draft-ietf-httpbis-cache-latest.xml

<?xml version="1.0" encoding="utf-8"?>
<?xml-stylesheet type='text/xsl' href='lib/myxml2rfc.xslt'?>
<!DOCTYPE rfc [
  <!ENTITY MAY "<bcp14>MAY</bcp14>">
  <!ENTITY MUST "<bcp14>MUST</bcp14>">
  <!ENTITY MUST-NOT "<bcp14>MUST NOT</bcp14>">
  <!ENTITY OPTIONAL "<bcp14>OPTIONAL</bcp14>">
  <!ENTITY RECOMMENDED "<bcp14>RECOMMENDED</bcp14>">
  <!ENTITY REQUIRED "<bcp14>REQUIRED</bcp14>">
  <!ENTITY SHALL "<bcp14>SHALL</bcp14>">
  <!ENTITY SHALL-NOT "<bcp14>SHALL NOT</bcp14>">
  <!ENTITY SHOULD "<bcp14>SHOULD</bcp14>">
  <!ENTITY SHOULD-NOT "<bcp14>SHOULD NOT</bcp14>">
  <!ENTITY ID-VERSION "latest">
  <!ENTITY mdash "&#8212;">
  <!ENTITY Note "<x:h xmlns:x='http://purl.org/net/xml2rfc/ext'>Note:</x:h>">
]>
<?rfc toc="yes" ?>
<?rfc tocdepth="4" ?>
<?rfc symrefs="yes" ?>
<?rfc sortrefs="yes" ?>
<?rfc compact="yes"?>
<?rfc subcompact="no" ?>
<?rfc linkmailto="no" ?>
<?rfc editing="no" ?>
<?rfc comments="yes"?>
<?rfc inline="yes"?>
<?rfc rfcedstyle="yes"?>
<?rfc-ext allow-markup-in-artwork="yes" ?>
<?rfc-ext html-pretty-print="prettyprint https://cdn.rawgit.com/google/code-prettify/master/loader/run_prettify.js"?>
<?rfc-ext include-references-in-index="yes" ?>
<?github-issue-label cache?>

<rfc category="std" docName="draft-ietf-httpbis-cache-&ID-VERSION;" ipr="pre5378Trust200902"
  obsoletes="7234" x:maturity-level="proposed" xmlns:x="http://purl.org/net/xml2rfc/ext" xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#'>
<x:feedback template="mailto:ietf-http-wg@w3.org?subject={docname},%20%22{section}%22&amp;body=&lt;{ref}&gt;:"/>
<front>

  <title>HTTP Caching</title>

  <author fullname="Roy T. Fielding" initials="R." surname="Fielding" role="editor">
    <organization>Adobe</organization>
    <address>
      <postal>
        <street>345 Park Ave</street>
        <city>San Jose</city>
        <region>CA</region>
        <code>95110</code>
        <country>United States of America</country>
      </postal>
      <email>fielding@gbiv.com</email>
      <uri>https://roy.gbiv.com/</uri>
    </address>
  </author>

  <author fullname="Mark Nottingham" initials="M." surname="Nottingham" role="editor">
    <organization>Fastly</organization>
    <address>
      <email>mnot@mnot.net</email>
      <uri>https://www.mnot.net/</uri>
    </address>
  </author>

  <author fullname="Julian F. Reschke" initials="J. F." surname="Reschke" role="editor">
    <organization abbrev="greenbytes">greenbytes GmbH</organization>
    <address>
      <postal>
        <street>Hafenweg 16</street>
        <city>Muenster</city><code>48155</code>
        <country>Germany</country>
      </postal>
      <email>julian.reschke@greenbytes.de</email>
      <uri>https://greenbytes.de/tech/webdav/</uri>
    </address>
  </author>

  <date/>

  <area>Applications and Real-Time</area>
  <workgroup>HTTP</workgroup>

  <keyword>Hypertext Transfer Protocol</keyword>
  <keyword>HTTP</keyword>
  <keyword>HTTP Caching</keyword>

<abstract>
<t>
   The Hypertext Transfer Protocol (HTTP) is a stateless application-level
   protocol for distributed, collaborative, hypertext information systems.
   This document defines HTTP caches and the associated header fields that
   control cache behavior or indicate cacheable response messages.
</t>
<t>
   This document obsoletes RFC 7234.
</t>
</abstract>

<note title="Editorial Note" removeInRFC="true">
  <t>
    Discussion of this draft takes place on the HTTP working group
    mailing list (ietf-http-wg@w3.org), which is archived at
    <eref target="https://lists.w3.org/Archives/Public/ietf-http-wg/"/>.
  </t>
  <t>
    Working Group information can be found at <eref target="https://httpwg.org/"/>;
    source code and issues list for this draft can be found at
    <eref target="https://github.com/httpwg/http-core"/>.
  </t>
  <t>
    The changes in this draft are summarized in <xref target="changes.since.06"/>.
  </t>
</note>

   </front>
   <middle>

<section title="Introduction" anchor="caching">
<t>
   The Hypertext Transfer Protocol (HTTP) is a stateless application-level
   request/response protocol that uses extensible semantics and
   self-descriptive messages for flexible interaction with network-based
   hypertext information systems. HTTP is defined by a series of documents
   that collectively form the HTTP/1.1 specification:
</t>
<ul>
   <li>"HTTP Semantics" <xref target="Semantics"/></li>
   <li>"HTTP Caching" (this document)</li>
   <li>"HTTP/1.1 Messaging" <xref target="Messaging"/></li>
</ul>
<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 related to caching and reusing response
   messages.
</t>

<iref item="cache" />
<t>
   An HTTP <x:dfn>cache</x:dfn> is a local store of response messages and the
   subsystem that controls storage, retrieval, and deletion of messages in it.
   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; employ a cache, though a cache cannot be used by a server that
   is acting as a tunnel.
</t>
<iref item="shared cache" />
<iref item="private cache" />
<t anchor="shared.and.private.caches">
   A <x:dfn>shared cache</x:dfn> is a cache that stores responses to be reused
   by more than one user; shared caches are usually (but not always) deployed
   as a part of an intermediary. A <x:dfn>private cache</x:dfn>, in contrast,
   is dedicated to a single user; often, they are deployed as a component of 
   a user agent.
</t>
<t>
   The goal of caching in HTTP is to significantly improve performance
   by reusing a prior response message to satisfy a current request.
   A stored response is considered "fresh", as defined in
   <xref target="expiration.model" />, if the response can be reused without
   "validation" (checking with the origin server to see if the cached response
   remains valid for this request).  A fresh response can therefore
   reduce both latency and network overhead each time it is reused.
   When a cached response is not fresh, it might still be reusable if it can
   be freshened by validation (<xref target="validation.model" />) or if the
   origin is unavailable (<xref target="serving.stale.responses" />).
</t>
<t>
   This document obsoletes <xref target="RFC7234" x:fmt="none">RFC 7234</xref>,
   with the changes being summarized in <xref target="changes.from.rfc.7234"/>.
</t>

<section title="Requirements Notation" anchor="intro.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>
   Conformance criteria and considerations regarding error handling
   are defined in <xref target="Semantics-conformance"/>.
</t>
</section>

<section title="Syntax Notation" anchor="notation">
<iref primary="true" item="Grammar" subitem="ALPHA"/>
<iref primary="true" item="Grammar" subitem="CR"/>
<iref primary="true" item="Grammar" subitem="CRLF"/>
<iref primary="true" item="Grammar" subitem="CTL"/>
<iref primary="true" item="Grammar" subitem="DIGIT"/>
<iref primary="true" item="Grammar" subitem="DQUOTE"/>
<iref primary="true" item="Grammar" subitem="HEXDIG"/>
<iref primary="true" item="Grammar" subitem="HTAB"/>
<iref primary="true" item="Grammar" subitem="LF"/>
<iref primary="true" item="Grammar" subitem="OCTET"/>
<iref primary="true" item="Grammar" subitem="SP"/>
<iref primary="true" item="Grammar" subitem="VCHAR"/>
<t>
   This specification uses the Augmented Backus-Naur Form (ABNF) notation of
   <xref target="RFC5234"/>, extended with the notation for case-sensitivity
   in strings defined in <xref target="RFC7405"/>.
</t>
<t>   
   It also uses a list extension, defined in <xref target="abnf.extension"/>,
   that allows for compact definition of comma-separated lists using a '#'
   operator (similar to how the '*' operator indicates repetition). <xref
   target="collected.abnf"/> shows the collected grammar with all list
   operators expanded to standard ABNF notation.
</t>
<t anchor="core.rules">
  <x:anchor-alias value="ALPHA"/>
  <x:anchor-alias value="CR"/>
  <x:anchor-alias value="CRLF"/>
  <x:anchor-alias value="CTL"/>
  <x:anchor-alias value="DIGIT"/>
  <x:anchor-alias value="DQUOTE"/>
  <x:anchor-alias value="HEXDIG"/>
  <x:anchor-alias value="HTAB"/>
  <x:anchor-alias value="LF"/>
  <x:anchor-alias value="OCTET"/>
  <x:anchor-alias value="SP"/>
  <x:anchor-alias value="VCHAR"/>
   The following core rules are included by
   reference, as defined in <xref target="RFC5234" x:fmt="," x:sec="B.1"/>:
   ALPHA (letters), CR (carriage return), CRLF (CR LF), CTL (controls),
   DIGIT (decimal 0-9), DQUOTE (double quote),
   HEXDIG (hexadecimal 0-9/A-F/a-f), HTAB (horizontal tab), LF (line feed),
   OCTET (any 8-bit sequence of data), SP (space), and
   VCHAR (any visible <xref target="USASCII"/> character).
</t>
<t anchor="imported.rules">
  <x:anchor-alias value="HTTP-date"/>
  <x:anchor-alias value="OWS"/>
  <x:anchor-alias value="field-name"/>
  <x:anchor-alias value="quoted-string"/>
  <x:anchor-alias value="token"/>
  The rules below are defined in <xref target="Semantics"/>:
</t>
<sourcecode type="abnf7230">
  <x:ref>HTTP-date</x:ref>     = &lt;HTTP-date, see <xref target="http.date"/>&gt;
  <x:ref>OWS</x:ref>           = &lt;OWS, see <xref target="whitespace"/>&gt;
  <x:ref>field-name</x:ref>    = &lt;field-name, see <xref target="field.names"/>&gt;
  <x:ref>quoted-string</x:ref> = &lt;quoted-string, see <xref target="quoted.strings"/>&gt;
  <x:ref>token</x:ref>         = &lt;token, see <xref target="tokens"/>&gt;
</sourcecode>
</section>

