactly, the cache &MAY; combine the stored entity-body in
+          the stored response with the updated entity-body received in the response and use the
+          result as the updated entity-body (see &combining-byte-ranges;).</t>
+        <t>The stored response headers are used for the updated response, except that <list
+            style="symbols">
+            <t>any stored Warning headers with warn-code 1xx (see <xref target="header.warning" />)
+              &MUST; be deleted from the stored response and the forwarded response.</t>
+            <t>any stored Warning headers with warn-code 2xx &MUST; be retained in the stored
+              response and the forwarded response.</t>
+            <t>any headers provided in the 304 or 206 response &MUST; replace the corresponding
+              headers from the stored response.</t>
+          </list>
+        </t>
+        <t>A cache &MUST; also replace any stored headers with corresponding headers received in the
+          incoming response, except for Warning headers as described immediately above. If a header
+          field-name in the incoming response matches more than one header in the stored response,
+          all such old headers &MUST; be replaced. It &MAY; store the combined
+        entity-body.</t>
+        <t><cref>ISSUE: discuss how to handle HEAD updates</cref></t>
+      </section>
+    </section>
+    <section anchor="header.fields" title="Header Field Definitions">
+      <t>This section defines the syntax and semantics of HTTP/1.1 header fields related to caching.</t>
+      <t>For entity-header fields, both sender and recipient refer to either the client or the
+        server, depending on who sends and who receives the entity.</t>
+      <section anchor="header.age" title="Age">
+        <iref item="Age header" primary="true" x:for-anchor="" />
+        <iref item="Headers" primary="true" subitem="Age" x:for-anchor="" />
+        <x:anchor-alias value="Age"/>
+        <x:anchor-alias value="Age-v"/>
+        <x:anchor-alias value="age-value"/>
+        <t>      The response-header field "Age" conveys the sender's estimate of the amount of time since
+          the response (or its validation) was generated at the origin server. Age values are
+          calculated as specified in <xref target="age.calculations" />.</t>
+</artwork></figure>
+<t>
+  where "request_time" is the time (according to the local clock) when the request that
+  elicited this response was sent.
+</t>
+<t>
+  The current_age of a stored response can then be calculated by adding the amount of
+  time (in seconds) since the stored response was last validated by the origin server to
+  the corrected_initial_age.
+</t>
+<t>
+  In summary:
+</t>
+<figure><artwork type="code">
+  age_value     - Age header field-value received with the response
+  date_value    - Date header field-value received with the response
+  request_time  - local time when the cache made the request
+                 resulting in the stored response
+  response_time - local time when the cache received the response
+  now           - current local time
+  apparent_age = max(0, response_time - date_value);
+  corrected_received_age = max(apparent_age, age_value);
+  response_delay = response_time - request_time;
+  corrected_initial_age = corrected_received_age + response_delay;
+  resident_time = now - response_time;
+  current_age   = corrected_initial_age + resident_time;
+</artwork></figure>
+</section>
+<section anchor="serving.stale.responses" title="Serving Stale Responses">
+<t>
+  A "stale" response is one that either has explicit expiry information, or is allowed to
+  have heuristic expiry calculated, but is not fresh according to the calculations in
+  <xref target="expiration.model" />.
+</t>
+<t>
+  Caches &MUST-NOT; return a stale response if it is prohibited by an explicit
+  in-protocol directive (e.g., by a "no-store" or "no-cache" cache directive, a
+  "must-revalidate" cache-response-directive, or an applicable "s-maxage" or
+  "proxy-revalidate" cache-response-directive; see <xref target="cache-response-directive"/>).
+</t>
+<t>
+  Caches &SHOULD-NOT; return stale responses unless they are
+  disconnected (i.e., it cannot contact the origin server or otherwise find a forward path)
+  or otherwise explicitly allowed (e.g., the max-stale request directive; see <xref target="cache-request-directive" />).
+</t>
+<t>
+  Stale responses &SHOULD; have a Warning header with the 110 warn-code (see <xref
+  target="header.warning" />). Likewise, the 112 warn-code &SHOULD; be sent on stale responses if
+  the cache is disconnected.
+</t>
+<t>
+  If a cache receives a first-hand response (either an entire response, or a 304 (Not
+  Modified) response) that it would normally forward to the requesting client, and the
+  received response is no longer fresh, the cache &SHOULD; forward it to the
+  requesting client without adding a new Warning (but without removing any existing
+  Warning headers). A cache &SHOULD-NOT; attempt to validate a response simply because
+  that response became stale in transit.
+</t>
+</section>
+</section>
+<section anchor="validation.model" title="Validation Model">
+<t>
+  Checking with the origin server to see if a stale or otherwise unusable cached response
+  can be reused is called "validating" or "revalidating." Doing so potentially avoids
+  the overhead of retransmitting the response body when the stored response is valid.
+</t>
+<t>
+  HTTP's conditional request mechanism &conditional; is used for this purpose. When a stored
+  response includes one or more validators, such as the field values of an ETag or
+  Last-Modified header field, then a validating request &SHOULD; be made conditional
+  to those field values.
+</t>
+<t>
+  A 304 (Not Modified) response status code indicates that the stored
+  response can be updated and reused; see <xref target="combining.headers"/>.
+</t>
+<t>
+  If instead the cache receives a full response (i.e., one with a response body), it is used to satisfy the
+  request and replace the stored response. <cref>Should there be a requirement here?</cref>
+</t>
+<t>
+  If a cache receives a 5xx response while attempting to validate a response, it &MAY;
+  either forward this response to the requesting client, or act as if the server failed to
+  respond. In the latter case, it &MAY; return a previously stored response (which &SHOULD; include the
+warn-code; see <xref target="header.warning"/>) unless the
+  stored response includes the "must-revalidate" cache directive (see <xref
+  target="serving.stale.responses" />).
+</t>
+</section>
+<section anchor="invalidation.after.updates.or.deletions" title="Request Methods that Invalidate">
+<t>
+  Because unsafe methods (&safe-methods;) have the potential for changing state on the
+  origin server, intervening caches can use them to keep their contents
+  up-to-date.
+</t>
+<t>
+  The following HTTP methods &MUST; cause a cache to invalidate the Request-URI as well
+  as the Location and Content-Location headers (if present):
+  <list style="symbols">
+    <t>PUT</t>
+    <t>DELETE</t>
+    <t>POST</t>
+  </list>
+</t>
+<t>
+  An invalidation based on the URI in a Location or Content-Location header &MUST-NOT;
+  be performed if the host part of that URI differs from the host part in the Request-URI.
+  This helps prevent denial of service attacks.
+</t>
+<t>
+  <cref>TODO: "host part" needs to be specified better.</cref>
+</t>
+<t>
+  A cache that passes through requests for methods it does not understand &SHOULD;
+  invalidate the Request-URI.
+</t>
+<t>
+  Here, "invalidate" means that the cache will either remove all stored responses related
+  to the Request-URI, or will mark these as "invalid" and in need of a mandatory validation
+  before they can be returned in response to a subsequent request.
+</t>
+<t>
+  Note that this does not guarantee that all appropriate responses are invalidated. For
+  example, the request that caused the change at the origin server might not have gone
+  through the cache where a response is stored.
+</t>
+<t>
+  <cref>TODO: specify that only successful (2xx, 3xx?) responses invalidate.</cref>
+</t>
+</section>
+<section anchor="caching.negotiated.responses" title="Caching Negotiated Responses">
+<t>
+  Use of server-driven content negotiation (&server-driven-negotiation;) alters
+  the conditions under which a cache can use the response for subsequent
+  requests.
+</t>
+<t>
+  When a cache receives a request that can be satisfied by a stored response
+  that includes a Vary header field (<xref target="header.vary"/>), it &MUST-NOT; use that response unless
+  all of the selecting request-headers in the presented request match the corresponding
+  stored request-headers from the original request.
+</t>
+<t>
+  The selecting request-headers from two requests are defined to match if and only if the
+  selecting request-headers in the first request can be transformed to the selecting
+  request-headers in the second request by adding or removing linear white space
+  <cref>[ref]</cref> at places where this is allowed by the corresponding ABNF, and/or
+  combining multiple message-header fields with the same field name following the rules
+  about message headers in &message-headers;. <cref>DISCUSS: header-specific canonicalisation</cref>
+</t>
+<t>
+  A Vary header field-value of "*" always fails to match, and subsequent requests to that
+  resource can only be properly interpreted by the origin server.
+</t>
+<t>
+  If no stored response matches, the cache &MAY; forward the presented request to the origin
+  server in a conditional request, and &SHOULD; include all ETags stored with
+  potentially suitable responses in an If-None-Match request header. If the server responds with 304 (Not Modified) and
+  includes an entity tag or Content-Location that indicates the entity to be used, that
+  cached response &MUST; be used to satisfy the presented request, and &SHOULD;
+  be used to update the corresponding stored response; see <xref target="combining.headers"/>.
+</t>
+<t>
+  If any of the stored responses contains only partial content, its entity-tag &SHOULD-NOT;
+  be included in the If-None-Match header field unless the request is for a range that would
+  be fully satisfied by that stored response.
+</t>
+<t>
+  If a cache receives a successful response whose Content-Location field matches that of an
+  existing stored response for the same Request-URI, whose entity-tag differs from that of
+  the existing stored response, and whose Date is more recent than that of the existing
+  response, the existing response &SHOULD-NOT; be returned in response to future
+  requests and &SHOULD; be deleted from the cache.<cref>DISCUSS: Not sure if this is necessary.</cref>
+</t>
+</section>
+<section anchor="combining.headers" title="Combining Responses">
+<t>
+  When a cache receives a 304 (Not Modified) response or a 206 (Partial Content) response,
+  it needs to update the stored response with the new one, so that the updated response can
+  be sent to the client.
+</t>
+<t>
+  If the status code is 304 (Not Modified), the cache &SHOULD; use the stored entity-body as
+  the updated entity-body. If the status code is 206 (Partial Content) and the ETag or
+  Last-Modified headers match exactly, the cache &MAY; combine the stored entity-body in
+  the stored response with the updated entity-body received in the response and use the
+  result as the updated entity-body (see &combining-byte-ranges;).
+</t>
+<t>
+  The stored response headers are used for the updated response, except that
+  <list style="symbols">
+    <t>any stored Warning headers with warn-code 1xx (see <xref target="header.warning" />)
+      &MUST; be deleted from the stored response and the forwarded response.</t>
+    <t>any stored Warning headers with warn-code 2xx &MUST; be retained in the stored
+      response and the forwarded response.</t>
+    <t>any headers provided in the 304 or 206 response &MUST; replace the corresponding
+      headers from the stored response.</t>
+  </list>
+</t>
+<t>
+  A cache &MUST; also replace any stored headers with corresponding headers received in the
+  incoming response, except for Warning headers as described immediately above. If a header
+  field-name in the incoming response matches more than one header in the stored response,
+  all such old headers &MUST; be replaced. It &MAY; store the combined
+  entity-body.
+</t>
+<t>
+  <cref>ISSUE: discuss how to handle HEAD updates</cref>
+</t>
+</section>
+</section>
+<section anchor="header.fields" title="Header Field Definitions">
+<t>This section defines the syntax and semantics of HTTP/1.1 header fields related to caching.</t>
+<t>For entity-header fields, both sender and recipient refer to either the client or the
+server, depending on who sends and who receives the entity.</t>
+<section anchor="header.age" title="Age">
+  <iref item="Age header" primary="true" x:for-anchor="" />
+  <iref item="Headers" primary="true" subitem="Age" x:for-anchor="" />
+  <x:anchor-alias value="Age"/>
+  <x:anchor-alias value="Age-v"/>
+  <x:anchor-alias value="age-value"/>
+<t>
+  The response-header field "Age" conveys the sender's estimate of the amount of time since
+  the response (or its validation) was generated at the origin server. Age values are
+  calculated as specified in <xref target="age.calculations" />.
+</t>
 <figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Age"/><iref primary="true" item="Grammar" subitem="Age-v"/>
   <x:ref>Age</x:ref>   = "Age" ":" <x:ref>OWS</x:ref> <x:ref>Age-v</x:ref>
   <x:ref>Age-v</x:ref> = <x:ref>delta-seconds</x:ref>
 </artwork></figure>
         <t anchor="rule.delta-seconds">
           <x:anchor-alias value="delta-seconds" /> Age field-values are non-negative decimal
           integers, representing time in seconds.</t>
         <figure>
           <artwork type="abnf2616"><iref item="Grammar" primary="true" subitem="delta-seconds" />
+<t anchor="rule.delta-seconds">
+  <x:anchor-alias value="delta-seconds" />
+  Age field-values are non-negative decimal integers, representing time in seconds.
+</t>
+<figure><artwork type="abnf2616"><iref item="Grammar" primary="true" subitem="delta-seconds" />
   <x:ref>delta-seconds</x:ref>  = 1*<x:ref>DIGIT</x:ref>
+</artwork>
+        </figure>
+        <t>If a cache receives a value larger than the largest positive integer it can represent, or
+          if any of its age calculations overflows, it &MUST; transmit an Age header with a
+          field-value of 2147483648 (2<x:sup>31</x:sup>). Caches &SHOULD; use an arithmetic type
+          of at least 31 bits of range.</t>
+        <t>The presence of an Age header field in a response implies that a response is not
+          first-hand. However, the converse is not true, since HTTP/1.0 caches may not implement the
+          Age header field.</t>
+      </section>
+      <section anchor="header.cache-control" title="Cache-Control">
+        <iref item="Cache-Control header" primary="true" x:for-anchor="" />
+        <iref item="Headers" primary="true" subitem="Cache-Control" x:for-anchor="" />
+        <x:anchor-alias value="Cache-Control"/>
+        <x:anchor-alias value="Cache-Control-v"/>
+        <x:anchor-alias value="cache-directive"/>
+        <x:anchor-alias value="cache-extension"/>
+        <x:anchor-alias value="cache-request-directive"/>
+        <x:anchor-alias value="cache-response-directive"/>
+        <t>The general-header field "Cache-Control" is used to specify directives that &MUST; be
+          obeyed by all caches along the request/response chain. The directives specify behavior
+          intended to prevent caches from adversely interfering with the request or response. Cache
+          directives are unidirectional in that the presence of a directive in a request does not
+          imply that the same directive is to be given in the response. <list>
+            <t>Note that HTTP/1.0 caches might not implement Cache-Control and might only implement
+              Pragma: no-cache (see <xref target="header.pragma" />).</t>
+          </list>
+        </t>
+        <t>Cache directives &MUST; be passed through by a proxy or gateway application,
+          regardless of their significance to that application, since the directives might be
+          applicable to all recipients along the request/response chain. It is not possible to
+          target a directive to a specific cache.</t>
+</artwork></figure>
+<t>
+  If a cache receives a value larger than the largest positive integer it can represent, or
+  if any of its age calculations overflows, it &MUST; transmit an Age header with a
+  field-value of 2147483648 (2<x:sup>31</x:sup>). Caches &SHOULD; use an arithmetic type
+  of at least 31 bits of range.
+</t>
+<t>
+  The presence of an Age header field in a response implies that a response is not
+  first-hand. However, the converse is not true, since HTTP/1.0 caches may not implement the
+  Age header field.
+</t>
+</section>
+<section anchor="header.cache-control" title="Cache-Control">
+  <iref item="Cache-Control header" primary="true" x:for-anchor="" />
+  <iref item="Headers" primary="true" subitem="Cache-Control" x:for-anchor="" />
+  <x:anchor-alias value="Cache-Control"/>
+  <x:anchor-alias value="Cache-Control-v"/>
+  <x:anchor-alias value="cache-directive"/>
+  <x:anchor-alias value="cache-extension"/>
+  <x:anchor-alias value="cache-request-directive"/>
+  <x:anchor-alias value="cache-response-directive"/>
+<t>
+  The general-header field "Cache-Control" is used to specify directives that &MUST; be
+  obeyed by all caches along the request/response chain. The directives specify behavior
+  intended to prevent caches from adversely interfering with the request or response. Cache
+  directives are unidirectional in that the presence of a directive in a request does not
+  imply that the same directive is to be given in the response.
+</t>
+<x:note>
+  <t>Note that HTTP/1.0 caches might not implement Cache-Control and might only implement
+  Pragma: no-cache (see <xref target="header.pragma" />).</t>
+</x:note>
+<t>
+  Cache directives &MUST; be passed through by a proxy or gateway application,
+  regardless of their significance to that application, since the directives might be
+  applicable to all recipients along the request/response chain. It is not possible to
+  target a directive to a specific cache.
+</t>
 <figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Cache-Control"/><iref primary="true" item="Grammar" subitem="Cache-Control-v"/><iref primary="true" item="Grammar" subitem="cache-extension"/>
   <x:ref>Cache-Control</x:ref>   = "Cache-Control" ":" <x:ref>OWS</x:ref> <x:ref>Cache-Control-v</x:ref>
 …
 </artwork></figure>
+        <section anchor="cache-request-directive" title="Request Cache-Control Directives">
+          <x:anchor-alias value="cache-request-directive" />
+          <figure>
+            <artwork type="abnf2616"><iref item="Grammar" primary="true" subitem="cache-request-directive" />
+<section anchor="cache-request-directive" title="Request Cache-Control Directives">
+  <x:anchor-alias value="cache-request-directive" />
+<figure><artwork type="abnf2616"><iref item="Grammar" primary="true" subitem="cache-request-directive" />
   <x:ref>cache-request-directive</x:ref> =
        "no-cache"
 …
      / "only-if-cached"
      / <x:ref>cache-extension</x:ref>
+</artwork>
+          </figure>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="no-cache" />
+            <iref item="no-cache" primary="true" subitem="Cache Directive" /> no-cache <list>
+              <t>The no-cache request directive indicates that a stored response &MUST-NOT; be
+                used to satisfy the request without successful validation on the origin server. </t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="no-store" />
+            <iref item="no-store" primary="true" subitem="Cache Directive" /> no-store <list>
+              <t>The no-store request directive indicates that a cache &MUST-NOT; store any part
+                of either this request or any response to it. This directive applies to both
+                non-shared and shared caches. "&MUST-NOT; store" in this context means that the
+                cache &MUST-NOT; intentionally store the information in non-volatile storage,
+                and &MUST; make a best-effort attempt to remove the information from volatile
+                storage as promptly as possible after forwarding it.</t>
+              <t>This directive is NOT a reliable or sufficient mechanism for ensuring privacy. In
+                particular, malicious or compromised caches might not recognize or obey this
+                directive, and communications networks may be vulnerable to eavesdropping.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="max-age" />
+            <iref item="max-age" primary="true" subitem="Cache Directive" /> max-age <list>
+              <t>The max-age request directive indicates that the client is willing to accept a
+                response whose age is no greater than the specified time in seconds. Unless
+                max-stale directive is also included, the client is not willing to accept a stale
+                response.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="max-stale" />
+            <iref item="max-stale" primary="true" subitem="Cache Directive" /> max-stale <list>
+              <t>The max-stale request directive indicates that the client is willing to accept a
+                response that has exceeded its expiration time. If max-stale is assigned a value,
+                then the client is willing to accept a response that has exceeded its expiration
+                time by no more than the specified number of seconds. If no value is assigned to
+                max-stale, then the client is willing to accept a stale response of any age. <cref source="mnot">of any staleness?</cref></t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="min-fresh" />
+            <iref item="min-fresh" primary="true" subitem="Cache Directive" /> min-fresh <list>
+              <t>The min-fresh request directive indicates that the client is willing to accept a
+                response whose freshness lifetime is no less than its current age plus the specified
+                time in seconds. That is, the client wants a response that will still be fresh for
+                at least the specified number of seconds.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="no-transform" />
+            <iref item="no-transform" primary="true" subitem="Cache Directive" /> no-transform <list>
+              <t>The no-transform request directive indicates that an intermediate cache or proxy
+                &MUST-NOT; change the Content-Encoding, Content-Range or Content-Type request
+                headers, nor the request entity-body.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="only-if-cached" />
+            <iref item="only-if-cached" primary="true" subitem="Cache Directive" /> only-if-cached <list>
+              <t>The only-if-cached request directive indicates that the client only wishes to
+                return a stored response. If it receives this directive, a cache &SHOULD; either
+                respond using a stored response that is consistent with the other constraints of the
+                request, or respond with a 504 (Gateway Timeout) status. If a group of caches is
+                being operated as a unified system with good internal connectivity, such a request
+                &MAY; be forwarded within that group of caches.</t>
+            </list>
+          </t>
+        </section>
+        <section anchor="cache-response-directive" title="Response Cache-Control Directives">
+          <x:anchor-alias value="cache-response-directive" />
+          <figure>
+            <artwork type="abnf2616"><iref item="Grammar" primary="true" subitem="cache-response-directive" />
+</artwork></figure>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="no-cache" />
+  <iref item="no-cache" primary="true" subitem="Cache Directive" />
+  no-cache
+  <list>
+    <t>The no-cache request directive indicates that a stored response &MUST-NOT; be
+      used to satisfy the request without successful validation on the origin server.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="no-store" />
+  <iref item="no-store" primary="true" subitem="Cache Directive" />
+  no-store
+  <list>
+    <t>The no-store request directive indicates that a cache &MUST-NOT; store any part
+      of either this request or any response to it. This directive applies to both
+      non-shared and shared caches. "&MUST-NOT; store" in this context means that the
+      cache &MUST-NOT; intentionally store the information in non-volatile storage,
+      and &MUST; make a best-effort attempt to remove the information from volatile
+      storage as promptly as possible after forwarding it.</t>
+    <t>This directive is NOT a reliable or sufficient mechanism for ensuring privacy. In
+      particular, malicious or compromised caches might not recognize or obey this
+      directive, and communications networks may be vulnerable to eavesdropping.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="max-age" />
+  <iref item="max-age" primary="true" subitem="Cache Directive" />
+  max-age
+  <list>
+    <t>The max-age request directive indicates that the client is willing to accept a
+      response whose age is no greater than the specified time in seconds. Unless
+      max-stale directive is also included, the client is not willing to accept a stale
+      response.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="max-stale" />
+  <iref item="max-stale" primary="true" subitem="Cache Directive" />
+  max-stale
+  <list>
+    <t>The max-stale request directive indicates that the client is willing to accept a
+      response that has exceeded its expiration time. If max-stale is assigned a value,
+      then the client is willing to accept a response that has exceeded its expiration
+      time by no more than the specified number of seconds. If no value is assigned to
+      max-stale, then the client is willing to accept a stale response of any age. <cref source="mnot">of any staleness?</cref></t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="min-fresh" />
+  <iref item="min-fresh" primary="true" subitem="Cache Directive" />
+  min-fresh
+  <list>
+    <t>The min-fresh request directive indicates that the client is willing to accept a
+      response whose freshness lifetime is no less than its current age plus the specified
+      time in seconds. That is, the client wants a response that will still be fresh for
+      at least the specified number of seconds.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="no-transform" />
+  <iref item="no-transform" primary="true" subitem="Cache Directive" />
+  no-transform
+  <list>
+    <t>The no-transform request directive indicates that an intermediate cache or proxy
+      &MUST-NOT; change the Content-Encoding, Content-Range or Content-Type request
+      headers, nor the request entity-body.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="only-if-cached" />
+  <iref item="only-if-cached" primary="true" subitem="Cache Directive" />
+  only-if-cached
+  <list>
+    <t>The only-if-cached request directive indicates that the client only wishes to
+      return a stored response. If it receives this directive, a cache &SHOULD; either
+      respond using a stored response that is consistent with the other constraints of the
+      request, or respond with a 504 (Gateway Timeout) status. If a group of caches is
+      being operated as a unified system with good internal connectivity, such a request
+      &MAY; be forwarded within that group of caches.</t>
+  </list>
+</t>
+</section>
+<section anchor="cache-response-directive" title="Response Cache-Control Directives">
+  <x:anchor-alias value="cache-response-directive" />
+<figure><artwork type="abnf2616"><iref item="Grammar" primary="true" subitem="cache-response-directive" />
   <x:ref>cache-response-directive</x:ref> =
        "public"
 …
      / "s-maxage" "=" <x:ref>delta-seconds</x:ref>
      / <x:ref>cache-extension</x:ref>
+</artwork>
+          </figure>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="public" />
+            <iref item="public" primary="true" subitem="Cache Directive" /> public <list>
+              <t>The public response directive indicates that the response &MAY; be cached, even
+                if it would normally be non-cacheable or cacheable only within a non-shared cache.
+                (See also Authorization, &header-authorization;, for additional details.) </t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="private" />
+            <iref item="private" primary="true" subitem="Cache Directive" /> private <list>
+              <t>The private response directive indicates that the response message is intended for
+                a single user and &MUST-NOT; be stored by a shared cache. A private (non-shared)
+                cache &MAY; store the response.</t>
+              <t>If the private response directive specifies one or more field-names, this
+                requirement is limited to the field-values associated with the listed response
+                headers. That is, the specified field-names(s) &MUST-NOT; be stored by a shared
+                cache, whereas the remainder of the response message &MAY; be.</t>
+              <t>
+                <x:h>Note:</x:h> This usage of the word private only controls where the response may
+                be stored, and cannot ensure the privacy of the message content.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="no-cache" />
+            <iref item="no-cache" primary="true" subitem="Cache Directive" /> no-cache <list>
+              <t>The no-cache response directive indicates that the response &MUST-NOT; be used to
+                satisfy a subsequent request without successful validation on the origin server.
+                This allows an origin server to prevent caching even by caches that have been
+                configured to return stale responses.</t>
+              <t>If the no-cache response directive specifies one or more field-names, this
+                requirement is limited to the field-values assosicated with the listed response
+                headers. That is, the specified field-name(s) &MUST-NOT; be sent in the response
+                to a subsequent request without successful validation on the origin server. This
+                allows an origin server to prevent the re-use of certain header fields in a
+                response, while still allowing caching of the rest of the response.</t>
+              <t>
+                <x:h>Note:</x:h> Most HTTP/1.0 caches will not recognize or obey this directive.