<section title="Delta Seconds" anchor="delta-seconds">
<t>
   The delta-seconds rule specifies a non-negative integer, representing time
   in seconds.
</t>
<sourcecode type="abnf7230"><iref item="Grammar" primary="true" subitem="delta-seconds" />
  <x:ref>delta-seconds</x:ref>  = 1*<x:ref>DIGIT</x:ref>
</sourcecode>
<t>
   A recipient parsing a delta-seconds value and converting it to binary form
   ought to use an arithmetic type of at least 31 bits of non-negative integer
   range.
   If a cache receives a delta-seconds value greater than the greatest integer
   it can represent, or if any of its subsequent calculations overflows,
   the cache &MUST; consider the value to be either 2147483648
   (2<sup>31</sup>) or the greatest positive integer it can conveniently
   represent.
</t>
<aside>
   <t>
       &Note; The value 2147483648 is here for historical reasons, effectively
       represents infinity (over 68 years), and does not need to be stored in
       binary form; an implementation could produce it as a canned string if
       any overflow occurs, even if the calculations are performed with an
       arithmetic type incapable of directly representing that number.
       What matters here is that an overflow be detected and not treated as a
       negative value in later calculations.
   </t>
</aside>
</section>
</section>

<section title="Overview of Cache Operation" anchor="caching.overview">
<iref item="cache key" />
<t>
   Proper cache operation preserves the semantics of HTTP transfers (<xref
   target="Semantics"/>) while reducing the transfer of information already
   held in the cache. Although caching is an entirely &OPTIONAL; feature of
   HTTP, it can be assumed that reusing a cached response is desirable and
   that such reuse is the default behavior when no requirement or local
   configuration prevents it. Therefore, HTTP cache requirements are focused
   on preventing a cache from either storing a non-reusable response or
   reusing a stored response inappropriately, rather than mandating that
   caches always store and reuse particular responses.
</t>
<iref item="cache key" />
<t>
   The base <x:dfn>cache key</x:dfn> consists of the request method and target
   URI used to retrieve the stored response; the method determines under which
   circumstances that response can be used to satisfy a request. However, many
   HTTP caches in common use today only cache GET responses, and therefore only
   use the URI as the cache key, forwarding other methods.
</t>
<t>
   If a request target is subject to content negotiation, the cache might
   store multiple responses for it. Caches differentiate these responses
   by incorporating values of the original request's selecting header fields
   into the cache key as well, as per <xref target="caching.negotiated.responses"/>.
</t>
<t>
  Furthermore, caches might incorporate additional material into the cache key.
  For example, user agent caches might include the referring site's identity,
  thereby "double keying" the cache to avoid some privacy risks (see <xref
  target="security.timing"/>).
</t>
<t>
   Most commonly, caches store the successful result of a retrieval
   request: i.e., a <x:ref>200 (OK)</x:ref> response to a GET request, which
   contains a representation of the resource identified by the request target
   (<xref target="GET"/>). However, it is also possible to store 
   redirects, negative results (e.g., <x:ref>404 (Not Found)</x:ref>),
   incomplete results (e.g., <x:ref>206 (Partial Content)</x:ref>), and
   responses to methods other than GET if the method's definition allows such
   caching and defines something suitable for use as a cache key.
</t>
<t>
   A cache is <x:dfn>disconnected</x:dfn> when it cannot contact the origin
   server or otherwise find a forward path for a given request. A
   disconnected cache can serve stale responses in some circumstances (<xref
   target="serving.stale.responses"/>).
</t>

</section>

<section title="Storing Responses in Caches" anchor="response.cacheability">
<t>
   A cache &MUST-NOT; store a response to any request, unless:
</t>
<ul>
  <li><t>The request method is understood by the cache, and</t></li>
  <li><t>the response status code is final (see <xref
  target="associating.response.to.request"/>), and</t></li>
  <li><t>the response status code is understood by the cache, and</t></li>
  <li><t>the "no-store" cache directive (see <xref
  target="header.cache-control" />) does not appear in the response, 
  and</t></li>
  <li><t>the "private" response directive (see <xref
  target="cache-response-directive.private" />) does not appear in the
  response, if the cache is shared, and</t></li>
  <li><t>the <x:ref>Authorization</x:ref> header field (see
  <xref target="header.authorization"/>) does not appear in the request, if the cache is
  shared, unless the response explicitly allows it (see <xref
  target="caching.authenticated.responses" />), and</t></li>
  <li><t>the response either:</t>
     <ul>
        <li>contains an <x:ref>Expires</x:ref> header field (see
        <xref target="header.expires"/>), or</li>
        <li>contains a max-age response directive (see <xref
        target="cache-response-directive.max-age" />), or</li>
        <li>contains a s-maxage response directive (see <xref
        target="cache-response-directive.s-maxage" />) and the cache is
        shared, or</li>
        <li>contains a Cache Control Extension (see <xref
        target="cache.control.extensions" />) that allows it to be cached,
        or</li>
        <li>has a status code that is defined as heuristically cacheable
        (see <xref target="heuristic.freshness" />), or</li>
        <li>contains a public response directive (see <xref
        target="cache-response-directive.public"/>).</li>
     </ul>
  </li>
</ul>
<t>
   Note that any of the requirements listed above can be overridden by a
   cache-control extension; see <xref target="cache.control.extensions" />.
</t>
<t>
   In this context, a cache has "understood" a request method or a response
   status code if it recognizes it and implements all specified
   caching-related behavior.
</t>
<t>
   Note that, in normal operation, some 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 title="Storing Incomplete Responses" anchor="incomplete.responses">
<t>
   A response message is considered complete when all of the octets indicated
   by its framing are available. Note that, when no explicit framing is
   provided, a response message that is ended by the connection's close is
   considered complete even though it might be indistinguishable from an
   incomplete response (see <xref target="Messaging" x:fmt=","
   x:rel="#message.body.length"/>). A cache &SHOULD; consider a close-terminated
   response incomplete if the connection termination is detected to be an
   error. A server that wishes to avoid premature termination resulting in an
   incorrect cached response &SHOULD; send the response with explicit framing.
</t>
<t>
   If the request method is GET, the response status code is <x:ref>200
   (OK)</x:ref>, and the entire response header section has been received, a
   cache &MAY; store an incomplete response message body if the stored response
   is recorded as incomplete. Likewise, a <x:ref>206 (Partial
   Content)</x:ref> response &MAY; be stored as if it were an incomplete
   <x:ref>200 (OK)</x:ref> response. However, a cache &MUST-NOT; store
   incomplete or partial-content responses if it does not support the
   <x:ref>Range</x:ref> and <x:ref>Content-Range</x:ref> header fields or if
   it does not understand the range units used in those fields.
</t>
<t>
   A cache &MAY; complete a stored incomplete response by making a subsequent
   range request (<xref target="header.range"/>) and combining the successful response with the
   stored response, as defined in <xref target="combining.responses"/>. A cache
   &MUST-NOT; use an incomplete response to answer requests unless the
   response has been made complete or the request is partial and specifies a
   range that is wholly within the incomplete response. A cache &MUST-NOT;
   send a partial response to a client without explicitly marking it as such
   using the <x:ref>206 (Partial Content)</x:ref> status code.
</t>
</section>


<section title="Storing Responses to Authenticated Requests" anchor="caching.authenticated.responses">
<t>
   A shared cache &MUST-NOT; use a cached response to a request with an
   <x:ref>Authorization</x:ref> header field (<xref target="header.authorization"/>) to
   satisfy any subsequent request unless a response directive that allows such
   responses to be stored is present.
</t>
<t>
   In this specification, the following <x:ref>Cache-Control</x:ref> response
   directives (<xref target="cache-response-directive"/>) have such an effect:
   must-revalidate, public, and s-maxage.
</t>
</section>