+              </t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="no-store" />
+            <iref item="no-store" primary="true" subitem="Cache Directive" /> no-store <list>
+              <t>The no-store response directive indicates that a cache &MUST-NOT; store any
+                part of either the immediate request or response. This directive applies to both
+                non-shared and shared caches. "&MUST-NOT; store" in this context means that the
+                cache &MUST-NOT; intentionally store the information in non-volatile storage,
+                and &MUST; make a best-effort attempt to remove the information from volatile
+                storage as promptly as possible after forwarding it.</t>
+              <t>This directive is NOT a reliable or sufficient mechanism for ensuring privacy. In
+                particular, malicious or compromised caches might not recognize or obey this
+                directive, and communications networks may be vulnerable to eavesdropping.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="must-revalidate" />
+            <iref item="must-revalidate" primary="true" subitem="Cache Directive" /> must-revalidate <list>
+              <t>The must-revalidate response directive indicates that once it has become stale, the response &MUST-NOT; be
+               used to satisfy subsequent requests without successful validation on the origin server.</t>
+              <t>The must-revalidate directive is necessary to support reliable operation for
+                certain protocol features. In all circumstances an HTTP/1.1 cache &MUST; obey
+                the must-revalidate directive; in particular, if the cache cannot reach the origin
+                server for any reason, it &MUST; generate a 504 (Gateway Timeout) response.</t>
+              <t>Servers &SHOULD; send the must-revalidate directive if and only if failure to
+                validate a request on the entity could result in incorrect operation, such as a
+                silently unexecuted financial transaction.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="proxy-revalidate" />
+            <iref item="proxy-revalidate" primary="true" subitem="Cache Directive" />
+            proxy-revalidate <list>
+              <t>The proxy-revalidate response directive has the same meaning as the must-revalidate
+                response directive, except that it does not apply to non-shared caches.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="max-age" />
+            <iref item="max-age" primary="true" subitem="Cache Directive" /> max-age <list>
+              <t>The max-age response directive indicates that response is to be considered stale
+                after its age is greater than the specified number of seconds.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="s-maxage" />
+            <iref item="s-maxage" primary="true" subitem="Cache Directive" /> s-maxage <list>
+              <t>The s-maxage response directive indicates that, in shared caches, the maximum age
+                specified by this directive overrides the maximum age specified by either the
+                max-age directive or the Expires header. The s-maxage directive also implies the
+                semantics of the proxy-revalidate response directive.</t>
+            </list>
+          </t>
+          <t>
+            <iref item="Cache Directives" primary="true" subitem="no-transform" />
+            <iref item="no-transform" primary="true" subitem="Cache Directive" /> no-transform <list>
+              <t>The no-transform response directive indicates that an intermediate cache or proxy
+                &MUST-NOT; change the Content-Encoding, Content-Range or Content-Type response
+                headers, nor the response entity-body.</t>
+            </list>
+          </t>
+        </section>
+        <section anchor="cache.control.extensions" title="Cache Control Extensions">
+          <t>The Cache-Control header field can be extended through the use of one or more
+            cache-extension tokens, each with an optional value. Informational extensions (those
+            that do not require a change in cache behavior) can be added without changing the
+            semantics of other directives. Behavioral extensions are designed to work by acting as
+            modifiers to the existing base of cache directives. Both the new directive and the
+            standard directive are supplied, such that applications that do not understand the new
+            directive will default to the behavior specified by the standard directive, and those
+            that understand the new directive will recognize it as modifying the requirements
+            associated with the standard directive. In this way, extensions to the cache-control
+            directives can be made without requiring changes to the base protocol.</t>
+          <t>This extension mechanism depends on an HTTP cache obeying all of the cache-control
+            directives defined for its native HTTP-version, obeying certain extensions, and ignoring
+            all directives that it does not understand.</t>
+          <t>For example, consider a hypothetical new response directive called "community" that
+            acts as a modifier to the private directive. We define this new directive to mean that,
+            in addition to any non-shared cache, any cache that is shared only by members of the
+            community named within its value may cache the response. An origin server wishing to
+            allow the UCI community to use an otherwise private response in their shared cache(s)
+            could do so by including</t>
+          <figure>
+            <artwork type="example">
+</artwork></figure>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="public" />
+  <iref item="public" primary="true" subitem="Cache Directive" />
+  public
+  <list>
+    <t>The public response directive indicates that the response &MAY; be cached, even
+      if it would normally be non-cacheable or cacheable only within a non-shared cache.
+      (See also Authorization, &header-authorization;, for additional details.) </t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="private" />
+  <iref item="private" primary="true" subitem="Cache Directive" />
+  private
+  <list>
+    <t>The private response directive indicates that the response message is intended for
+      a single user and &MUST-NOT; be stored by a shared cache. A private (non-shared)
+      cache &MAY; store the response.</t>
+    <t>If the private response directive specifies one or more field-names, this
+      requirement is limited to the field-values associated with the listed response
+      headers. That is, the specified field-names(s) &MUST-NOT; be stored by a shared
+      cache, whereas the remainder of the response message &MAY; be.</t>
+    <t>
+      <x:h>Note:</x:h> This usage of the word private only controls where the response may
+      be stored, and cannot ensure the privacy of the message content.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="no-cache" />
+  <iref item="no-cache" primary="true" subitem="Cache Directive" />
+  no-cache
+  <list>
+    <t>The no-cache response directive indicates that the response &MUST-NOT; be used to
+      satisfy a subsequent request without successful validation on the origin server.
+      This allows an origin server to prevent caching even by caches that have been
+      configured to return stale responses.</t>
+    <t>If the no-cache response directive specifies one or more field-names, this
+      requirement is limited to the field-values assosicated with the listed response
+      headers. That is, the specified field-name(s) &MUST-NOT; be sent in the response
+      to a subsequent request without successful validation on the origin server. This
+      allows an origin server to prevent the re-use of certain header fields in a
+      response, while still allowing caching of the rest of the response.</t>
+    <t>
+      <x:h>Note:</x:h> Most HTTP/1.0 caches will not recognize or obey this directive.
+    </t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="no-store" />
+  <iref item="no-store" primary="true" subitem="Cache Directive" />
+  no-store
+  <list>
+    <t>The no-store response directive indicates that a cache &MUST-NOT; store any
+      part of either the immediate request or response. This directive applies to both
+      non-shared and shared caches. "&MUST-NOT; store" in this context means that the
+      cache &MUST-NOT; intentionally store the information in non-volatile storage,
+      and &MUST; make a best-effort attempt to remove the information from volatile
+      storage as promptly as possible after forwarding it.</t>
+    <t>This directive is NOT a reliable or sufficient mechanism for ensuring privacy. In
+      particular, malicious or compromised caches might not recognize or obey this
+      directive, and communications networks may be vulnerable to eavesdropping.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="must-revalidate" />
+  <iref item="must-revalidate" primary="true" subitem="Cache Directive" />
+  must-revalidate
+  <list>
+    <t>The must-revalidate response directive indicates that once it has become stale, the response &MUST-NOT; be
+     used to satisfy subsequent requests without successful validation on the origin server.</t>
+    <t>The must-revalidate directive is necessary to support reliable operation for
+      certain protocol features. In all circumstances an HTTP/1.1 cache &MUST; obey
+      the must-revalidate directive; in particular, if the cache cannot reach the origin
+      server for any reason, it &MUST; generate a 504 (Gateway Timeout) response.</t>
+    <t>Servers &SHOULD; send the must-revalidate directive if and only if failure to
+      validate a request on the entity could result in incorrect operation, such as a
+      silently unexecuted financial transaction.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="proxy-revalidate" />
+  <iref item="proxy-revalidate" primary="true" subitem="Cache Directive" />
+  proxy-revalidate
+  <list>
+    <t>The proxy-revalidate response directive has the same meaning as the must-revalidate
+      response directive, except that it does not apply to non-shared caches.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="max-age" />
+  <iref item="max-age" primary="true" subitem="Cache Directive" />
+  max-age
+  <list>
+    <t>The max-age response directive indicates that response is to be considered stale
+      after its age is greater than the specified number of seconds.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="s-maxage" />
+  <iref item="s-maxage" primary="true" subitem="Cache Directive" />
+  s-maxage
+  <list>
+    <t>The s-maxage response directive indicates that, in shared caches, the maximum age
+      specified by this directive overrides the maximum age specified by either the
+      max-age directive or the Expires header. The s-maxage directive also implies the
+      semantics of the proxy-revalidate response directive.</t>
+  </list>
+</t>
+<t>
+  <iref item="Cache Directives" primary="true" subitem="no-transform" />
+  <iref item="no-transform" primary="true" subitem="Cache Directive" />
+  no-transform
+  <list>
+    <t>The no-transform response directive indicates that an intermediate cache or proxy
+      &MUST-NOT; change the Content-Encoding, Content-Range or Content-Type response
+      headers, nor the response entity-body.</t>
+  </list>
+</t>
+</section>
+<section anchor="cache.control.extensions" title="Cache Control Extensions">
+<t>
+  The Cache-Control header field can be extended through the use of one or more
+  cache-extension tokens, each with an optional value. Informational extensions (those
+  that do not require a change in cache behavior) can be added without changing the
+  semantics of other directives. Behavioral extensions are designed to work by acting as
+  modifiers to the existing base of cache directives. Both the new directive and the
+  standard directive are supplied, such that applications that do not understand the new
+  directive will default to the behavior specified by the standard directive, and those
+  that understand the new directive will recognize it as modifying the requirements
+  associated with the standard directive. In this way, extensions to the cache-control
+  directives can be made without requiring changes to the base protocol.
+</t>
+<t>
+  This extension mechanism depends on an HTTP cache obeying all of the cache-control
+  directives defined for its native HTTP-version, obeying certain extensions, and ignoring
+  all directives that it does not understand.
+</t>
+<t>
+  For example, consider a hypothetical new response directive called "community" that
+  acts as a modifier to the private directive. We define this new directive to mean that,
+  in addition to any non-shared cache, any cache that is shared only by members of the
+  community named within its value may cache the response. An origin server wishing to
+  allow the UCI community to use an otherwise private response in their shared cache(s)
+  could do so by including
+</t>
+<figure><artwork type="example">
   Cache-Control: private, community="UCI"
+</artwork>
+          </figure>
+          <t>A cache seeing this header field will act correctly even if the cache does not
+            understand the community cache-extension, since it will also see and understand the
+            private directive and thus default to the safe behavior.</t>
+          <t>Unrecognized cache directives &MUST; be ignored; it is assumed that any cache
+            directive likely to be unrecognized by an HTTP/1.1 cache will be combined with standard
+            directives (or the response's default cacheability) such that the cache behavior will
+            remain minimally correct even if the cache does not understand the extension(s).</t>
+        </section>
+      </section>
+      <section anchor="header.expires" title="Expires">
+        <iref item="Expires header" primary="true" x:for-anchor="" />
+        <iref item="Headers" primary="true" subitem="Expires" x:for-anchor="" />
+        <x:anchor-alias value="Expires"/>
+        <x:anchor-alias value="Expires-v"/>
+        <t>The entity-header field "Expires" gives the date/time after which the response is
+          considered stale. See <xref target="expiration.model" /> for further discussion of the
+          freshness model.</t>
+        <t>The presence of an Expires field does not imply that the original resource will change or
+          cease to exist at, before, or after that time.</t>
+        <t>The field-value is an absolute date and time as defined by HTTP-date in &full-date;;
+          it &MUST; be sent in rfc1123-date format.</t>
+</artwork></figure>
+<t>
+  A cache seeing this header field will act correctly even if the cache does not
+  understand the community cache-extension, since it will also see and understand the
+  private directive and thus default to the safe behavior.
+</t>
+<t>
+  Unrecognized cache directives &MUST; be ignored; it is assumed that any cache
+  directive likely to be unrecognized by an HTTP/1.1 cache will be combined with standard
+  directives (or the response's default cacheability) such that the cache behavior will
+  remain minimally correct even if the cache does not understand the extension(s).
+</t>
+</section>
+</section>
+<section anchor="header.expires" title="Expires">
+  <iref item="Expires header" primary="true" x:for-anchor="" />
+  <iref item="Headers" primary="true" subitem="Expires" x:for-anchor="" />
+  <x:anchor-alias value="Expires"/>
+  <x:anchor-alias value="Expires-v"/>
+<t>
+  The entity-header field "Expires" gives the date/time after which the response is
+  considered stale. See <xref target="expiration.model" /> for further discussion of the
+  freshness model.
+</t>
+<t>
+  The presence of an Expires field does not imply that the original resource will change or
+  cease to exist at, before, or after that time.
+</t>
+<t>
+  The field-value is an absolute date and time as defined by HTTP-date in &full-date;;
+  it &MUST; be sent in rfc1123-date format.
+</t>
 <figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Expires"/><iref primary="true" item="Grammar" subitem="Expires-v"/>
   <x:ref>Expires</x:ref>   = "Expires" ":" <x:ref>OWS</x:ref> <x:ref>Expires-v</x:ref>
   <x:ref>Expires-v</x:ref> = <x:ref>HTTP-date</x:ref>
 </artwork></figure>
         <t>For example</t>
         <figure>
           <artwork type="example">
+<figure>
+  <preamble>For example</preamble>
+<artwork type="example">
   Expires: Thu, 01 Dec 1994 16:00:00 GMT
+</artwork>
+        </figure>
+        <t>
+          <list>
+            <t>
+              <x:h>Note:</x:h> if a response includes a Cache-Control field with the max-age
+              directive (see <xref target="cache-response-directive" />), that directive overrides
+              the Expires field. Likewise, the s-maxage directive overrides Expires in shared caches.</t>
+          </list>
+        </t>
+        <t>HTTP/1.1 servers &SHOULD-NOT; send Expires dates more than one year in the future.</t>
+        <t>HTTP/1.1 clients and caches &MUST; treat other invalid date formats, especially
+          including the value "0", as in the past (i.e., "already expired").</t>
+      </section>
+      <section anchor="header.pragma" title="Pragma">
+        <iref item="Pragma header" primary="true" x:for-anchor="" />
+        <iref item="Headers" primary="true" subitem="Pragma" x:for-anchor="" />
+        <x:anchor-alias value="extension-pragma"/>
+        <x:anchor-alias value="Pragma"/>
+        <x:anchor-alias value="Pragma-v"/>
+        <x:anchor-alias value="pragma-directive"/>
+        <t>The general-header field "Pragma" is used to include implementation-specific directives
+          that might apply to any recipient along the request/response chain. All pragma directives
+          specify optional behavior from the viewpoint of the protocol; however, some systems
+          &MAY; require that behavior be consistent with the directives.</t>
+</artwork></figure>
+<x:note>
+    <t>
+      <x:h>Note:</x:h> if a response includes a Cache-Control field with the max-age
+      directive (see <xref target="cache-response-directive" />), that directive overrides
+      the Expires field. Likewise, the s-maxage directive overrides Expires in shared caches.</t>
+</x:note>
+<t>
+  HTTP/1.1 servers &SHOULD-NOT; send Expires dates more than one year in the future.
+</t>
+<t>
+  HTTP/1.1 clients and caches &MUST; treat other invalid date formats, especially
+  including the value "0", as in the past (i.e., "already expired").