<section title="Combining Partial Content" anchor="combining.responses">
<t>
   A response might transfer only a partial representation if the
   connection closed prematurely or if the request used one or more Range
   specifiers (<xref target="header.range"/>).  After several such transfers, a cache might have
   received several ranges of the same representation.  A cache &MAY; combine
   these ranges into a single stored response, and reuse that response to
   satisfy later requests, if they all share the same strong validator and
   the cache complies with the client requirements in <xref target="combining.byte.ranges"/>.
</t>
<t>
   When combining the new response with one or more stored responses, a cache
   &MUST; use the header fields provided in the new response, aside from
   <x:ref>Content-Range</x:ref>, to replace all instances of the
   corresponding header fields in the stored response.
</t>
</section>

</section>


<section title="Constructing Responses from Caches" anchor="constructing.responses.from.caches">
<t>
   When presented with a request, a cache &MUST-NOT; reuse a stored response,
   unless:
</t>
<ul>
   <li><t>The presented effective request URI (<xref target="effective.request.uri"/>) and
   that of the stored response match, and</t></li>
   <li><t>the request method associated with the stored response allows it to
   be used for the presented request, and</t></li>
   <li><t>selecting header fields nominated by the stored response (if any)
   match those presented (see <xref target="caching.negotiated.responses"
   />), and</t></li>
   <li><t>the stored response does not contain the no-cache cache directive
   (<xref target="cache-response-directive.no-cache"/>), unless it is
   successfully validated (<xref target="validation.model"/>), and</t></li>
   <li><t>the stored response is either:</t>
     <ul>
        <li>fresh (see <xref target="expiration.model" />), or</li>
        <li>allowed to be served stale (see <xref
        target="serving.stale.responses" />), or</li>
        <li>successfully validated (see <xref target="validation.model"
        />).</li>
     </ul>
   </li>
</ul>
<t>
   Note that any of the requirements listed above can be overridden by a
   cache-control extension; see <xref target="cache.control.extensions" />.
</t>
<t>
   When a stored response is used to satisfy a request without validation, a
   cache &MUST; generate an <x:ref>Age</x:ref> header field (<xref
   target="header.age"/>), replacing any present in the response with a value
   equal to the stored response's current_age; see <xref
   target="age.calculations" />.
</t>
<t>
   A cache &MUST; write through requests with methods that are unsafe
   (<xref target="safe.methods"/>) to the origin server; i.e., a cache is not allowed to
   generate a 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" />.
</t>
<t>
   When more than one suitable response is stored, a cache &MUST; use the 
   most recent one (as determined by the <x:ref>Date</x:ref> header
   field). It can also forward the request with "Cache-Control: max-age=0" or
   "Cache-Control: no-cache" to disambiguate which response to use.
</t>
<t>
   A cache that does not have a clock available &MUST-NOT; use stored
   responses without revalidating them upon every use.
</t>

<section title="Calculating Cache Keys with Vary" anchor="caching.negotiated.responses">
<t>
   When a cache receives a request that can be satisfied by a stored response
   that has a <x:ref>Vary</x:ref> header field (<xref target="header.vary"/>),
   it &MUST-NOT; use that response unless all of the selecting header fields
   nominated by the Vary header field match in both the original request
   (i.e., that associated with the stored response), and the presented
   request.
</t>
<t>
   The selecting header fields from two requests are defined to match if and
   only if those in the first request can be transformed to those in the
   second request by applying any of the following:
</t>
<ul>
  <li>
     adding or removing whitespace, where allowed in the header field's
     syntax
  </li>
  <li>
     combining multiple header fields with the same field name
     (see <xref target="field.values"/>)
  </li>
  <li>
     normalizing both header field values in a way that is known to have
     identical semantics, according to the header field's specification
     (e.g., reordering field values when order is not significant;
     case-normalization, where values are defined to be case-insensitive)
  </li>
</ul>
<t>
   If (after any normalization that might take place) a header field is absent
   from a request, it can only match another request if it is also absent
   there.
</t>
<t>
   A <x:ref>Vary</x:ref> header field-value of "*" always fails to match.
</t>
<t>
   The stored response with matching selecting header fields is known as the
   selected response.
</t>
<t>
   If multiple selected responses are available (potentially including
   responses without a Vary header field), the cache will need to choose one to use.
   When a selecting header field has a known mechanism for doing so (e.g., qvalues on
   <x:ref>Accept</x:ref> and similar request header fields), that mechanism &MAY; be
   used to select preferred responses; of the remainder, the most recent
   response (as determined by the <x:ref>Date</x:ref> header field) is used, as
   per <xref target="constructing.responses.from.caches"/>.
</t>
<t>
   Note that in practice, some resources might send the Vary header field on
   responses inconsistently. When a cache has multiple responses for a given
   target URI, and one or more omits the Vary header field, it &SHOULD; use the
   most recent non-empty value available to select an appropriate response
   for the request.
</t>
<t>
   If no selected response is available, the cache cannot satisfy the 
   presented request. Typically, it is forwarded to the origin server 
   in a (possibly conditional; see <xref target="validation.model"/>) request.
</t>
</section>

<section title="Freshness" anchor="expiration.model">
<iref item="fresh" />
<iref item="stale" />
<t>
   A <x:dfn>fresh</x:dfn> response is one whose age has not yet exceeded its
   freshness lifetime. Conversely, a <x:dfn>stale</x:dfn>
   response is one where it has.
</t>
<iref item="freshness lifetime" />
<iref item="explicit expiration time" />
<iref item="heuristic expiration time" />
<t>
   A response's <x:dfn>freshness lifetime</x:dfn> is the length of time
   between its generation by the origin server and its expiration time. An
   <x:dfn>explicit expiration time</x:dfn> is the time at which the origin
   server intends that a stored response can no longer be used by a cache
   without further validation, whereas a <x:dfn>heuristic expiration
   time</x:dfn> is assigned by a cache when no explicit expiration time is
   available.
</t>
<iref item="age" />
<t>
   A response's <x:dfn>age</x:dfn> is the time that has passed since it was
   generated by, or successfully validated with, the origin server.  
</t>
<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
   <x:ref>Expires</x:ref> header field (<xref target="header.expires" />) or
   the max-age response directive (<xref
   target="cache-response-directive.max-age" />). Generally, origin servers
   will assign future explicit expiration times to responses in the belief
   that the representation 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 to indicate that the
   response is already stale. Compliant caches will normally validate a stale
   cached response before reusing it for subsequent requests (see <xref
   target="serving.stale.responses" />).
</t>
<t>
   Since origin servers do not always provide explicit expiration times,
   caches are also allowed to use a heuristic to determine an expiration time
   under certain circumstances (see <xref target="heuristic.freshness"/>).
</t>
<t>
  The calculation to determine if a response is fresh is:
</t>
<artwork type="code">
   response_is_fresh = (freshness_lifetime &gt; current_age)
</artwork>
<t>
   freshness_lifetime is defined in <xref
   target="calculating.freshness.lifetime" />; current_age is defined in
   <xref target="age.calculations" />.
</t>
<t>
   Clients can send the max-age or min-fresh request directives (<xref
   target="cache-request-directive" />) to constrain or relax freshness
   calculations for the corresponding response. However, caches are not
   required to honor them.
</t>
<t>
   When calculating freshness, to avoid common problems in date parsing:
</t>
<ul>
   <li>Although all date formats are specified to be case-sensitive, 
      a cache recipient &SHOULD; match day, week, and time-zone names
      case-insensitively.</li>
   <li>If a cache recipient's internal implementation of time has less
      resolution than the value of an HTTP-date, the recipient &MUST;
      internally represent a parsed <x:ref>Expires</x:ref> date as the
      nearest time equal to or earlier than the received value.</li>
   <li>A cache recipient &MUST-NOT; allow local time zones to influence the
      calculation or comparison of an age or expiration time.</li>
   <li>A cache recipient &SHOULD; consider a date with a zone abbreviation
      other than GMT or UTC to be invalid for calculating expiration.</li>
</ul>
<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 title="Calculating Freshness Lifetime" anchor="calculating.freshness.lifetime">
<t>
   A cache can calculate the freshness lifetime (denoted as
   freshness_lifetime) of a response by using the first match of the following:
</t>
<ul>
   <li>If the cache is shared and the s-maxage response directive
   (<xref target="cache-response-directive.s-maxage" />) is present, use its value,
   or</li>
   <li>If the max-age response directive (<xref
   target="cache-response-directive.max-age" />) is present, use its value, or</li>
   <li>If the <x:ref>Expires</x:ref> response header field
   (<xref target="header.expires" />) is present, use its value minus the
    value of the <x:ref>Date</x:ref> response header field, or</li>
   <li>Otherwise, no explicit expiration time is present in the response. A
   heuristic freshness lifetime might be applicable; see <xref
   target="heuristic.freshness" />.</li>
</ul>
<t>
   Note that this calculation is not vulnerable to clock skew, since all of
   the information comes from the origin server.
</t>
<t>
   When there is more than one value present for a given directive (e.g., two
   <x:ref>Expires</x:ref> header fields, multiple Cache-Control: max-age
   directives), the directive's value is considered invalid. Caches are
   encouraged to consider responses that have invalid freshness information to
   be stale.
</t>
</section>