+</t>
+</section>
+<section anchor="header.pragma" title="Pragma">
+  <iref item="Pragma header" primary="true" x:for-anchor="" />
+  <iref item="Headers" primary="true" subitem="Pragma" x:for-anchor="" />
+  <x:anchor-alias value="extension-pragma"/>
+  <x:anchor-alias value="Pragma"/>
+  <x:anchor-alias value="Pragma-v"/>
+  <x:anchor-alias value="pragma-directive"/>
+<t
+  >The general-header field "Pragma" is used to include implementation-specific directives
+  that might apply to any recipient along the request/response chain. All pragma directives
+  specify optional behavior from the viewpoint of the protocol; however, some systems
+  &MAY; require that behavior be consistent with the directives.
+</t>
 <figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Pragma"/><iref primary="true" item="Grammar" subitem="Pragma-v"/><iref primary="true" item="Grammar" subitem="pragma-directive"/><iref primary="true" item="Grammar" subitem="extension-pragma"/>
   <x:ref>Pragma</x:ref>            = "Pragma" ":" <x:ref>OWS</x:ref> <x:ref>Pragma-v</x:ref>
 …
   <x:ref>extension-pragma</x:ref>  = <x:ref>token</x:ref> [ "=" ( <x:ref>token</x:ref> / <x:ref>quoted-string</x:ref> ) ]
 </artwork></figure>
+        <t>When the no-cache directive is present in a request message, an application &SHOULD;
+          forward the request toward the origin server even if it has a cached copy of what is being
+          requested. This pragma directive has the same semantics as the no-cache response directive
+          (see <xref target="cache-response-directive" />) and is defined here for backward
+          compatibility with HTTP/1.0. Clients &SHOULD; include both header fields when a
+          no-cache request is sent to a server not known to be HTTP/1.1 compliant. HTTP/1.1 caches
+          &SHOULD; treat "Pragma: no-cache" as if the client had sent "Cache-Control: no-cache".</t>
+        <t>
+          <list>
+            <t>
+              <x:h>Note:</x:h> because the meaning of "Pragma: no-cache" as a response-header field
+              is not actually specified, it does not provide a reliable replacement for
+              "Cache-Control: no-cache" in a response.</t>
+          </list>
+        </t>
+        <t>This mechanism is deprecated; no new Pragma directives will be defined in HTTP.</t>
+      </section>
+      <section anchor="header.vary" title="Vary">
+        <iref item="Vary header" primary="true" x:for-anchor="" />
+        <iref item="Headers" primary="true" subitem="Vary" x:for-anchor="" />
+        <x:anchor-alias value="Vary"/>
+        <x:anchor-alias value="Vary-v"/>
+        <t>The "Vary" response-header field's value indicates the set of request-header fields that
+          determines, while the response is fresh, whether a cache is permitted to use the
+          response to reply to a subsequent request without validation; see <xref
+          target="caching.negotiated.responses" />.</t>
+        <t>In uncacheable or stale responses, the Vary field value advises the user agent about
+          the criteria that were used to select the representation.</t>
+<t>
+  When the no-cache directive is present in a request message, an application &SHOULD;
+  forward the request toward the origin server even if it has a cached copy of what is being
+  requested. This pragma directive has the same semantics as the no-cache response directive
+  (see <xref target="cache-response-directive" />) and is defined here for backward
+  compatibility with HTTP/1.0. Clients &SHOULD; include both header fields when a
+  no-cache request is sent to a server not known to be HTTP/1.1 compliant. HTTP/1.1 caches
+  &SHOULD; treat "Pragma: no-cache" as if the client had sent "Cache-Control: no-cache".
+</t>
+<x:note>
+  <t>
+    <x:h>Note:</x:h> because the meaning of "Pragma: no-cache" as a response-header field
+    is not actually specified, it does not provide a reliable replacement for
+    "Cache-Control: no-cache" in a response.</t>
+</x:note>
+<t>
+  This mechanism is deprecated; no new Pragma directives will be defined in HTTP.
+</t>
+</section>
+<section anchor="header.vary" title="Vary">
+  <iref item="Vary header" primary="true" x:for-anchor="" />
+  <iref item="Headers" primary="true" subitem="Vary" x:for-anchor="" />
+  <x:anchor-alias value="Vary"/>
+  <x:anchor-alias value="Vary-v"/>
+<t>
+  The "Vary" response-header field's value indicates the set of request-header fields that
+  determines, while the response is fresh, whether a cache is permitted to use the
+  response to reply to a subsequent request without validation; see <xref
+  target="caching.negotiated.responses" />.
+</t>
+<t>
+  In uncacheable or stale responses, the Vary field value advises the user agent about
+  the criteria that were used to select the representation.
+</t>
 <figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Vary"/><iref primary="true" item="Grammar" subitem="Vary-v"/>
   <x:ref>Vary</x:ref>   = "Vary" ":" <x:ref>OWS</x:ref> <x:ref>Vary-v</x:ref>
   <x:ref>Vary-v</x:ref> = "*" / 1#<x:ref>field-name</x:ref>
 </artwork></figure>
+        <t>The set of header fields named by the Vary field value is known as the selecting
+          request-headers.</t>
+        <t>Servers &SHOULD; include a Vary header field with any cacheable response that is
+          subject to server-driven negotiation. Doing so allows a cache to properly interpret future
+          requests on that resource and informs the user agent about the presence of negotiation on
+          that resource. A server &MAY; include a Vary header field with a non-cacheable
+          response that is subject to server-driven negotiation, since this might provide the user
+          agent with useful information about the dimensions over which the response varies at the
+          time of the response.</t>
+        <t>A Vary field value of "*" signals that unspecified parameters not limited to the
+          request-headers (e.g., the network address of the client), play a role in the selection of
+          the response representation; therefore, a cache cannot determine whether this response is
+          appropriate. The "*" value &MUST-NOT; be generated by a proxy server;
+          it may only be generated by an origin server.</t>
+        <t>The field-names given are not limited to the set of standard request-header fields
+          defined by this specification. Field names are case-insensitive.</t>
+      </section>
+      <section anchor="header.warning" title="Warning">
+        <iref item="Warning header" primary="true" x:for-anchor="" />
+        <iref item="Headers" primary="true" subitem="Warning" x:for-anchor="" />
+        <x:anchor-alias value="Warning"/>
+        <x:anchor-alias value="Warning-v"/>
+        <x:anchor-alias value="warning-value"/>
+        <x:anchor-alias value="warn-agent"/>
+        <x:anchor-alias value="warn-code"/>
+        <x:anchor-alias value="warn-date"/>
+        <x:anchor-alias value="warn-text"/>
+        <t>The general-header field "Warning" is used to carry additional information about the status
+          or transformation of a message that might not be reflected in the message. This
+          information is typically used to warn about possible incorrectness introduced by caching
+          operations or transformations applied to the entity body of the message.</t>
+        <t>Warnings can be used for other purposes, both cache-related and otherwise. The use of a
+          warning, rather than an error status code, distinguish these responses from true failures.</t>
+        <t>Warning headers can in general be applied to any message, however some warn-codes are
+          specific to caches and can only be applied to response messages.</t>
+<t>
+  The set of header fields named by the Vary field value is known as the selecting
+  request-headers.
+</t>
+<t>
+  Servers &SHOULD; include a Vary header field with any cacheable response that is
+  subject to server-driven negotiation. Doing so allows a cache to properly interpret future
+  requests on that resource and informs the user agent about the presence of negotiation on
+  that resource. A server &MAY; include a Vary header field with a non-cacheable
+  response that is subject to server-driven negotiation, since this might provide the user
+  agent with useful information about the dimensions over which the response varies at the
+  time of the response.
+</t>
+<t>
+  A Vary field value of "*" signals that unspecified parameters not limited to the
+  request-headers (e.g., the network address of the client), play a role in the selection of
+  the response representation; therefore, a cache cannot determine whether this response is
+  appropriate. The "*" value &MUST-NOT; be generated by a proxy server;
+  it may only be generated by an origin server.
+</t>
+<t>
+  The field-names given are not limited to the set of standard request-header fields
+  defined by this specification. Field names are case-insensitive.
+</t>
+</section>
+<section anchor="header.warning" title="Warning">
+  <iref item="Warning header" primary="true" x:for-anchor="" />
+  <iref item="Headers" primary="true" subitem="Warning" x:for-anchor="" />
+  <x:anchor-alias value="Warning"/>
+  <x:anchor-alias value="Warning-v"/>
+  <x:anchor-alias value="warning-value"/>
+  <x:anchor-alias value="warn-agent"/>
+  <x:anchor-alias value="warn-code"/>
+  <x:anchor-alias value="warn-date"/>
+  <x:anchor-alias value="warn-text"/>
+<t>
+  The general-header field "Warning" is used to carry additional information about the status
+  or transformation of a message that might not be reflected in the message. This
+  information is typically used to warn about possible incorrectness introduced by caching
+  operations or transformations applied to the entity body of the message.
+</t>
+<t>
+  Warnings can be used for other purposes, both cache-related and otherwise. The use of a
+  warning, rather than an error status code, distinguish these responses from true failures.
+</t>
+<t>
+  Warning headers can in general be applied to any message, however some warn-codes are
+  specific to caches and can only be applied to response messages.
+</t>
 <figure><artwork type="abnf2616"><iref primary="true" item="Grammar" subitem="Warning"/><iref primary="true" item="Grammar" subitem="Warning-v"/><iref primary="true" item="Grammar" subitem="warning-value"/><iref primary="true" item="Grammar" subitem="warn-code"/><iref primary="true" item="Grammar" subitem="warn-agent"/><iref primary="true" item="Grammar" subitem="warn-text"/><iref primary="true" item="Grammar" subitem="warn-date"/>
   <x:ref>Warning</x:ref>    = "Warning" ":" <x:ref>OWS</x:ref> <x:ref>Warning-v</x:ref>
 …
   <x:ref>warn-date</x:ref>  = <x:ref>DQUOTE</x:ref> <x:ref>HTTP-date</x:ref> <x:ref>DQUOTE</x:ref>
 </artwork></figure>
+        <t>Multiple warnings can be attached to a response (either by the origin server or by
+          a cache), including multiple warnings with the same code number. For example, a server
+          might provide the same warning with texts in both English and Basque.</t>
+        <t>When this occurs, the user agent &SHOULD; inform the user of as many of them as
+          possible, in the order that they appear in the response. If it is not possible to inform
+          the user of all of the warnings, the user agent &SHOULD; follow these heuristics:
+            <list style="symbols">
+            <t>Warnings that appear early in the response take priority over those appearing later
+              in the response.</t>
+            <t>Warnings in the user's preferred character set take priority over warnings in other
+              character sets but with identical warn-codes and warn-agents.</t>
+          </list>
+        </t>
+        <t>Systems that generate multiple Warning headers &SHOULD; order them with this user
+          agent behavior in mind. New Warning headers &SHOULD; be added after any existing
+          Warning headers.</t>
+        <t>Warnings are assigned three digit warn-codes. The first digit indicates whether the
+          Warning is required to be deleted from a stored response after validation: <list
+            style="symbols">
+            <t>1xx Warnings that describe the freshness or validation status of the response, and so
+              &MUST; be deleted by caches after validation. They &MUST-NOT; be generated by a cache
+              except when validating a cached entry, and &MUST-NOT; be generated by clients.</t>
+            <t>2xx Warnings that describe some aspect of the entity body or entity headers that is
+              not rectified by a validation (for example, a lossy compression of the entity bodies)
+              and &MUST-NOT; be deleted by caches after validation, unless a full response is
+              returned, in which case they &MUST; be.</t>
+          </list>
+        </t>
+        <t>The warn-text &SHOULD; be in a natural language and character set that is most likely
+          to be intelligible to the human user receiving the response. This decision can be based on
+          any available knowledge, such as the location of the cache or user, the Accept-Language
+          field in a request, the Content-Language field in a response, etc. The default language is
+          English and the default character set is ISO-8859-1 (<xref target="ISO-8859-1" />).</t>
+        <t>If a character set other than ISO-8859-1 is used, it &MUST; be encoded in the
+          warn-text using the method described in <xref target="RFC2047" />.</t>
+        <t>If an implementation sends a message with one or more Warning headers to a receiver whose
+          version is HTTP/1.0 or lower, then the sender &MUST; include in each warning-value a
+          warn-date that matches the Date header in the message.</t>
+        <t>If an implementation receives a message with a warning-value that includes a warn-date,
+          and that warn-date is different from the Date value in the response, then that
+          warning-value &MUST; be deleted from the message before storing, forwarding, or using
+          it. (preventing the consequences of naive caching of Warning header fields.) If all of the
+          warning-values are deleted for this reason, the Warning header &MUST; be deleted as
+          well.</t>
+        <t>The following warn-codes are defined by this specification, each with a recommended
+          warn-text in English, and a description of its meaning.</t>
+        <t>110 Response is stale <list>
+            <t>&SHOULD; be included whenever the returned response is stale.</t>
+          </list>
+        </t>
+        <t>111 Revalidation failed <list>
+            <t>&SHOULD; be included if a cache returns a stale response because an attempt to
+              validate the response failed, due to an inability to reach the server.</t>
+          </list>
+        </t>
+        <t>112 Disconnected operation <list>
+            <t>&SHOULD; be included if the cache is intentionally disconnected from the rest of
+              the network for a period of time.</t>
+          </list>
+        </t>
+        <t>113 Heuristic expiration <list>
+            <t>&SHOULD; be included if the cache heuristically chose a freshness lifetime
+              greater than 24 hours and the response's age is greater than 24 hours.</t>
+          </list>
+        </t>
+        <t>199 Miscellaneous warning <list>
+            <t>The warning text can include arbitrary information to be presented to a human
+              user, or logged. A system receiving this warning &MUST-NOT; take any automated
+              action, besides presenting the warning to the user.</t>
+          </list>
+        </t>
+        <t>214 Transformation applied <list>
+            <t>&MUST; be added by an intermediate cache or proxy if it applies any
+              transformation changing the content-coding (as specified in the Content-Encoding
+              header) or media-type (as specified in the Content-Type header) of the response, or
+              the entity-body of the response, unless this Warning code already appears in the
+              response.</t>
+          </list>
+        </t>
+        <t>299 Miscellaneous persistent warning <list>
+            <t>The warning text can include arbitrary information to be presented to a human
+              user, or logged. A system receiving this warning &MUST-NOT; take any automated
+              action.</t>