<section title="Calculating Heuristic Freshness" anchor="heuristic.freshness">
  <iref item="heuristically cacheable"/>
<t>
   Since origin servers do not always provide explicit expiration times, a
   cache &MAY; assign a heuristic expiration time when an explicit time is not
   specified, employing algorithms that use other header field values (such as
   the <x:ref>Last-Modified</x:ref> time) to estimate a plausible expiration
   time. This specification does not provide specific algorithms, but does
   impose worst-case constraints on their results.
</t>
<t>
   A cache &MUST-NOT; use heuristics to determine freshness when an explicit
   expiration time is present in the stored response. Because of the
   requirements in <xref target="response.cacheability"/>, this means that,
   effectively, heuristics can only be used on responses without explicit
   freshness whose status codes are defined as "<x:dfn>heuristically cacheable</x:dfn>" (e.g., see
   <xref target="overview.of.status.codes"/>), and those responses without
   explicit freshness that have been marked as explicitly cacheable (e.g.,
   with a "public" response directive).
</t>
<t>
   Note that in previous specifications heuristically cacheable response status
   codes were called "cacheable by default."
</t>
<t>
   If the response has a <x:ref>Last-Modified</x:ref> header field
   (<xref target="header.last-modified"/>), caches are encouraged to use a heuristic
   expiration value that is no more than some fraction of the interval since
   that time. A typical setting of this fraction might be 10%.
</t>
<aside>
   <t>
      &Note; <xref target="RFC2616" x:fmt="of" x:sec="13.9"/> prohibited caches
      from calculating heuristic freshness for URIs with query components
      (i.e., those containing '?'). In practice, this has not been widely
      implemented. Therefore, origin servers are encouraged to send explicit
      directives (e.g., Cache-Control: no-cache) if they wish to preclude
      caching.
   </t>
</aside>
</section>

<section title="Calculating Age" anchor="age.calculations">
<t>
   The <x:ref>Age</x:ref> header field is used to convey an estimated
   age of the response message when obtained from a cache. The Age field value
   is the cache's estimate of the number of seconds 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 following data is used for the age calculation:
</t>
<dl newline="false">
  <dt><x:dfn>age_value</x:dfn></dt>
  <dd>
     The term "age_value" denotes the value of the <x:ref>Age</x:ref>
     header field (<xref target="header.age"/>), in a form appropriate for
     arithmetic operation; or 0, if not available.
  </dd>
  <dt><x:dfn>date_value</x:dfn></dt>
  <dd>
     The term "date_value" denotes the value of
     the Date header field, in a form appropriate for arithmetic
     operations. See <xref target="header.date"/> for the definition of the Date header
     field, and for requirements regarding responses without it.
  </dd>
  <dt><x:dfn>now</x:dfn></dt>
  <dd>
     The term "now" means "the current value of the clock at the host
     performing the calculation". A host ought to use NTP (<xref
     target="RFC5905"/>) or some similar protocol to synchronize its
     clocks to Coordinated Universal Time.
  </dd>
  <dt><x:dfn>request_time</x:dfn></dt>
  <dd>
     The current value of the clock at the host at the time the request
     resulting in the stored response was made.
  </dd>
  <dt><x:dfn>response_time</x:dfn></dt>
  <dd>
     The current value of the clock at the host at the time the response
     was received.
  </dd>
</dl>
<t>
   A response's age can be calculated in two entirely independent ways:
</t>
<ol>
   <li>the "apparent_age": response_time 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.</li>
   <li>the "corrected_age_value", if all of the caches along the response
   path implement HTTP/1.1 or greater. A cache &MUST; interpret this value
   relative to the time the request was initiated, not the time that the
   response was received.</li>
</ol>
<artwork type="code">
  apparent_age = max(0, response_time - date_value);

  response_delay = response_time - request_time;
  corrected_age_value = age_value + response_delay;  
</artwork>
<t>
   These are combined as
</t>
<artwork type="code">
  corrected_initial_age = max(apparent_age, corrected_age_value);
</artwork>
<t>
   unless the cache is confident in the value of the <x:ref>Age</x:ref> header
   field (e.g., because there are no HTTP/1.0 hops in the <x:ref>Via</x:ref>
   header field), in which case the corrected_age_value &MAY; be used as the
   corrected_initial_age.
</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>
<artwork type="code">
  resident_time = now - response_time;
  current_age = corrected_initial_age + resident_time;
</artwork>
</section>

<section title="Serving Stale Responses" anchor="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>
   A cache &MUST-NOT; generate 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>
   A cache &MUST-NOT; generate a stale response unless it is disconnected
   or doing so is explicitly permitted by the client or origin server
   (e.g., by the max-stale request directive in <xref
   target="cache-request-directive" />, by extension directives such as those
   defined in <xref target="RFC5861"/>, or by configuration in accordance
   with an out-of-band contract).
</t>
</section>
</section>

<section title="Validation" anchor="validation.model">
<t>
   When a cache has one or more stored responses for a requested URI, but
   cannot serve any of them (e.g., because they are not fresh, or one cannot
   be selected; see <xref target="caching.negotiated.responses"/>), it can use
   the conditional request mechanism <xref target="preconditions"/> in the forwarded request to
   give the next inbound server an opportunity to select a valid stored
   response to use, updating the stored metadata in the process, or to replace
   the stored response(s) with a new response. This process is known as
   "validating" or "revalidating" the stored response.
</t>

<section title="Sending a Validation Request" anchor="validation.sent"><iref item="validator" />
<t>
   When generating a conditional request for validation, a cache starts with
   either a request it is attempting to satisfy, or &mdash; if it is initiating
   the request independently &mdash; it synthesises a request using a stored
   response by copying the method, request-target, and request header fields
   identified by the Vary header field <xref
   target="caching.negotiated.responses"/>.
</t>
<t>
   It then updates that request with one or more precondition header fields.
   These contain validator metadata sourced from stored response(s) that have
   the same cache key.
</t>
<t>
   The precondition header fields are then compared by recipients to
   determine whether any stored response is equivalent to a current
   representation of the resource.
</t>
<t>
   One such validator is the timestamp given in a <x:ref>Last-Modified</x:ref>
   header field (<xref target="header.last-modified"/>), which can be used in an
   <x:ref>If-Modified-Since</x:ref> header field for response validation, or
   in an <x:ref>If-Unmodified-Since</x:ref> or <x:ref>If-Range</x:ref> header
   field for representation selection (i.e., the client is referring
   specifically to a previously obtained representation with that timestamp).
</t>
<t>
   Another validator is the entity-tag given in an <x:ref>ETag</x:ref> header
   field (<xref target="header.etag"/>). One or more entity-tags, indicating one or more
   stored responses, can be used in an <x:ref>If-None-Match</x:ref> header
   field for response validation, or in an <x:ref>If-Match</x:ref> or
   <x:ref>If-Range</x:ref> header field for representation selection (i.e.,
   the client is referring specifically to one or more previously obtained
   representations with the listed entity-tags).
</t>
</section>

<section title="Handling a Received Validation Request" anchor="validation.received">
<t>
   Each client in the request chain may have its own cache, so it is common
   for a cache at an intermediary to receive conditional requests from other
   (outbound) caches. Likewise, some user agents make use of conditional
   requests to limit data transfers to recently modified representations or to
   complete the transfer of a partially retrieved representation.
</t>
<t>
   If a cache receives a request that can be satisfied by reusing one of
   its stored <x:ref>200 (OK)</x:ref> or <x:ref>206 (Partial Content)</x:ref>
   responses, the cache &SHOULD; evaluate any applicable conditional header
   field preconditions received in that request with respect to the
   corresponding validators contained within the selected response.
   A cache &MUST-NOT; evaluate conditional header fields that are only
   applicable to an origin server, found in a request with semantics that
   cannot be satisfied with a cached response, or applied to a target resource
   for which it has no stored responses; such preconditions are likely
   intended for some other (inbound) server.
</t>
<t>
   The proper evaluation of conditional requests by a cache depends on the
   received precondition header fields and their precedence, as defined in
   <xref target="precedence"/>. The <x:ref>If-Match</x:ref> and
   <x:ref>If-Unmodified-Since</x:ref> conditional header fields are not
   applicable to a cache.
</t>
<t>
   A request containing an <x:ref>If-None-Match</x:ref> header field
   (<xref target="header.if-none-match"/>) indicates that the client wants to validate one
   or more of its own stored responses in comparison to whichever stored
   response is selected by the cache.
   If the field-value is "*", or if the field-value is a list of entity-tags
   and at least one of them matches the entity-tag of the selected stored
   response, a cache recipient &SHOULD; generate a
   <x:ref>304 (Not Modified)</x:ref> response (using the metadata of the
   selected stored response) instead of sending that stored response.
</t>
<t>
   When a cache decides to revalidate its own stored responses for a
   request that contains an <x:ref>If-None-Match</x:ref> list of entity-tags,
   the cache &MAY; combine the received list with a list of entity-tags
   from its own stored set of responses (fresh or stale) and send the union of
   the two lists as a replacement <x:ref>If-None-Match</x:ref> header
   field value in the forwarded request.
   If a stored response contains only partial content, the
   cache &MUST-NOT; include its entity-tag in the union unless the request is
   for a range that would be fully satisfied by that partial stored response.
   If the response to the forwarded request is
   <x:ref>304 (Not Modified)</x:ref> and has an ETag header field value with
   an entity-tag that is not in the client's list, the cache &MUST;
   generate a <x:ref>200 (OK)</x:ref> response for the client by reusing its
   corresponding stored response, as updated by the 304 response metadata
   (<xref target="freshening.responses"/>).
</t>
<t>
   If an <x:ref>If-None-Match</x:ref> header field is not present, a request
   containing an <x:ref>If-Modified-Since</x:ref> header field
   (<xref target="header.if-modified-since"/>) indicates that the client wants to validate
   one or more of its own stored responses by modification date.
   A cache recipient &SHOULD; generate a <x:ref>304 (Not Modified)</x:ref>
   response (using the metadata of the selected stored response) if one of the
   following cases is true:
   1) the selected stored response has a <x:ref>Last-Modified</x:ref>
   field-value that is earlier than or equal to the conditional timestamp;
   2) no <x:ref>Last-Modified</x:ref> field is present in the selected stored
   response, but it has a <x:ref>Date</x:ref> field-value that is earlier than
   or equal to the conditional timestamp; or,
   3) neither <x:ref>Last-Modified</x:ref> nor <x:ref>Date</x:ref> is present
   in the selected stored response, but the cache recorded it as having been
   received at a time earlier than or equal to the conditional timestamp.
</t>
<t>
   A cache that implements partial responses to range requests, as defined in
   <xref target="header.range"/>, also needs to evaluate a received <x:ref>If-Range</x:ref> header
   field (<xref target="header.if-range"/>) with respect to its selected stored response.
</t>
</section>