+          </list>
+        </t>
+      </section>
+    </section>
+    <section anchor="history.lists" title="History Lists">
+      <t>User agents often have history mechanisms, such as "Back" buttons and history lists, that
+        can be used to redisplay an entity retrieved earlier in a session.</t>
+      <t>History mechanisms and caches are different. In particular history mechanisms
+        &SHOULD-NOT; try to show a correct view of the current state of a resource. Rather, a
+        history mechanism is meant to show exactly what the user saw at the time when the resource
+        was retrieved.</t>
+      <t>By default, an expiration time does not apply to history mechanisms. If the entity is still
+        in storage, a history mechanism &SHOULD; display it even if the entity has expired,
+        unless the user has specifically configured the agent to refresh expired history documents.</t>
+      <t>This is not to be construed to prohibit the history mechanism from telling the user that a
+        view might be stale. <list>
+          <t>
+            <x:h>Note:</x:h> if history list mechanisms unnecessarily prevent users from viewing
+            stale resources, this will tend to force service authors to avoid using HTTP expiration
+            controls and cache controls when they would otherwise like to. Service authors may
+            consider it important that users not be presented with error messages or warning
+            messages when they use navigation controls (such as BACK) to view previously fetched
+            resources. Even though sometimes such resources ought not be cached, or ought to expire
+            quickly, user interface considerations may force service authors to resort to other
+            means of preventing caching (e.g. "once-only" URLs) in order not to suffer the effects
+            of improperly functioning history mechanisms.</t>
+        </list>
+      </t>
+    </section>
+    <section anchor="IANA.considerations" title="IANA Considerations">
+      <section anchor="message.header.registration" title="Message Header Registration">
+        <t>The Message Header Registry located at <eref
+            target="http://www.iana.org/assignments/message-headers/message-header-index.html" />
+          should be updated with the permanent registrations below (see <xref target="RFC3864" />):</t>
+        <!--AUTOGENERATED FROM extract-header-defs.xslt, do not edit manually-->
+        <texttable align="left" anchor="iana.header.registration.table" suppress-title="true">
+          <ttcol>Header Field Name</ttcol>
+          <ttcol>Protocol</ttcol>
+          <ttcol>Status</ttcol>
+          <ttcol>Reference</ttcol>
+          <c>Age</c>
+          <c>http</c>
+          <c>standard</c>
+          <c>
+            <xref target="header.age" />
+          </c>
+          <c>Cache-Control</c>
+          <c>http</c>
+          <c>standard</c>
+          <c>
+            <xref target="header.cache-control" />
+          </c>
+          <c>Expires</c>
+          <c>http</c>
+          <c>standard</c>
+          <c>
+            <xref target="header.expires" />
+          </c>
+          <c>Pragma</c>
+          <c>http</c>
+          <c>standard</c>
+          <c>
+            <xref target="header.pragma" />
+          </c>
+          <c>Vary</c>
+          <c>http</c>
+          <c>standard</c>
+          <c>
+            <xref target="header.vary" />
+          </c>
+          <c>Warning</c>
+          <c>http</c>
+          <c>standard</c>
+          <c>
+            <xref target="header.warning" />
+          </c>
+        </texttable>
+        <!--(END)-->
+        <t>The change controller is: "IETF (iesg@ietf.org) - Internet Engineering Task Force".</t>
+      </section>
+    </section>
+    <section anchor="security.considerations" title="Security Considerations">
+      <t>Caches expose additional potential vulnerabilities, since the contents of the cache
+        represent an attractive target for malicious exploitation. Because cache contents persist
+        after an HTTP request is complete, an attack on the cache can reveal information long after
+        a user believes that the information has been removed from the network. Therefore, cache
+        contents should be protected as sensitive information.</t>
+    </section>
+    <section anchor="ack" title="Acknowledgments">
+      <t>Much of the content and presentation of the caching design is due to suggestions and
+        comments from individuals including: Shel Kaphan, Paul Leach, Koen Holtman, David Morris,
+        and Larry Masinter.</t>
+    </section>
+  </middle>
+  <back>
+    <references title="Normative References">
+      <reference anchor="ISO-8859-1">
+        <front>
+          <title> Information technology -- 8-bit single-byte coded graphic character sets -- Part
+: Latin alphabet No. 1 </title>
+          <author>
+            <organization>International Organization for Standardization</organization>
+          </author>
+          <date year="1998" />
+        </front>
+        <seriesInfo name="ISO/IEC" value="8859-1:1998" />
+      </reference>
+      <reference anchor="Part1">
+        <front>
+          <title abbrev="HTTP/1.1">HTTP/1.1, part 1: URIs, Connections, and Message Parsing</title>
+          <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+            <organization abbrev="Day Software">Day Software</organization>
+            <address><email>fielding@gbiv.com</email></address>
+          </author>
+          <author fullname="Jim Gettys" initials="J." surname="Gettys">
+            <organization>One Laptop per Child</organization>
+            <address><email>jg@laptop.org</email></address>
+          </author>
+          <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+            <organization abbrev="HP">Hewlett-Packard Company</organization>
+            <address><email>JeffMogul@acm.org</email></address>
+          </author>
+          <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>henrikn@microsoft.com</email></address>
+          </author>
+          <author fullname="Larry Masinter" initials="L." surname="Masinter">
+            <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+            <address><email>LMM@acm.org</email></address>
+          </author>
+          <author fullname="Paul J. Leach" initials="P." surname="Leach">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>paulle@microsoft.com</email></address>
+          </author>
+          <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+            <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+            <address><email>timbl@w3.org</email></address>
+          </author>
+          <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+            <organization abbrev="W3C">World Wide Web Consortium</organization>
+            <address><email>ylafon@w3.org</email></address>
+          </author>
+          <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+            <organization abbrev="greenbytes">greenbytes GmbH</organization>
+            <address><email>julian.reschke@greenbytes.de</email></address>
+          </author>
+          <date month="&ID-MONTH;" year="&ID-YEAR;" />
+        </front>
+        <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p1-messaging-&ID-VERSION;" />
+        <x:source basename="p1-messaging" href="p1-messaging.xml" />
+      </reference>
+      <reference anchor="Part2">
+        <front>
+          <title abbrev="HTTP/1.1">HTTP/1.1, part 2: Message Semantics</title>
+          <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+            <organization abbrev="Day Software">Day Software</organization>
+            <address><email>fielding@gbiv.com</email></address>
+          </author>
+          <author fullname="Jim Gettys" initials="J." surname="Gettys">
+            <organization>One Laptop per Child</organization>
+            <address><email>jg@laptop.org</email></address>
+          </author>
+          <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+            <organization abbrev="HP">Hewlett-Packard Company</organization>
+            <address><email>JeffMogul@acm.org</email></address>
+          </author>
+          <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>henrikn@microsoft.com</email></address>
+          </author>
+          <author fullname="Larry Masinter" initials="L." surname="Masinter">
+            <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+            <address><email>LMM@acm.org</email></address>
+          </author>
+          <author fullname="Paul J. Leach" initials="P." surname="Leach">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>paulle@microsoft.com</email></address>
+          </author>
+          <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+            <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+            <address><email>timbl@w3.org</email></address>
+          </author>
+          <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+            <organization abbrev="W3C">World Wide Web Consortium</organization>
+            <address><email>ylafon@w3.org</email></address>
+          </author>
+          <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+            <organization abbrev="greenbytes">greenbytes GmbH</organization>
+            <address><email>julian.reschke@greenbytes.de</email></address>
+          </author>
+          <date month="&ID-MONTH;" year="&ID-YEAR;" />
+        </front>
+        <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p2-semantics-&ID-VERSION;" />
+        <x:source basename="p2-semantics" href="p2-semantics.xml" />
+      </reference>
+      <reference anchor="Part3">
+        <front>
+          <title abbrev="HTTP/1.1">HTTP/1.1, part 3: Message Payload and Content Negotiation</title>
+          <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+            <organization abbrev="Day Software">Day Software</organization>
+            <address><email>fielding@gbiv.com</email></address>
+          </author>
+          <author fullname="Jim Gettys" initials="J." surname="Gettys">
+            <organization>One Laptop per Child</organization>
+            <address><email>jg@laptop.org</email></address>
+          </author>
+          <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+            <organization abbrev="HP">Hewlett-Packard Company</organization>
+            <address><email>JeffMogul@acm.org</email></address>
+          </author>
+          <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>henrikn@microsoft.com</email></address>
+          </author>
+          <author fullname="Larry Masinter" initials="L." surname="Masinter">
+            <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+            <address><email>LMM@acm.org</email></address>
+          </author>
+          <author fullname="Paul J. Leach" initials="P." surname="Leach">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>paulle@microsoft.com</email></address>
+          </author>
+          <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+            <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+            <address><email>timbl@w3.org</email></address>
+          </author>
+          <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+            <organization abbrev="W3C">World Wide Web Consortium</organization>
+            <address><email>ylafon@w3.org</email></address>
+          </author>
+          <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+            <organization abbrev="greenbytes">greenbytes GmbH</organization>
+            <address><email>julian.reschke@greenbytes.de</email></address>
+          </author>
+          <date month="&ID-MONTH;" year="&ID-YEAR;" />
+        </front>
+        <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p3-payload-&ID-VERSION;" />
+        <x:source basename="p3-payload" href="p3-payload.xml" />
+      </reference>
+      <reference anchor="Part4">
+        <front>
+          <title abbrev="HTTP/1.1">HTTP/1.1, part 4: Conditional Requests</title>
+          <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+            <organization abbrev="Day Software">Day Software</organization>
+            <address><email>fielding@gbiv.com</email></address>
+          </author>
+          <author fullname="Jim Gettys" initials="J." surname="Gettys">
+            <organization>One Laptop per Child</organization>
+            <address><email>jg@laptop.org</email></address>
+          </author>
+          <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+            <organization abbrev="HP">Hewlett-Packard Company</organization>
+            <address><email>JeffMogul@acm.org</email></address>
+          </author>
+          <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>henrikn@microsoft.com</email></address>
+          </author>
+          <author fullname="Larry Masinter" initials="L." surname="Masinter">
+            <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+            <address><email>LMM@acm.org</email></address>
+          </author>
+          <author fullname="Paul J. Leach" initials="P." surname="Leach">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>paulle@microsoft.com</email></address>
+          </author>
+          <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+            <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+            <address><email>timbl@w3.org</email></address>
+          </author>
+          <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+            <organization abbrev="W3C">World Wide Web Consortium</organization>
+            <address><email>ylafon@w3.org</email></address>
+          </author>
+          <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+            <organization abbrev="greenbytes">greenbytes GmbH</organization>
+            <address><email>julian.reschke@greenbytes.de</email></address>
+          </author>
+          <date month="&ID-MONTH;" year="&ID-YEAR;" />
+        </front>
+        <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p4-conditional-&ID-VERSION;" />
+        <x:source basename="p4-conditional" href="p4-conditional.xml" />
+      </reference>
+      <reference anchor="Part5">
+        <front>
+          <title abbrev="HTTP/1.1">HTTP/1.1, part 5: Range Requests and Partial Responses</title>
+          <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+            <organization abbrev="Day Software">Day Software</organization>
+            <address><email>fielding@gbiv.com</email></address>
+          </author>
+          <author fullname="Jim Gettys" initials="J." surname="Gettys">
+            <organization>One Laptop per Child</organization>
+            <address><email>jg@laptop.org</email></address>
+          </author>
+          <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+            <organization abbrev="HP">Hewlett-Packard Company</organization>
+            <address><email>JeffMogul@acm.org</email></address>
+          </author>
+          <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>henrikn@microsoft.com</email></address>
+          </author>
+          <author fullname="Larry Masinter" initials="L." surname="Masinter">
+            <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+            <address><email>LMM@acm.org</email></address>
+          </author>
+          <author fullname="Paul J. Leach" initials="P." surname="Leach">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>paulle@microsoft.com</email></address>
+          </author>
+          <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+            <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+            <address><email>timbl@w3.org</email></address>
+          </author>
+          <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+            <organization abbrev="W3C">World Wide Web Consortium</organization>
+            <address><email>ylafon@w3.org</email></address>
+          </author>
+          <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+            <organization abbrev="greenbytes">greenbytes GmbH</organization>
+            <address><email>julian.reschke@greenbytes.de</email></address>
+          </author>
+          <date month="&ID-MONTH;" year="&ID-YEAR;" />
+        </front>
+        <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p5-range-&ID-VERSION;" />
+        <x:source basename="p5-range" href="p5-range.xml" />
+      </reference>
+      <reference anchor="Part7">
+        <front>
+          <title abbrev="HTTP/1.1">HTTP/1.1, part 7: Authentication</title>
+          <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+            <organization abbrev="Day Software">Day Software</organization>
+            <address><email>fielding@gbiv.com</email></address>
+          </author>
+          <author fullname="Jim Gettys" initials="J." surname="Gettys">
+            <organization>One Laptop per Child</organization>
+            <address><email>jg@laptop.org</email></address>
+          </author>
+          <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+            <organization abbrev="HP">Hewlett-Packard Company</organization>
+            <address><email>JeffMogul@acm.org</email></address>
+          </author>
+          <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>henrikn@microsoft.com</email></address>
+          </author>
+          <author fullname="Larry Masinter" initials="L." surname="Masinter">
+            <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+            <address><email>LMM@acm.org</email></address>
+          </author>
+          <author fullname="Paul J. Leach" initials="P." surname="Leach">
+            <organization abbrev="Microsoft">Microsoft Corporation</organization>
+            <address><email>paulle@microsoft.com</email></address>
+          </author>
+          <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+            <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+            <address><email>timbl@w3.org</email></address>
+          </author>
+          <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+            <organization abbrev="W3C">World Wide Web Consortium</organization>
+            <address><email>ylafon@w3.org</email></address>
+          </author>
+          <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+            <organization abbrev="greenbytes">greenbytes GmbH</organization>
+            <address><email>julian.reschke@greenbytes.de</email></address>
+          </author>
+          <date month="&ID-MONTH;" year="&ID-YEAR;" />
+        </front>
+        <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p7-auth-&ID-VERSION;" />
+        <x:source basename="p7-auth" href="p7-auth.xml" />
+      </reference>
+      <reference anchor="RFC2047">
+        <front>
+          <title abbrev="Message Header Extensions">MIME (Multipurpose Internet Mail Extensions)
+            Part Three: Message Header Extensions for Non-ASCII Text</title>
+          <author fullname="Keith Moore" initials="K." surname="Moore">
+            <organization>University of Tennessee</organization>
+            <address><email>moore@cs.utk.edu</email></address>
+          </author>
+          <date month="November" year="1996" />
+        </front>
+        <seriesInfo name="RFC" value="2047" />
+      </reference>
+      <reference anchor="RFC2119">
+        <front>
+          <title>Key words for use in RFCs to Indicate Requirement Levels</title>
+          <author fullname="Scott Bradner" initials="S." surname="Bradner">
+            <organization>Harvard University</organization>
+            <address><email>sob@harvard.edu</email></address>
+          </author>
+          <date month="March" year="1997" />
+        </front>
+        <seriesInfo name="BCP" value="14" />
+        <seriesInfo name="RFC" value="2119" />
+      </reference>
+      <reference anchor="RFC5234">
+        <front>
+          <title abbrev="ABNF for Syntax Specifications">Augmented BNF for Syntax Specifications: ABNF</title>
+          <author initials="D." surname="Crocker" fullname="Dave Crocker" role="editor">
+            <organization>Brandenburg InternetWorking</organization>
+            <address>
+            <postal>
+            <street>675 Spruce Dr.</street>
+            <city>Sunnyvale</city>
+            <region>CA</region>
+            <code>94086</code>
+            <country>US</country></postal>
+            <phone>+1.408.246.8253</phone>
+            <email>dcrocker@bbiw.net</email></address>
+          </author>
+          <author initials="P." surname="Overell" fullname="Paul Overell">
+            <organization>THUS plc.</organization>
+            <address>
+            <postal>
+            <street>1/2 Berkeley Square</street>
+            <street>99 Berkely Street</street>
+            <city>Glasgow</city>
+            <code>G3 7HR</code>
+            <country>UK</country></postal>
+            <email>paul.overell@thus.net</email></address>
+          </author>
+          <date month="January" year="2008"/>
+        </front>
+        <seriesInfo name="STD" value="68"/>
+        <seriesInfo name="RFC" value="5234"/>
+      </reference>
+    </references>
+    <references title="Informative References">
+      <reference anchor="RFC1305">
+        <front>
+          <title>Network Time Protocol (Version 3) Specification, Implementation</title>
+          <author fullname="David L. Mills" initials="D." surname="Mills">
+            <organization>University of Delaware, Electrical Engineering Department</organization>
+            <address><email>mills@udel.edu</email></address>
+          </author>
+          <date month="March" year="1992" />
+        </front>
+        <seriesInfo name="RFC" value="1305" />
+      </reference>
+      <reference anchor="RFC2616">
+        <front>
+          <title>Hypertext Transfer Protocol -- HTTP/1.1</title>
+          <author fullname="R. Fielding" initials="R." surname="Fielding">
+            <organization>University of California, Irvine</organization>
+            <address><email>fielding@ics.uci.edu</email></address>
+          </author>
+          <author fullname="J. Gettys" initials="J." surname="Gettys">
+            <organization>W3C</organization>
+            <address><email>jg@w3.org</email></address>
+          </author>
+          <author fullname="J. Mogul" initials="J." surname="Mogul">
+            <organization>Compaq Computer Corporation</organization>
+            <address><email>mogul@wrl.dec.com</email></address>
+          </author>
+          <author fullname="H. Frystyk" initials="H." surname="Frystyk">
+            <organization>MIT Laboratory for Computer Science</organization>
+            <address><email>frystyk@w3.org</email></address>
+          </author>
+          <author fullname="L. Masinter" initials="L." surname="Masinter">
+            <organization>Xerox Corporation</organization>
+            <address><email>masinter@parc.xerox.com</email></address>
+          </author>
+          <author fullname="P. Leach" initials="P." surname="Leach">
+            <organization>Microsoft Corporation</organization>
+            <address><email>paulle@microsoft.com</email></address>
+          </author>
+          <author fullname="T. Berners-Lee" initials="T." surname="Berners-Lee">
+            <organization>W3C</organization>
+            <address><email>timbl@w3.org</email></address>
+          </author>
+          <date month="June" year="1999" />
+        </front>
+        <seriesInfo name="RFC" value="2616" />
+      </reference>
+      <reference anchor="RFC3864">
+        <front>
+          <title>Registration Procedures for Message Header Fields</title>
+          <author fullname="G. Klyne" initials="G." surname="Klyne">
+            <organization>Nine by Nine</organization>
+            <address><email>GK-IETF@ninebynine.org</email></address>
+          </author>
+          <author fullname="M. Nottingham" initials="M." surname="Nottingham">
+            <organization>BEA Systems</organization>
+            <address><email>mnot@pobox.com</email></address>
+          </author>
+          <author fullname="J. Mogul" initials="J." surname="Mogul">
+            <organization>HP Labs</organization>
+            <address><email>JeffMogul@acm.org</email></address>
+          </author>
+          <date month="September" year="2004" />
+        </front>
+        <seriesInfo name="BCP" value="90" />
+        <seriesInfo name="RFC" value="3864" />
+      </reference>
+    </references>
+    <section anchor="compatibility" title="Compatibility with Previous Versions">
+      <section anchor="changes.from.rfc.2068" title="Changes from RFC 2068">
+        <t>A case was missed in the Cache-Control model of HTTP/1.1; s-maxage was introduced to add
+          this missing case. (Sections <xref format="counter" target="response.cacheability" />,
+            <xref format="counter" target="header.cache-control" />).</t>
+        <t>Transfer-coding and message lengths all interact in ways that required fixing exactly
+          when chunked encoding is used (to allow for transfer encoding that may not be self
+          delimiting); it was important to straighten out exactly how message lengths are computed.
+          (see also <xref target="Part1" />, <xref target="Part3" /> and <xref target="Part5" />)
+          <cref source="jre">This used to refer to the text about non-modifiable headers, and will have to be updated later on.</cref>
+          </t>
+        <t>Proxies should be able to add Content-Length when appropriate.
+          <cref source="jre">This used to refer to the text about non-modifiable headers, and will have to be updated later on.</cref>
+        </t>
+        <t>Range request responses would become very verbose if all meta-data were always returned;
+          by allowing the server to only send needed headers in a 206 response, this problem can be
+          avoided. (<xref target="combining.headers" />)</t>
+        <t>The Cache-Control: max-age directive was not properly defined for responses. (<xref target="cache-response-directive"/>)</t>
+        <t>Warnings could be cached incorrectly, or not updated appropriately. (Section <xref
+            format="counter" target="expiration.model" />, <xref format="counter"
+            target="combining.headers" />, <xref format="counter" target="header.cache-control" />,
+          and <xref format="counter" target="header.warning" />) Warning also needed to be a general
+          header, as PUT or other methods may have need for it in requests.</t>
+      </section>
+      <section anchor="changes.from.rfc.2616" title="Changes from RFC 2616">
+        <t>Clarify denial of service attack avoidance requirement. (<xref
+            target="invalidation.after.updates.or.deletions" />)</t>
+      </section>
+<t>
+  Multiple warnings can be attached to a response (either by the origin server or by
+  a cache), including multiple warnings with the same code number. For example, a server
+  might provide the same warning with texts in both English and Basque.
+</t>
+<t>
+  When this occurs, the user agent &SHOULD; inform the user of as many of them as
+  possible, in the order that they appear in the response. If it is not possible to inform
+  the user of all of the warnings, the user agent &SHOULD; follow these heuristics:
+  <list style="symbols">
+    <t>Warnings that appear early in the response take priority over those appearing later
+      in the response.</t>
+    <t>Warnings in the user's preferred character set take priority over warnings in other
+      character sets but with identical warn-codes and warn-agents.</t>
+  </list>
+</t>
+<t>
+  Systems that generate multiple Warning headers &SHOULD; order them with this user
+  agent behavior in mind. New Warning headers &SHOULD; be added after any existing
+  Warning headers.
+</t>
+<t>
+  Warnings are assigned three digit warn-codes. The first digit indicates whether the
+  Warning is required to be deleted from a stored response after validation:
+  <list style="symbols">
+    <t>1xx Warnings that describe the freshness or validation status of the response, and so
+      &MUST; be deleted by caches after validation. They &MUST-NOT; be generated by a cache
+      except when validating a cached entry, and &MUST-NOT; be generated by clients.</t>
+    <t>2xx Warnings that describe some aspect of the entity body or entity headers that is
+      not rectified by a validation (for example, a lossy compression of the entity bodies)
+      and &MUST-NOT; be deleted by caches after validation, unless a full response is
+      returned, in which case they &MUST; be.</t>
+  </list>
+</t>
+<t>
+  The warn-text &SHOULD; be in a natural language and character set that is most likely
+  to be intelligible to the human user receiving the response. This decision can be based on
+  any available knowledge, such as the location of the cache or user, the Accept-Language
+  field in a request, the Content-Language field in a response, etc. The default language is
+  English and the default character set is ISO-8859-1 (<xref target="ISO-8859-1" />).
+</t>
+<t>
+  If a character set other than ISO-8859-1 is used, it &MUST; be encoded in the
+  warn-text using the method described in <xref target="RFC2047" />.
+</t>
+<t>
+  If an implementation sends a message with one or more Warning headers to a receiver whose
+  version is HTTP/1.0 or lower, then the sender &MUST; include in each warning-value a
+  warn-date that matches the Date header in the message.
+</t>
+<t>
+  If an implementation receives a message with a warning-value that includes a warn-date,
+  and that warn-date is different from the Date value in the response, then that
+  warning-value &MUST; be deleted from the message before storing, forwarding, or using
+  it. (preventing the consequences of naive caching of Warning header fields.) If all of the
+  warning-values are deleted for this reason, the Warning header &MUST; be deleted as
+  well.
+</t>
+<t>
+  The following warn-codes are defined by this specification, each with a recommended
+  warn-text in English, and a description of its meaning.
+</t>
+<t>
+Response is stale
+  <list>
+    <t>&SHOULD; be included whenever the returned response is stale.</t>
+  </list>
+</t>
+<t>
+Revalidation failed
+  <list>
+    <t>&SHOULD; be included if a cache returns a stale response because an attempt to
+      validate the response failed, due to an inability to reach the server.</t>
+  </list>
+</t>
+<t>
+Disconnected operation
+  <list>
+    <t>&SHOULD; be included if the cache is intentionally disconnected from the rest of
+      the network for a period of time.</t>
+  </list>
+</t>
+<t>
+Heuristic expiration
+  <list>
+    <t>&SHOULD; be included if the cache heuristically chose a freshness lifetime
+      greater than 24 hours and the response's age is greater than 24 hours.</t>
+  </list>
+</t>
+<t>
+Miscellaneous warning
+  <list>
+    <t>The warning text can include arbitrary information to be presented to a human
+      user, or logged. A system receiving this warning &MUST-NOT; take any automated
+      action, besides presenting the warning to the user.</t>
+  </list>
+</t>
+<t>
+Transformation applied
+  <list>
+    <t>&MUST; be added by an intermediate cache or proxy if it applies any
+      transformation changing the content-coding (as specified in the Content-Encoding
+      header) or media-type (as specified in the Content-Type header) of the response, or
+      the entity-body of the response, unless this Warning code already appears in the
+      response.</t>
+  </list>
+</t>
+<t>
+Miscellaneous persistent warning
+  <list>
+    <t>The warning text can include arbitrary information to be presented to a human
+      user, or logged. A system receiving this warning &MUST-NOT; take any automated
+      action.</t>
+  </list>
+</t>
+</section>
+</section>
+<section anchor="history.lists" title="History Lists">
+<t>
+  User agents often have history mechanisms, such as "Back" buttons and history lists, that
+  can be used to redisplay an entity retrieved earlier in a session.
+</t>
+<t>
+  History mechanisms and caches are different. In particular history mechanisms
+  &SHOULD-NOT; try to show a correct view of the current state of a resource. Rather, a
+  history mechanism is meant to show exactly what the user saw at the time when the resource
+  was retrieved.
+  </t>
+<t>
+  By default, an expiration time does not apply to history mechanisms. If the entity is still
+  in storage, a history mechanism &SHOULD; display it even if the entity has expired,
+  unless the user has specifically configured the agent to refresh expired history documents.
+</t>
+<t>
+  This is not to be construed to prohibit the history mechanism from telling the user that a
+  view might be stale.