<section title="Handling a Validation Response" anchor="validation.response">
<t>
   Cache handling of a response to a conditional request is dependent upon its
   status code:
</t>
<ul>
   <li>
     A <x:ref>304 (Not Modified)</x:ref> response status code indicates
     that the stored response can be updated and reused; see <xref
     target="freshening.responses"/>.
   </li>
   <li>
     A full response (i.e., one with a payload body) indicates that none
     of the stored responses nominated in the conditional request is
     suitable. Instead, the cache &MUST; use the full response to
     satisfy the request and &MAY; replace the stored response(s).
   </li>
   <li>
     However, if a cache receives a <x:ref>5xx (Server Error)</x:ref>
     response while attempting to validate a response, it can either
     forward this response to the requesting client, or act as if the
     server failed to respond. In the latter case, the cache &MAY; send a
     previously stored response (see <xref
     target="serving.stale.responses" />).
   </li>
</ul>
</section>

<section title="Freshening Stored Responses upon Validation" anchor="freshening.responses">
<t>
   When a cache receives a <x:ref>304 (Not Modified)</x:ref> response and
   already has one or more stored <x:ref>200 (OK)</x:ref> responses for the
   applicable cache key, the cache needs to identify which (if any) are to
   be updated by the new information provided, and then do so.
</t>
<iref item="strong validator" />
<t>
   The stored response(s) to update are identified by using the first match
   (if any) of the following:
</t>
<ul>
   <li>
     If the new response contains a <x:dfn>strong validator</x:dfn> (see
     <xref target="weak.and.strong.validators"/>), then that strong validator
     identifies the selected representation for update. All of the stored
     responses with the same strong validator are identified for update. If
     none of the stored responses contain the same strong validator, then the
     cache &MUST-NOT; use the new response to update any stored responses.
   </li>
   <li>
     If the new response contains a weak validator and that validator
     corresponds to one of the cache's stored responses, then the most
     recent of those matching stored responses is identified for update.
   </li>
   <li>
     If the new response does not include any form of validator (such as in
     the case where a client generates an If-Modified-Since request from a
     source other than the Last-Modified response header field), and there is
     only one stored response, and that stored response also lacks a
     validator, then that stored response is identified for update.
   </li>
</ul>
<t>
   For each stored response identified for update, the cache &MUST; use the
   header fields provided in the <x:ref>304 (Not Modified)</x:ref> response
   to replace all instances of the corresponding header fields in the stored
   response.
</t>
</section>

<section title="Freshening Responses with HEAD" anchor="head.effects">
<t>
   A response to the HEAD method is identical to what an equivalent request
   made with a GET would have been, except it lacks a body. This property
   of HEAD responses can be used to invalidate or update a cached GET
   response if the more efficient conditional GET request mechanism is not
   available (due to no validators being present in the stored response) or
   if transmission of the representation body is not desired even if it has
   changed.
</t>
<t>
   When a cache makes an inbound HEAD request for a given request target and
   receives a <x:ref>200 (OK)</x:ref> response, the cache &SHOULD; update or
   invalidate each of its stored GET responses that could have been selected
   for that request (see <xref target="caching.negotiated.responses"/>).
</t>
<t>
   For each of the stored responses that could have been selected, if the
   stored response and HEAD response have matching values for any received
   validator fields (<x:ref>ETag</x:ref> and <x:ref>Last-Modified</x:ref>)
   and, if the HEAD response has a <x:ref>Content-Length</x:ref> header field,
   the value of <x:ref>Content-Length</x:ref> matches that of the stored
   response, the cache &SHOULD; update the stored response as described below;
   otherwise, the cache &SHOULD; consider the stored response to be stale.
</t>
<t>
   If a cache updates a stored response with the metadata provided in a HEAD
   response, the cache &MUST; use the header fields provided in the HEAD
   response to replace all instances of the corresponding header fields in
   the stored response and append new header fields to the stored response's
   header section unless otherwise restricted by the
   <x:ref>Cache-Control</x:ref> header field.
</t>
</section>
</section>


<section title="Invalidation" anchor="invalidation">
<t>
   Because unsafe request methods (<xref target="safe.methods"/>) such as PUT, POST or DELETE
   have the potential for changing state on the origin server, intervening
   caches can use them to keep their contents up to date.
</t>
<t>
   A cache &MUST; invalidate the effective Request URI
   (<xref target="effective.request.uri"/>) as well as the URI(s) in the
   <x:ref>Location</x:ref> and <x:ref>Content-Location</x:ref> response header
   fields (if present) when a non-error status code is received in response to
   an unsafe request method.
</t>
<t>
   However, a cache &MUST-NOT; invalidate a URI from a <x:ref>Location</x:ref>
   or <x:ref>Content-Location</x:ref> response header field if the host part of
   that URI differs from the host part in the effective request URI
   (<xref target="effective.request.uri"/>). This helps prevent denial-of-service attacks.
</t>
<t>
   A cache &MUST; invalidate the effective request URI 
   (<xref target="effective.request.uri"/>) when it receives a non-error response 
   to a request with a method whose safety is unknown.
</t>
<t>
   Here, a "non-error response" is one with a <x:ref>2xx (Successful)</x:ref>
   or <x:ref>3xx (Redirection)</x:ref> status code. "Invalidate" means that
   the cache will either remove all stored responses related to the effective
   request URI or will mark these as "invalid" and in need of a mandatory
   validation before they can be sent in response to a subsequent request.
</t>
<t>
   Note that this does not guarantee that all appropriate responses are
   invalidated. For example, a state-changing request might invalidate
   responses in the caches it travels through, but relevant responses still
   might be stored in other caches that it has not.
</t>
</section>
</section>



<section title="Header Field Definitions" anchor="header.field.definitions">
<t>
   This section defines the syntax and semantics of HTTP header fields
   related to caching.
</t>
<?BEGININC build/draft-ietf-httpbis-cache-latest.iana-headers ?>
<!--AUTOGENERATED FROM extract-header-defs.xslt, do not edit manually-->
<table align="left" anchor="iana.header.registration.table">
   <thead>
      <tr>
         <th>Header Field Name</th>
         <th>Status</th>
         <th>Reference</th>
      </tr>
   </thead>
   <tbody>
      <tr>
         <td>Age</td>
         <td>standard</td>
         <td>
            <xref target="header.age"/>
         </td>
      </tr>
      <tr>
         <td>Cache-Control</td>
         <td>standard</td>
         <td>
            <xref target="header.cache-control"/>
         </td>
      </tr>
      <tr>
         <td>Expires</td>
         <td>standard</td>
         <td>
            <xref target="header.expires"/>
         </td>
      </tr>
      <tr>
         <td>Pragma</td>
         <td>standard</td>
         <td>
            <xref target="header.pragma"/>
         </td>
      </tr>
      <tr>
         <td>Warning</td>
         <td>obsoleted</td>
         <td>
            <xref target="header.warning"/>
         </td>
      </tr>
   </tbody>
</table>
<!--(END)-->

<?ENDINC build/draft-ietf-httpbis-cache-latest.iana-headers ?>

<section title="Age" anchor="header.age">
   <iref item="Age header field" primary="true" x:for-anchor="" />
   <x:anchor-alias value="Age"/>
   <x:anchor-alias value="age-value"/>
<t>
   The "Age" header field conveys the sender's estimate of the amount
   of time since the response was generated or successfully validated at the
   origin server. Age values are calculated as specified in <xref
   target="age.calculations" />.
</t>
<sourcecode type="abnf7230"><iref primary="true" item="Grammar" subitem="Age"/>
  <x:ref>Age</x:ref> = <x:ref>delta-seconds</x:ref>
</sourcecode>
<t>
   The Age field-value is a non-negative integer, representing time in seconds
   (see <xref target="delta-seconds"/>).
</t>
<t>
   The presence of an Age header field implies that the response was not
   generated or validated by the origin server for this request. However,
   lack of an Age header field does not imply the origin was contacted, since
   the response might have been received from an HTTP/1.0 cache that does not
   implement Age.
</t>
</section>

<section title="Cache-Control" anchor="header.cache-control">
   <iref item="Cache-Control header field" primary="true" x:for-anchor="" />
   <x:anchor-alias value="Cache-Control"/>
   <x:anchor-alias value="cache-directive"/>
<t>
   The "Cache-Control" header field is used to specify directives for caches
   along the request/response chain. Such cache directives are unidirectional
   in that the presence of a directive in a request does not imply that the
   same directive is present in the response, or to be repeated in it.
</t>
<t>
   See <xref target="cache.control.extensions"/> for information about how
   Cache-Control directives defined elsewhere are handled.
</t>
<aside>
   <t>
       &Note; Some HTTP/1.0 caches might not implement Cache-Control.
   </t>
</aside>
<t>
   A proxy, whether or not it implements a cache, &MUST; pass cache directives
   through in forwarded messages, 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>
<t>
   Cache directives are identified by a token, to be compared case-insensitively,
   and have an optional argument, that can use both token and quoted-string
   syntax. For the directives defined below that define arguments, recipients
   ought to accept both forms, even if one is documented to be preferred. For
   any directive not defined by this specification, a recipient &MUST; accept
   both forms. 
</t>
<sourcecode type="abnf7230"><iref primary="true" item="Grammar" subitem="Cache-Control"/><iref primary="true" item="Grammar" subitem="cache-directive"/>
  <x:ref>Cache-Control</x:ref>   = 1#<x:ref>cache-directive</x:ref>

  <x:ref>cache-directive</x:ref> = <x:ref>token</x:ref> [ "=" ( <x:ref>token</x:ref> / <x:ref>quoted-string</x:ref> ) ]
</sourcecode>
<t>
   For the cache directives defined below, no argument is defined (nor allowed)
   unless stated otherwise.
</t>
<?BEGININC build/draft-ietf-httpbis-cache-latest.cache-directives ?>
<!--AUTOGENERATED FROM extract-cache-directives-defs.xslt, do not edit manually-->
<table align="left" anchor="iana.cache.directive.registration.table">
   <thead>
      <tr>
         <th>Cache Directive</th>
         <th>Reference</th>
      </tr>
   </thead>
   <tbody>
      <tr>
         <td>max-age</td>
         <td>
            <xref target="cache-request-directive.max-age"/>, <xref target="cache-response-directive.max-age"/>
         </td>
      </tr>
      <tr>
         <td>max-stale</td>
         <td>
            <xref target="cache-request-directive.max-stale"/>
         </td>
      </tr>
      <tr>
         <td>min-fresh</td>
         <td>
            <xref target="cache-request-directive.min-fresh"/>
         </td>
      </tr>
      <tr>
         <td>must-revalidate</td>
         <td>
            <xref target="cache-response-directive.must-revalidate"/>
         </td>
      </tr>
      <tr>
         <td>no-cache</td>
         <td>
            <xref target="cache-request-directive.no-cache"/>, <xref target="cache-response-directive.no-cache"/>
         </td>
      </tr>
      <tr>
         <td>no-store</td>
         <td>
            <xref target="cache-request-directive.no-store"/>, <xref target="cache-response-directive.no-store"/>
         </td>
      </tr>
      <tr>
         <td>no-transform</td>
         <td>
            <xref target="cache-request-directive.no-transform"/>, <xref target="cache-response-directive.no-transform"/>
         </td>
      </tr>
      <tr>
         <td>only-if-cached</td>
         <td>
            <xref target="cache-request-directive.only-if-cached"/>
         </td>
      </tr>
      <tr>
         <td>private</td>
         <td>
            <xref target="cache-response-directive.private"/>
         </td>
      </tr>
      <tr>
         <td>proxy-revalidate</td>
         <td>
            <xref target="cache-response-directive.proxy-revalidate"/>
         </td>
      </tr>
      <tr>
         <td>public</td>
         <td>
            <xref target="cache-response-directive.public"/>
         </td>
      </tr>
      <tr>
         <td>s-maxage</td>
         <td>
            <xref target="cache-response-directive.s-maxage"/>
         </td>
      </tr>
   </tbody>
</table>
<!--(END)-->

<?ENDINC build/draft-ietf-httpbis-cache-latest.cache-directives ?>

<section title="Request Cache-Control Directives" anchor="cache-request-directive">

<t>
   This section defines cache request directives. They are advisory; caches
   &MAY; implement them, but are not required to.
</t>


<section title="max-age" anchor="cache-request-directive.max-age">
   <iref item="max-age (cache directive)" primary="true" />
<t>
   Argument syntax:
</t>
<ul empty="true">
  <li><x:ref>delta-seconds</x:ref> (see <xref target="delta-seconds"/>)</li>
</ul>
<t>
   The "max-age" request directive indicates that the client prefers a
   response whose age is less than or equal to the specified number of
   seconds. Unless the max-stale request directive is also present, the
   client does not wish to receive a stale response.
</t>
<t>
   This directive uses the token form of the argument syntax:
   e.g., 'max-age=5' not 'max-age="5"'. A sender &SHOULD-NOT; generate the
   quoted-string form.
</t>
</section>

<section title="max-stale" anchor="cache-request-directive.max-stale">
   <iref item="max-stale (cache directive)" primary="true" />
<t>
   Argument syntax:
</t>
<ul empty="true">
  <li><x:ref>delta-seconds</x:ref> (see <xref target="delta-seconds"/>)</li>
</ul>
<t>
   The "max-stale" request directive indicates that the client is willing to
   accept a response that has exceeded its freshness lifetime. If a value is
   present, then the client is willing to accept a response that has exceeded
   its freshness lifetime 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.
</t>
<t>
   This directive uses the token form of the argument syntax:
   e.g., 'max-stale=10' not 'max-stale="10"'. A sender &SHOULD-NOT; generate
   the quoted-string form.
</t>
</section>

<section title="min-fresh" anchor="cache-request-directive.min-fresh">
   <iref item="min-fresh (cache directive)" primary="true" />
<t>
   Argument syntax:
</t>
<ul empty="true">
  <li><x:ref>delta-seconds</x:ref> (see <xref target="delta-seconds"/>)</li>
</ul>
<t>
   The "min-fresh" request directive indicates that the client prefers 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>
<t>
   This directive uses the token form of the argument syntax:
   e.g., 'min-fresh=20' not 'min-fresh="20"'. A sender &SHOULD-NOT; generate
   the quoted-string form.
</t>
</section>

<section title="no-cache" anchor="cache-request-directive.no-cache">
   <iref item="no-cache (cache directive)" primary="true" />
<t>
   The "no-cache" request directive indicates that the client prefers
   stored response not be used to satisfy the request without successful
   validation on the origin server.
</t>
</section>
 
<section title="no-store" anchor="cache-request-directive.no-store">
   <iref item="no-store (cache directive)" primary="true" />
<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 private 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 might be
   vulnerable to eavesdropping.
</t>
<t>
   Note that if a request containing this directive is satisfied from a
   cache, the no-store request directive does not apply to the already
   stored response.
</t>
</section>

<section title="no-transform" anchor="cache-request-directive.no-transform">
   <iref item="no-transform (cache directive)" primary="true" />
<t>
   The "no-transform" request directive indicates that the client is asking
   for intermediares (whether or not they implement a cache) to avoid
   transforming the payload, as defined in <xref
   target="message.transformations"/>.
</t>
</section>

<section title="only-if-cached" anchor="cache-request-directive.only-if-cached">
   <iref item="only-if-cached (cache directive)" primary="true" />
<t>
   The "only-if-cached" request directive indicates that the client only
   wishes to obtain a stored response. Caches that honor this request
   directive &SHOULD;, upon receiving it, either respond using a stored
   response that is consistent with the other constraints of the request, or
   respond with a <x:ref>504 (Gateway Timeout)</x:ref> status code.
</t>
</section>
</section>

<section title="Response Cache-Control Directives" anchor="cache-response-directive">
   <x:anchor-alias value="cache-response-directive" />

<t>
   This section defines cache response directives. A cache &MUST; obey the
   requirements of the Cache-Control directives defined in this section.
</t>

<section title="must-revalidate" anchor="cache-response-directive.must-revalidate">
   <iref item="must-revalidate (cache directive)" primary="true" />
<t>
   The "must-revalidate" response directive indicates that once it has become
   stale, the response &MUST-NOT; be used to satisfy any other request
   without forwarding it for validation and receiving a successful response;
   see <xref target="validation.model"/>.
</t>
<t>
   The must-revalidate directive is necessary to support reliable operation
   for certain protocol features. In all circumstances a cache &MUST; obey
   the must-revalidate directive; in particular, if a cache is disconnected,
   it &MUST; generate a <x:ref>504 (Gateway Timeout)</x:ref> response.
</t>
<t>
   The must-revalidate directive ought to be used by servers if and only
   if failure to validate a request on the representation could result in
   incorrect operation, such as a silently unexecuted financial
   transaction.
</t>
<t>
   The must-revalidate directive also has the effect of allowing a stored
   response to be used to satisfy a request with an Authorization header
   field; see <xref target="caching.authenticated.responses"/>.
</t>
</section>

<section title="no-cache" anchor="cache-response-directive.no-cache">
   <iref item="no-cache (cache directive)" primary="true" />
<t>
   Argument syntax:
</t>
<ul empty="true">
  <li>#<x:ref>field-name</x:ref></li>
</ul>
<t>
   The "no-cache" response directive indicates that the response &MUST-NOT;
   be used to satisfy any other request without forwarding it for validation
   and receiving a successful response; see <xref target="validation.model"/>.
</t>
<t>
   This allows an origin server to prevent a cache from using
   it to satisfy a request without contacting it, even by caches that have
   been configured to send stale responses.
</t>
<t>
   If the no-cache response directive specifies one or more field-names,
   then a cache &MAY; use the response to satisfy a subsequent request,
   subject to any other restrictions on caching. However, any header fields
   in the response that have the field-name(s) listed &MUST-NOT; be sent
   in the response to a subsequent request without successful revalidation
   with 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>
   The field-names given are not limited to the set of header
   fields defined by this specification. Field names are case-insensitive.
</t>
<t>
   This directive uses the quoted-string form of the argument syntax.
   A sender &SHOULD-NOT; generate the token form (even if quoting appears not
   to be needed for single-entry lists).
</t>
<t>
   &Note; Although it has been back-ported to many implementations, some
   HTTP/1.0 caches will not recognize or obey this directive. Also, no-cache
   response directives with field-names are often handled by caches
   as if an unqualified no-cache directive was received; i.e., the special
   handling for the qualified form is not widely implemented.
</t>
</section>

<section title="no-store" anchor="cache-response-directive.no-store">
   <iref item="no-store (cache directive)" primary="true" />
<t>
   The "no-store" response directive indicates that a cache &MUST-NOT; store
   any part of either the immediate request or response, and &MUST-NOT; use
   the response to satisfy any other request.
</t>
<t>
   This directive applies to both private 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 might be
   vulnerable to eavesdropping.
</t>
</section>

<section title="no-transform" anchor="cache-response-directive.no-transform">
   <iref item="no-transform (cache directive)" primary="true" />
<t>
   The "no-transform" response directive indicates that an intermediary
   (regardless of whether it implements a cache) &MUST-NOT; transform the
   payload, as defined in <xref target="message.transformations"/>.
</t>
</section>

<section title="public" anchor="cache-response-directive.public">
   <iref item="public (cache directive)" primary="true" />
<t>
   The "public" response directive indicates that any cache &MAY; store the
   response, even if the response would normally be non-cacheable or cacheable
   only within a private cache. (See <xref
   target="caching.authenticated.responses"/> for additional details related to
   the use of public in response to a request containing
   <x:ref>Authorization</x:ref>, and <xref target="response.cacheability"/> for
   details of how public affects responses that would normally not be stored,
   due to their status codes not being defined as heuristically cacheable; see
   <xref target="heuristic.freshness"/>.)
</t>
</section>

<section title="private" anchor="cache-response-directive.private">
   <iref item="private (cache directive)" primary="true" />
<t>
   Argument syntax:
</t>
<ul empty="true">
  <li>#<x:ref>field-name</x:ref></li>
</ul>
<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 cache &MAY; store the response and reuse it for later requests,
   even if the response would normally be non-cacheable.
</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 header fields. That is, a shared cache &MUST-NOT; store
   the specified field-names(s), whereas it &MAY; store the remainder of the
   response message.
</t>
<t>
   The field-names given are not limited to the set of header
   fields defined by this specification. Field names are case-insensitive.
</t>
<t>
   This directive uses the quoted-string form of the argument syntax.
   A sender &SHOULD-NOT; generate the token form (even if quoting appears not
   to be needed for single-entry lists).
</t>
<t>
   &Note; This usage of the word "private" only controls
   where the response can be stored; it cannot ensure the privacy of the
   message content. Also, private response directives with field-names are
   often handled by caches as if an unqualified private directive
   was received; i.e., the special handling for the qualified form is not
   widely implemented.
</t>
</section>

<section title="proxy-revalidate" anchor="cache-response-directive.proxy-revalidate">
   <iref item="proxy-revalidate (cache directive)" primary="true" />
<t>
   The "proxy-revalidate" response directive has the same meaning as the
   must-revalidate response directive, except that it does not apply to
   private caches.
</t>
</section>

<section title="max-age" anchor="cache-response-directive.max-age">
   <iref item="max-age (cache directive)" primary="true" />
<t>
   Argument syntax:
</t>
<ul empty="true">
  <li><x:ref>delta-seconds</x:ref> (see <xref target="delta-seconds"/>)</li>
</ul>
<t>
   The "max-age" response directive indicates that the response is to be
   considered stale after its age is greater than the specified number of
   seconds.
</t>
<t>
   This directive uses the token form of the argument syntax:
   e.g., 'max-age=5' not 'max-age="5"'. A sender &SHOULD-NOT; generate the
   quoted-string form.
</t>
</section>      

<section title="s-maxage" anchor="cache-response-directive.s-maxage">
   <iref item="s-maxage (cache directive)" primary="true" />
<t>
   Argument syntax:
</t>
<ul empty="true">
  <li><x:ref>delta-seconds</x:ref> (see <xref target="delta-seconds"/>)</li>
</ul>
<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 <x:ref>Expires</x:ref>
   header field. The s-maxage directive also implies the semantics of the
   proxy-revalidate response directive.
</t>
<t>
   The s-maxage directive also has the effect of allowing a stored
   response to be used to satisfy a request with an Authorization header
   field; see <xref target="caching.authenticated.responses"/>.
</t>
<t>
   This directive uses the token form of the argument syntax:
   e.g., 's-maxage=10' not 's-maxage="10"'. A sender &SHOULD-NOT; generate
   the quoted-string form.
</t>
</section>
</section>

<section title="Cache Control Extensions" anchor="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.
   A cache &MUST; ignore unrecognized cache directives.
</t>
<t>
   Informational extensions (those that do not require a change in cache
   behavior) can be added without changing the semantics of other directives.
</t>
<t>   
   Behavioral extensions are designed to work by acting as modifiers to the
   existing base of cache directives.
   Both the new directive and the old directive are supplied, such that
   applications that do not understand the new directive will default to the
   behavior specified by the old directive, and those that understand the
   new directive will recognize it as modifying the requirements associated
   with the old directive. In this way, extensions to the existing
   cache-control directives can be made without breaking deployed caches.
</t>
<t>
   For example, consider a hypothetical new response directive called
   "community" that acts as a modifier to the private directive: in addition
   to private caches, any cache that is shared only by members of the named
   community is allowed to 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>
<artwork type="example">
  Cache-Control: private, community="UCI"
</artwork>
<t>
   A cache that recognizes such a community cache-extension could broaden its
   behavior in accordance with that extension.  A cache that does not
   recognize the community cache-extension would ignore it and adhere to the
   private directive.
</t>
<t>
   New extension directives ought to consider defining:
</t>
<ul>
  <li>What it means for a directive to be specified multiple times,</li>
  <li>When the directive does not take an argument, what it means when an
  argument is present,</li>
  <li>When the directive requires an argument, what it means when it is
  missing,</li>
  <li>Whether the directive is specific to requests, responses, or able
  to be used in either.</li>
</ul>
</section>

<section title="Cache Directive Registry" anchor="cache.directive.registry">
<t>
   The "Hypertext Transfer Protocol (HTTP) Cache Directive Registry" defines the namespace for the
   cache directives. It has been created and is now maintained at
   <eref target="https://www.iana.org/assignments/http-cache-directives"/>.
</t>
<t>
   A registration &MUST; include the following fields:
</t>
<ul>
  <li>Cache Directive Name</li>
  <li>Pointer to specification text</li>
</ul>
<t>
   Values to be added to this namespace require IETF Review (see <xref
   target="RFC8126" x:fmt="," x:sec="4.8"/>).
</t>
</section>
</section>

<section title="Expires" anchor="header.expires">
   <iref item="Expires header field" primary="true" x:for-anchor="" />
   <x:anchor-alias value="Expires"/>
<t>
   The "Expires" header field 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 Expires value is an HTTP-date timestamp, as defined in <xref target="http.date"/>.
</t>
<sourcecode type="abnf7230"><iref primary="true" item="Grammar" subitem="Expires"/>
  <x:ref>Expires</x:ref> = <x:ref>HTTP-date</x:ref>
</sourcecode>
<t>
   For example
</t>
<artwork type="example">
  Expires: Thu, 01 Dec 1994 16:00:00 GMT
</artwork>
<t>
   A cache recipient &MUST; interpret invalid date formats, especially the
   value "0", as representing a time in the past (i.e., "already expired").
</t>
<t>
   If a response includes a <x:ref>Cache-Control</x:ref> field with
   the max-age directive (<xref target="cache-response-directive.max-age"/>),
   a recipient &MUST; ignore the Expires field.
   Likewise, if a response includes the s-maxage directive
   (<xref target="cache-response-directive.s-maxage" />), a shared cache
   recipient &MUST; ignore the Expires field. In both these cases, the value
   in Expires is only intended for recipients that have not yet implemented
   the Cache-Control field.
</t>
<t>
   An origin server without a clock &MUST-NOT; generate an Expires field
   unless its value represents a fixed time in the past (always expired)
   or its value has been associated with the resource by a system or user
   with a reliable clock.
</t>
<t>
   Historically, HTTP required the Expires field-value to be no more than a 
   year in the future. While longer freshness lifetimes are no longer 
   prohibited, extremely large values have been demonstrated to cause 
   problems (e.g., clock overflows due to use of 32-bit integers for 
   time values), and many caches will evict a response far sooner than 
   that.
</t>
</section>