+  <list>
+    <t>
+      <x:h>Note:</x:h> if history list mechanisms unnecessarily prevent users from viewing
+      stale resources, this will tend to force service authors to avoid using HTTP expiration
+      controls and cache controls when they would otherwise like to. Service authors may
+      consider it important that users not be presented with error messages or warning
+      messages when they use navigation controls (such as BACK) to view previously fetched
+      resources. Even though sometimes such resources ought not be cached, or ought to expire
+      quickly, user interface considerations may force service authors to resort to other
+      means of preventing caching (e.g. "once-only" URLs) in order not to suffer the effects
+      of improperly functioning history mechanisms.</t>
+  </list>
+</t>
+</section>
+<section anchor="IANA.considerations" title="IANA Considerations">
+<section anchor="message.header.registration" title="Message Header Registration">
+<t>
+  The Message Header Registry located at <eref
+  target="http://www.iana.org/assignments/message-headers/message-header-index.html" />
+  should be updated with the permanent registrations below (see <xref target="RFC3864" />):
+</t>
+<!--AUTOGENERATED FROM extract-header-defs.xslt, do not edit manually-->
+<texttable align="left" anchor="iana.header.registration.table" suppress-title="true">
+  <ttcol>Header Field Name</ttcol>
+  <ttcol>Protocol</ttcol>
+  <ttcol>Status</ttcol>
+  <ttcol>Reference</ttcol>
+  <c>Age</c>
+  <c>http</c>
+  <c>standard</c>
+  <c>
+    <xref target="header.age" />
+  </c>
+  <c>Cache-Control</c>
+  <c>http</c>
+  <c>standard</c>
+  <c>
+    <xref target="header.cache-control" />
+  </c>
+  <c>Expires</c>
+  <c>http</c>
+  <c>standard</c>
+  <c>
+    <xref target="header.expires" />
+  </c>
+  <c>Pragma</c>
+  <c>http</c>
+  <c>standard</c>
+  <c>
+    <xref target="header.pragma" />
+  </c>
+  <c>Vary</c>
+  <c>http</c>
+  <c>standard</c>
+  <c>
+    <xref target="header.vary" />
+  </c>
+  <c>Warning</c>
+  <c>http</c>
+  <c>standard</c>
+  <c>
+    <xref target="header.warning" />
+  </c>
+</texttable>
+<!--(END)-->
+<t>
+  The change controller is: "IETF (iesg@ietf.org) - Internet Engineering Task Force".
+</t>
+</section>
+</section>
+<section anchor="security.considerations" title="Security Considerations">
+<t>
+  Caches expose additional potential vulnerabilities, since the contents of the cache
+  represent an attractive target for malicious exploitation. Because cache contents persist
+  after an HTTP request is complete, an attack on the cache can reveal information long after
+  a user believes that the information has been removed from the network. Therefore, cache
+  contents should be protected as sensitive information.
+</t>
+</section>
+<section anchor="ack" title="Acknowledgments">
+<t>
+  Much of the content and presentation of the caching design is due to suggestions and
+  comments from individuals including: Shel Kaphan, Paul Leach, Koen Holtman, David Morris,
+  and Larry Masinter.
+</t>
+</section>
+</middle>
+<back>
+<references title="Normative References">
+  <reference anchor="ISO-8859-1">
+    <front>
+      <title> Information technology -- 8-bit single-byte coded graphic character sets -- Part
+: Latin alphabet No. 1 </title>
+      <author>
+        <organization>International Organization for Standardization</organization>
+      </author>
+      <date year="1998" />
+    </front>
+    <seriesInfo name="ISO/IEC" value="8859-1:1998" />
+  </reference>
+  <reference anchor="Part1">
+    <front>
+      <title abbrev="HTTP/1.1">HTTP/1.1, part 1: URIs, Connections, and Message Parsing</title>
+      <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+        <organization abbrev="Day Software">Day Software</organization>
+        <address><email>fielding@gbiv.com</email></address>
+      </author>
+      <author fullname="Jim Gettys" initials="J." surname="Gettys">
+        <organization>One Laptop per Child</organization>
+        <address><email>jg@laptop.org</email></address>
+      </author>
+      <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+        <organization abbrev="HP">Hewlett-Packard Company</organization>
+        <address><email>JeffMogul@acm.org</email></address>
+      </author>
+      <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>henrikn@microsoft.com</email></address>
+      </author>
+      <author fullname="Larry Masinter" initials="L." surname="Masinter">
+        <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+        <address><email>LMM@acm.org</email></address>
+      </author>
+      <author fullname="Paul J. Leach" initials="P." surname="Leach">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>paulle@microsoft.com</email></address>
+      </author>
+      <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+        <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+        <address><email>timbl@w3.org</email></address>
+      </author>
+      <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+        <organization abbrev="W3C">World Wide Web Consortium</organization>
+        <address><email>ylafon@w3.org</email></address>
+      </author>
+      <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+        <organization abbrev="greenbytes">greenbytes GmbH</organization>
+        <address><email>julian.reschke@greenbytes.de</email></address>
+      </author>
+      <date month="&ID-MONTH;" year="&ID-YEAR;" />
+    </front>
+    <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p1-messaging-&ID-VERSION;" />
+    <x:source basename="p1-messaging" href="p1-messaging.xml" />
+  </reference>
+  <reference anchor="Part2">
+    <front>
+      <title abbrev="HTTP/1.1">HTTP/1.1, part 2: Message Semantics</title>
+      <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+        <organization abbrev="Day Software">Day Software</organization>
+        <address><email>fielding@gbiv.com</email></address>
+      </author>
+      <author fullname="Jim Gettys" initials="J." surname="Gettys">
+        <organization>One Laptop per Child</organization>
+        <address><email>jg@laptop.org</email></address>
+      </author>
+      <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+        <organization abbrev="HP">Hewlett-Packard Company</organization>
+        <address><email>JeffMogul@acm.org</email></address>
+      </author>
+      <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>henrikn@microsoft.com</email></address>
+      </author>
+      <author fullname="Larry Masinter" initials="L." surname="Masinter">
+        <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+        <address><email>LMM@acm.org</email></address>
+      </author>
+      <author fullname="Paul J. Leach" initials="P." surname="Leach">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>paulle@microsoft.com</email></address>
+      </author>
+      <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+        <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+        <address><email>timbl@w3.org</email></address>
+      </author>
+      <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+        <organization abbrev="W3C">World Wide Web Consortium</organization>
+        <address><email>ylafon@w3.org</email></address>
+      </author>
+      <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+        <organization abbrev="greenbytes">greenbytes GmbH</organization>
+        <address><email>julian.reschke@greenbytes.de</email></address>
+      </author>
+      <date month="&ID-MONTH;" year="&ID-YEAR;" />
+    </front>
+    <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p2-semantics-&ID-VERSION;" />
+    <x:source basename="p2-semantics" href="p2-semantics.xml" />
+  </reference>
+  <reference anchor="Part3">
+    <front>
+      <title abbrev="HTTP/1.1">HTTP/1.1, part 3: Message Payload and Content Negotiation</title>
+      <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+        <organization abbrev="Day Software">Day Software</organization>
+        <address><email>fielding@gbiv.com</email></address>
+      </author>
+      <author fullname="Jim Gettys" initials="J." surname="Gettys">
+        <organization>One Laptop per Child</organization>
+        <address><email>jg@laptop.org</email></address>
+      </author>
+      <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+        <organization abbrev="HP">Hewlett-Packard Company</organization>
+        <address><email>JeffMogul@acm.org</email></address>
+      </author>
+      <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>henrikn@microsoft.com</email></address>
+      </author>
+      <author fullname="Larry Masinter" initials="L." surname="Masinter">
+        <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+        <address><email>LMM@acm.org</email></address>
+      </author>
+      <author fullname="Paul J. Leach" initials="P." surname="Leach">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>paulle@microsoft.com</email></address>
+      </author>
+      <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+        <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+        <address><email>timbl@w3.org</email></address>
+      </author>
+      <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+        <organization abbrev="W3C">World Wide Web Consortium</organization>
+        <address><email>ylafon@w3.org</email></address>
+      </author>
+      <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+        <organization abbrev="greenbytes">greenbytes GmbH</organization>
+        <address><email>julian.reschke@greenbytes.de</email></address>
+      </author>
+      <date month="&ID-MONTH;" year="&ID-YEAR;" />
+    </front>
+    <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p3-payload-&ID-VERSION;" />
+    <x:source basename="p3-payload" href="p3-payload.xml" />
+  </reference>
+  <reference anchor="Part4">
+    <front>
+      <title abbrev="HTTP/1.1">HTTP/1.1, part 4: Conditional Requests</title>
+      <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+        <organization abbrev="Day Software">Day Software</organization>
+        <address><email>fielding@gbiv.com</email></address>
+      </author>
+      <author fullname="Jim Gettys" initials="J." surname="Gettys">
+        <organization>One Laptop per Child</organization>
+        <address><email>jg@laptop.org</email></address>
+      </author>
+      <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+        <organization abbrev="HP">Hewlett-Packard Company</organization>
+        <address><email>JeffMogul@acm.org</email></address>
+      </author>
+      <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>henrikn@microsoft.com</email></address>
+      </author>
+      <author fullname="Larry Masinter" initials="L." surname="Masinter">
+        <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+        <address><email>LMM@acm.org</email></address>
+      </author>
+      <author fullname="Paul J. Leach" initials="P." surname="Leach">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>paulle@microsoft.com</email></address>
+      </author>
+      <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+        <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+        <address><email>timbl@w3.org</email></address>
+      </author>
+      <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+        <organization abbrev="W3C">World Wide Web Consortium</organization>
+        <address><email>ylafon@w3.org</email></address>
+      </author>
+      <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+        <organization abbrev="greenbytes">greenbytes GmbH</organization>
+        <address><email>julian.reschke@greenbytes.de</email></address>
+      </author>
+      <date month="&ID-MONTH;" year="&ID-YEAR;" />
+    </front>
+    <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p4-conditional-&ID-VERSION;" />
+    <x:source basename="p4-conditional" href="p4-conditional.xml" />
+  </reference>
+  <reference anchor="Part5">
+    <front>
+      <title abbrev="HTTP/1.1">HTTP/1.1, part 5: Range Requests and Partial Responses</title>
+      <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+        <organization abbrev="Day Software">Day Software</organization>
+        <address><email>fielding@gbiv.com</email></address>
+      </author>
+      <author fullname="Jim Gettys" initials="J." surname="Gettys">
+        <organization>One Laptop per Child</organization>
+        <address><email>jg@laptop.org</email></address>
+      </author>
+      <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+        <organization abbrev="HP">Hewlett-Packard Company</organization>
+        <address><email>JeffMogul@acm.org</email></address>
+      </author>
+      <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>henrikn@microsoft.com</email></address>
+      </author>
+      <author fullname="Larry Masinter" initials="L." surname="Masinter">
+        <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+        <address><email>LMM@acm.org</email></address>
+      </author>
+      <author fullname="Paul J. Leach" initials="P." surname="Leach">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>paulle@microsoft.com</email></address>
+      </author>
+      <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+        <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+        <address><email>timbl@w3.org</email></address>
+      </author>
+      <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+        <organization abbrev="W3C">World Wide Web Consortium</organization>
+        <address><email>ylafon@w3.org</email></address>
+      </author>
+      <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+        <organization abbrev="greenbytes">greenbytes GmbH</organization>
+        <address><email>julian.reschke@greenbytes.de</email></address>
+      </author>
+      <date month="&ID-MONTH;" year="&ID-YEAR;" />
+    </front>
+    <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p5-range-&ID-VERSION;" />
+    <x:source basename="p5-range" href="p5-range.xml" />
+  </reference>
+  <reference anchor="Part7">
+    <front>
+      <title abbrev="HTTP/1.1">HTTP/1.1, part 7: Authentication</title>
+      <author fullname="Roy T. Fielding" initials="R." role="editor" surname="Fielding">
+        <organization abbrev="Day Software">Day Software</organization>
+        <address><email>fielding@gbiv.com</email></address>
+      </author>
+      <author fullname="Jim Gettys" initials="J." surname="Gettys">
+        <organization>One Laptop per Child</organization>
+        <address><email>jg@laptop.org</email></address>
+      </author>
+      <author fullname="Jeffrey C. Mogul" initials="J." surname="Mogul">
+        <organization abbrev="HP">Hewlett-Packard Company</organization>
+        <address><email>JeffMogul@acm.org</email></address>
+      </author>
+      <author fullname="Henrik Frystyk Nielsen" initials="H." surname="Frystyk">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>henrikn@microsoft.com</email></address>
+      </author>
+      <author fullname="Larry Masinter" initials="L." surname="Masinter">
+        <organization abbrev="Adobe Systems">Adobe Systems, Incorporated</organization>
+        <address><email>LMM@acm.org</email></address>
+      </author>
+      <author fullname="Paul J. Leach" initials="P." surname="Leach">
+        <organization abbrev="Microsoft">Microsoft Corporation</organization>
+        <address><email>paulle@microsoft.com</email></address>
+      </author>
+      <author fullname="Tim Berners-Lee" initials="T." surname="Berners-Lee">
+        <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
+        <address><email>timbl@w3.org</email></address>
+      </author>
+      <author fullname="Yves Lafon" initials="Y." role="editor" surname="Lafon">
+        <organization abbrev="W3C">World Wide Web Consortium</organization>
+        <address><email>ylafon@w3.org</email></address>
+      </author>
+      <author fullname="Julian F. Reschke" initials="J. F." role="editor" surname="Reschke">
+        <organization abbrev="greenbytes">greenbytes GmbH</organization>
+        <address><email>julian.reschke@greenbytes.de</email></address>
+      </author>