<section title="Pragma" anchor="header.pragma">
   <iref item="Pragma header field" primary="true" x:for-anchor="" />
   <x:anchor-alias value="extension-pragma"/>
   <x:anchor-alias value="Pragma"/>
   <x:anchor-alias value="pragma-directive"/>
<t>
   The "Pragma" header field was defined for HTTP/1.0 caches, so that clients
   could specify a "no-cache" request (as <x:ref>Cache-Control</x:ref> was
   not defined until HTTP/1.1).
</t>
<t>
   However, support for Cache-Control is now widespread. As a result, this
   specification deprecates Pragma.
</t>

<aside>
   <t>
      &Note; Because the meaning of "Pragma: no-cache" in responses was never
      specified, it does not provide a reliable replacement for
      "Cache-Control: no-cache" in them.
   </t>
</aside>
</section>

<section title="Warning" anchor="header.warning">
   <rdf:Description>
     <status xmlns="urn:ietf:id:draft-ietf-httpbis-p2-semantics#">obsoleted</status>
   </rdf:Description>
   <iref item="Warning header field" primary="true" x:for-anchor="" />
   <x:anchor-alias value="Warning"/>
<t>
   The "Warning" header field was used to carry additional information
   about the status or transformation of a message that might not be reflected
   in the status code. This specification obsoletes it, as it is not widely
   generated or surfaced to users. The information it carried can be gleaned
   from examining other header fields, such as <x:ref>Age</x:ref>.
</t>
</section>
</section>

<section title="Relationship to Applications" anchor="history.lists">
<t>
   Applications using HTTP often specify additional forms of caching. For
   example, Web browsers often have history mechanisms such as "Back" buttons
   that can be used to redisplay a representation retrieved earlier in a
   session.
</t>
<t>
   Likewise, some Web browsers implement caching of images and other assets
   within a page view; they may or may not honor HTTP caching semantics.
</t>
<t>
   The requirements in this specification do not necessarily apply to how
   applications use data after it is retrieved from a HTTP cache. That is, a
   history mechanism can display a previous representation even if it has
   expired, and an application can use cached data in other ways beyond its
   freshness lifetime.
</t>
<t>
   This does not prohibit the application from taking HTTP caching into
   account; for example, a history mechanism might tell the user that a view
   is stale, or it might honor cache directives (e.g., Cache-Control:
   no-store).
</t>
</section>


<section title="Security Considerations" anchor="security.considerations">
<t>
   This section is meant to inform developers, information providers, and
   users of known security concerns specific to HTTP caching.
   More general security considerations are addressed in HTTP messaging
   <xref target="Messaging"/> and semantics <xref target="Semantics"/>.
</t>
<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
   need to be protected as sensitive information.
</t>

<section title="Cache Poisoning">
<t>
   Various attacks might be amplified by being stored in a shared cache. Such
   "cache poisoning" attacks use the cache to distribute a malicious payload
   to many clients, and are especially effective when an attacker can use
   implementation flaws, elevated privileges, or other techniques to insert
   such a response into a cache.
</t>
<t>
   One common attack vector for cache poisoning is to exploit differences in
   message parsing on proxies and in user agents; see <xref
   target="message.body.length"/> for the relevant requirements regarding
   HTTP/1.1.
</t>
</section>

<section title="Timing Attacks" anchor="security.timing">
<t>
   Because one of the primary uses of a cache is to optimise performance,
   its use can "leak" information about what resources have been previously
   requested. 
</t>
<t>
   For example, if a user visits a site and their browser caches some of its
   responses, and then navigates to a second site, that site can attempt to
   load responses that it knows exists on the first site. If they load very
   quickly, it can be assumed that the user has visited that site, or even
   a specific page on it.
</t>
<t>
   Such "timing attacks" can be mitigated by adding more information to the
   cache key, such as the identity of the referring site (to prevent the
   attack described above). This is sometimes called "double keying."
</t>
</section>

<section title="Caching of Sensitive Information">
<t>
   Implementation and deployment flaws (as well as misunderstanding of cache
   operation) might lead to caching of sensitive information (e.g.,
   authentication credentials) that is thought to be private, exposing it to
   unauthorized parties.
</t>
<t>
   Note that the Set-Cookie response header field <xref target="RFC6265"/>
   does not inhibit caching; a cacheable response with a Set-Cookie header
   field can be (and often is) used to satisfy subsequent requests to caches.
   Servers who wish to control caching of these responses are encouraged to
   emit appropriate Cache-Control response header fields.
</t>
</section>
</section>


<section title="IANA Considerations" anchor="iana.considerations">
<t>
   The change controller for the following registrations is:
   "IETF (iesg@ietf.org) - Internet Engineering Task Force".
</t>

<section title="Header Field Registration" anchor="header.field.registration">
<t>
   Please update the "Hypertext Transfer Protocol (HTTP) Header Field
   Registry" registry at <eref
   target="https://www.iana.org/assignments/http-headers"/> with the
   header field names listed in the two tables of <xref
   target="header.field.definitions"/>.
</t>
</section>

<section title="Cache Directive Registration" anchor="cache.directive.registration">
<t>
   Please update the
   "Hypertext Transfer Protocol (HTTP) Cache Directive Registry"
   at <eref target="https://www.iana.org/assignments/http-cache-directives"/>
   with the registration procedure of <xref target="cache.directive.registry"/>
   and the cache directive names summarized in the table of
   <xref target="header.cache-control"/>.
</t>
</section>

<section title="Warn Code Registry" anchor="warn.code.registration">
<t>
   Please add a note to the "Hypertext Transfer Protocol (HTTP) Warn Codes"
   registry at <eref
   target="https://www.iana.org/assignments/http-warn-codes"/> to the effect
   that Warning is obsoleted.
</t>
</section>
</section>
</middle>

<back>
<references title="Normative References">

  <reference anchor="Messaging">
    <x:source basename="draft-ietf-httpbis-messaging-latest" href="draft-ietf-httpbis-messaging-latest.xml">
      <x:has anchor="associating.response.to.request"/>
      <x:has anchor="message.body.length"/>
    </x:source>
  </reference>

  <reference anchor="Semantics">
    <x:source basename="draft-ietf-httpbis-semantics-latest" href="draft-ietf-httpbis-semantics-latest.xml">
      <x:defines>Content-Length</x:defines>
      <x:has anchor="GET"/>
      <x:has anchor="abnf.extension"/>
      <x:has anchor="combining.byte.ranges"/>
      <x:has anchor="effective.request.uri"/>
      <x:has anchor="field.components"/>
      <x:has anchor="header.authorization"/>
      <x:has anchor="header.date"/>
      <x:has anchor="header.etag"/>
      <x:has anchor="header.if-modified-since"/>
      <x:has anchor="header.if-none-match"/>
      <x:has anchor="header.if-range"/>
      <x:has anchor="header.last-modified"/>
      <x:has anchor="header.range"/>
      <x:has anchor="header.vary"/>
      <x:has anchor="header.via"/>
      <x:has anchor="http.date"/>
      <x:has anchor="field.names"/>
      <x:has anchor="field.values"/>
      <x:has anchor="message.transformations"/>
      <x:has anchor="overview.of.status.codes"/>
      <x:has anchor="precedence"/>
      <x:has anchor="preconditions"/>
      <x:has anchor="quoted.strings"/>
      <x:has anchor="safe.methods"/>
      <x:has anchor="tokens"/>
      <x:has anchor="uri"/>
      <x:has anchor="weak.and.strong.validators"/>
      <x:has anchor="whitespace"/>
      <x:has anchor="Semantics-acks" target="acks"/>
      <x:has anchor="Semantics-conformance" target="conformance"/>
    </x:source>
  </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="RFC3986">
   <front>
    <title abbrev='URI Generic Syntax'>Uniform Resource Identifier (URI): Generic Syntax</title>
    <author initials='T.' surname='Berners-Lee' fullname='Tim Berners-Lee'>
      <organization abbrev="W3C/MIT">World Wide Web Consortium</organization>
      <address>
         <email>timbl@w3.org</email>
         <uri>http://www.w3.org/People/Berners-Lee/</uri>
      </address>
    </author>
    <author initials='R.' surname='Fielding' fullname='Roy T. Fielding'>
      <organization abbrev="Day Software">Day Software</organization>
      <address>
        <email>fielding@gbiv.com</email>
        <uri>http://roy.gbiv.com/</uri>
      </address>
    </author>
    <author initials='L.' surname='Masinter' fullname='Larry Masinter'>
      <organization>Adobe</organization>
      <address>
        <email>LMM@acm.org</email>
        <uri>http://larry.masinter.net/</uri>
      </address>
    </author>
    <date month='January' year='2005'></date>
   </front>
   <seriesInfo name="STD" value="66"/>
   <seriesInfo name="RFC" value="3986"/>
  </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>
          <email>dcrocker@bbiw.net</email>
        </address>  
      </author>
      <author initials="P." surname="Overell" fullname="Paul Overell">
        <organization>THUS plc.</organization>
        <address>
          <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>

  <reference anchor="RFC7405">
    <front>
      <title>Case-Sensitive String Support in ABNF</title>
      <author initials="P." surname="Kyzivat" fullname="Dave Kyzivat">
        <address>
          <email>pkyzivat@alum.mit.edu</email>
        </address>
      </author>
      <date month="December" year="2014"/>
    </front>
    <seriesInfo name="RFC" value="7405"/