+      <date month="&ID-MONTH;" year="&ID-YEAR;" />
+    </front>
+    <seriesInfo name="Internet-Draft" value="draft-ietf-httpbis-p7-auth-&ID-VERSION;" />
+    <x:source basename="p7-auth" href="p7-auth.xml" />
+  </reference>
+  <reference anchor="RFC2047">
+    <front>
+      <title abbrev="Message Header Extensions">MIME (Multipurpose Internet Mail Extensions)
+        Part Three: Message Header Extensions for Non-ASCII Text</title>
+      <author fullname="Keith Moore" initials="K." surname="Moore">
+        <organization>University of Tennessee</organization>
+        <address><email>moore@cs.utk.edu</email></address>
+      </author>
+      <date month="November" year="1996" />
+    </front>
+    <seriesInfo name="RFC" value="2047" />
+  </reference>
+  <reference anchor="RFC2119">
+    <front>
+      <title>Key words for use in RFCs to Indicate Requirement Levels</title>
+      <author fullname="Scott Bradner" initials="S." surname="Bradner">
+        <organization>Harvard University</organization>
+        <address><email>sob@harvard.edu</email></address>
+      </author>
+      <date month="March" year="1997" />
+    </front>
+    <seriesInfo name="BCP" value="14" />
+    <seriesInfo name="RFC" value="2119" />
+  </reference>
+  <reference anchor="RFC5234">
+    <front>
+      <title abbrev="ABNF for Syntax Specifications">Augmented BNF for Syntax Specifications: ABNF</title>
+      <author initials="D." surname="Crocker" fullname="Dave Crocker" role="editor">
+        <organization>Brandenburg InternetWorking</organization>
+        <address>
+        <postal>
+        <street>675 Spruce Dr.</street>
+        <city>Sunnyvale</city>
+        <region>CA</region>
+        <code>94086</code>
+        <country>US</country></postal>
+        <phone>+1.408.246.8253</phone>
+        <email>dcrocker@bbiw.net</email></address>
+      </author>
+      <author initials="P." surname="Overell" fullname="Paul Overell">
+        <organization>THUS plc.</organization>
+        <address>
+        <postal>
+        <street>1/2 Berkeley Square</street>
+        <street>99 Berkely Street</street>
+        <city>Glasgow</city>
+        <code>G3 7HR</code>
+        <country>UK</country></postal>
+        <email>paul.overell@thus.net</email></address>
+      </author>
+      <date month="January" year="2008"/>
+    </front>
+    <seriesInfo name="STD" value="68"/>
+    <seriesInfo name="RFC" value="5234"/>
+  </reference>
+</references>
+<references title="Informative References">
+  <reference anchor="RFC1305">
+    <front>
+      <title>Network Time Protocol (Version 3) Specification, Implementation</title>
+      <author fullname="David L. Mills" initials="D." surname="Mills">
+        <organization>University of Delaware, Electrical Engineering Department</organization>
+        <address><email>mills@udel.edu</email></address>
+      </author>
+      <date month="March" year="1992" />
+    </front>
+    <seriesInfo name="RFC" value="1305" />
+  </reference>
+  <reference anchor="RFC2616">
+    <front>
+      <title>Hypertext Transfer Protocol -- HTTP/1.1</title>
+      <author fullname="R. Fielding" initials="R." surname="Fielding">
+        <organization>University of California, Irvine</organization>
+        <address><email>fielding@ics.uci.edu</email></address>
+      </author>
+      <author fullname="J. Gettys" initials="J." surname="Gettys">
+        <organization>W3C</organization>
+        <address><email>jg@w3.org</email></address>
+      </author>
+      <author fullname="J. Mogul" initials="J." surname="Mogul">
+        <organization>Compaq Computer Corporation</organization>
+        <address><email>mogul@wrl.dec.com</email></address>
+      </author>
+      <author fullname="H. Frystyk" initials="H." surname="Frystyk">
+        <organization>MIT Laboratory for Computer Science</organization>
+        <address><email>frystyk@w3.org</email></address>
+      </author>
+      <author fullname="L. Masinter" initials="L." surname="Masinter">
+        <organization>Xerox Corporation</organization>
+        <address><email>masinter@parc.xerox.com</email></address>
+      </author>
+      <author fullname="P. Leach" initials="P." surname="Leach">
+        <organization>Microsoft Corporation</organization>
+        <address><email>paulle@microsoft.com</email></address>
+      </author>
+      <author fullname="T. Berners-Lee" initials="T." surname="Berners-Lee">
+        <organization>W3C</organization>
+        <address><email>timbl@w3.org</email></address>
+      </author>
+      <date month="June" year="1999" />
+    </front>
+    <seriesInfo name="RFC" value="2616" />
+  </reference>
+  <reference anchor="RFC3864">
+    <front>
+      <title>Registration Procedures for Message Header Fields</title>
+      <author fullname="G. Klyne" initials="G." surname="Klyne">
+        <organization>Nine by Nine</organization>
+        <address><email>GK-IETF@ninebynine.org</email></address>
+      </author>
+      <author fullname="M. Nottingham" initials="M." surname="Nottingham">
+        <organization>BEA Systems</organization>
+        <address><email>mnot@pobox.com</email></address>
+      </author>
+      <author fullname="J. Mogul" initials="J." surname="Mogul">
+        <organization>HP Labs</organization>
+        <address><email>JeffMogul@acm.org</email></address>
+      </author>
+      <date month="September" year="2004" />
+    </front>
+    <seriesInfo name="BCP" value="90" />
+    <seriesInfo name="RFC" value="3864" />
+  </reference>
+</references>
+<section anchor="compatibility" title="Compatibility with Previous Versions">
+<section anchor="changes.from.rfc.2068" title="Changes from RFC 2068">
+<t>
+  A case was missed in the Cache-Control model of HTTP/1.1; s-maxage was introduced to add
+  this missing case.
+  (Sections <xref format="counter" target="response.cacheability" />, <xref format="counter" target="header.cache-control" />).
+</t>
+<t>
+  Transfer-coding and message lengths all interact in ways that required fixing exactly
+  when chunked encoding is used (to allow for transfer encoding that may not be self
+  delimiting); it was important to straighten out exactly how message lengths are computed.
+  (see also <xref target="Part1" />, <xref target="Part3" /> and <xref target="Part5" />)
+  <cref source="jre">This used to refer to the text about non-modifiable headers, and will have to be updated later on.</cref>
+</t>
+<t>
+  Proxies should be able to add Content-Length when appropriate.
+  <cref source="jre">This used to refer to the text about non-modifiable headers, and will have to be updated later on.</cref>
+</t>
+<t
+  >Range request responses would become very verbose if all meta-data were always returned;
+  by allowing the server to only send needed headers in a 206 response, this problem can be
+  avoided.
+  (<xref target="combining.headers" />)
+</t>
+<t>
+  The Cache-Control: max-age directive was not properly defined for responses.
+  (<xref target="cache-response-directive"/>)
+</t>
+<t>
+  Warnings could be cached incorrectly, or not updated appropriately. (Section <xref
+    format="counter" target="expiration.model" />, <xref format="counter"
+    target="combining.headers" />, <xref format="counter" target="header.cache-control" />,
+  and <xref format="counter" target="header.warning" />) Warning also needed to be a general
+  header, as PUT or other methods may have need for it in requests.
+</t>
+</section>
+<section anchor="changes.from.rfc.2616" title="Changes from RFC 2616">
+<t>
+  Clarify denial of service attack avoidance requirement.
+  (<xref target="invalidation.after.updates.or.deletions" />)
+</t>
+</section>
     </section>
 …
 </artwork></figure></section>
+    <section anchor="change.log" title="Change Log (to be removed by RFC Editor before publication)">
+      <section title="Since RFC2616">
+        <t>Extracted relevant partitions from <xref target="RFC2616" />.</t>
+      </section>
+      <section title="Since draft-ietf-httpbis-p6-cache-00">
+        <t>Closed issues: <list style="symbols">
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/9" />: "Trailer"
+                (<eref target="http://purl.org/NET/http-errata#trailer-hop" />)</t>
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/12" />: "Invalidation
+              after Update or Delete" (<eref target="http://purl.org/NET/http-errata#invalidupd" />)</t>
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/35" />: "Normative and
+              Informative references"</t>
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/48" />: "Date
+              reference typo"</t>
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/49" />: "Connection
+              header text"</t>
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/65" />: "Informative
+              references"</t>
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/66" />: "ISO-8859-1
+              Reference"</t>
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/86" />: "Normative
+              up-to-date references"</t>
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/87" />: "typo in
+.2.2"</t>
+          </list>
+        </t>
+        <t>Other changes: <list style="symbols">
+            <t>Use names of RFC4234 core rules DQUOTE and HTAB (work in progress on <eref
+                target="http://tools.ietf.org/wg/httpbis/trac/ticket/36" />)</t>
+          </list>
+        </t>
+      </section>
+      <section title="Since draft-ietf-httpbis-p6-cache-01">
+        <t>Closed issues: <list style="symbols">
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/82" />: "rel_path not
+              used"</t>
+          </list>
+        </t>
+        <t>Other changes: <list style="symbols">
+            <t>Get rid of duplicate BNF rule names ("host" -&gt; "uri-host") (work in progress
+              on <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/36" />)</t>
+            <t>Add explicit references to BNF syntax and rules imported from other parts of the
+              specification.</t>
+          </list>
+        </t>
+      </section>
+      <section anchor="changes.since.02" title="Since draft-ietf-httpbis-p6-cache-02">
+        <t>Ongoing work on IANA Message Header Registration (<eref
+            target="http://tools.ietf.org/wg/httpbis/trac/ticket/40" />): <list style="symbols">
+            <t>Reference RFC 3984, and update header registrations for headers defined in this
+              document.</t>
+          </list>
+        </t>
+      </section>
+      <section anchor="changes.since.03" title="Since draft-ietf-httpbis-p6-cache-03">
+        <t>Closed issues: <list style="symbols">
+            <t>
+              <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/106" />: "Vary header
+              classification"</t>
+          </list>
+        </t>
+      </section>
+      <section anchor="changes.since.04" title="Since draft-ietf-httpbis-p6-cache-04">
+        <t>
+          Ongoing work on ABNF conversion (<eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/36"/>):
+          <list style="symbols">
+            <t>
+              Use "/" instead of "|" for alternatives.
+            </t>
+            <t>
+              Introduce new ABNF rules for "bad" whitespace ("BWS"), optional
+              whitespace ("OWS") and required whitespace ("RWS").
+            </t>
+            <t>
+              Rewrite ABNFs to spell out whitespace rules, factor out
+              header value format definitions.
+            </t>
+          </list>
+        </t>
+      </section>
+      <section anchor="changes.since.05" title="Since draft-ietf-httpbis-p6-cache-05">
+        <t>
+          This is a total rewrite of this part of the specification.
+        </t>
+        <t>
+          In addition: Ongoing work on ABNF conversion (<eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/36"/>):
+          <list style="symbols">
+            <t>
+              Add appendix containing collected and expanded ABNF, reorganize ABNF introduction.
+            </t>
+          </list>
+        </t>
+      </section>
+    </section>
+<section anchor="change.log" title="Change Log (to be removed by RFC Editor before publication)">
+<section title="Since RFC2616">
+  <t>Extracted relevant partitions from <xref target="RFC2616" />.</t>
+</section>
+<section title="Since draft-ietf-httpbis-p6-cache-00">
+<t>
+  Closed issues:
+  <list style="symbols">
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/9" />: "Trailer"
+        (<eref target="http://purl.org/NET/http-errata#trailer-hop" />)</t>
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/12" />: "Invalidation after Update or Delete" (<eref target="http://purl.org/NET/http-errata#invalidupd" />)</t>
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/35" />: "Normative and Informative references"</t>
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/48" />: "Date reference typo"</t>
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/49" />: "Connection header text"</t>
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/65" />: "Informative references"</t>
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/66" />: "ISO-8859-1 Reference"</t>
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/86" />: "Normative up-to-date references"</t>
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/87" />: "typo in 13.2.2"</t>
+  </list>
+</t>
+<t>
+  Other changes:
+  <list style="symbols">
+    <t>Use names of RFC4234 core rules DQUOTE and HTAB (work in progress on <eref
+        target="http://tools.ietf.org/wg/httpbis/trac/ticket/36" />)</t>
+  </list>
+</t>
+</section>
+<section title="Since draft-ietf-httpbis-p6-cache-01">
+<t>
+  Closed issues:
+  <list style="symbols">
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/82" />: "rel_path not used"</t>
+  </list>
+</t>
+<t>
+  Other changes:
+  <list style="symbols">
+    <t>Get rid of duplicate BNF rule names ("host" -&gt; "uri-host") (work in progress
+      on <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/36" />)</t>
+    <t>Add explicit references to BNF syntax and rules imported from other parts of the
+      specification.</t>
+  </list>
+</t>
+</section>
+<section anchor="changes.since.02" title="Since draft-ietf-httpbis-p6-cache-02">
+<t>
+  Ongoing work on IANA Message Header Registration (<eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/40" />):
+  <list style="symbols">
+    <t>Reference RFC 3984, and update header registrations for headers defined in this
+      document.</t>
+  </list>
+</t>
+</section>
+<section anchor="changes.since.03" title="Since draft-ietf-httpbis-p6-cache-03">
+<t>
+  Closed issues:
+  <list style="symbols">
+    <t>
+      <eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/106" />: "Vary header classification"</t>
+  </list>
+</t>
+</section>
+<section anchor="changes.since.04" title="Since draft-ietf-httpbis-p6-cache-04">
+<t>
+  Ongoing work on ABNF conversion (<eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/36"/>):
+  <list style="symbols">
+    <t>
+      Use "/" instead of "|" for alternatives.
+    </t>
+    <t>
+      Introduce new ABNF rules for "bad" whitespace ("BWS"), optional
+      whitespace ("OWS") and required whitespace ("RWS").
+    </t>
+    <t>
+      Rewrite ABNFs to spell out whitespace rules, factor out
+      header value format definitions.
+    </t>
+  </list>
+</t>
+</section>
+<section anchor="changes.since.05" title="Since draft-ietf-httpbis-p6-cache-05">
+  <t>
+    This is a total rewrite of this part of the specification.
+  </t>
+  <t>
+    In addition: Ongoing work on ABNF conversion (<eref target="http://tools.ietf.org/wg/httpbis/trac/ticket/36"/>):
+    <list style="symbols">
+      <t>
+        Add appendix containing collected and expanded ABNF, reorganize ABNF introduction.
+      </t>
+    </list>
+  </t>
+</section>
+</section>
   </back>
 </rfc>

draft-ietf-httpbis/latest/p7-auth.html

r538	r540
462	462	<tr>
463	463	<td class="header left"></td>
464		<td class="header right">March 5, 2009</td>
	464	<td class="header right">March 6, 2009</td>
465	465	</tr>
466	466	</table>

Note: See TracChangeset for help on using the changeset viewer.

Context Navigation

Changeset 540 for draft-ietf-httpbis

Legend:

Download in other formats: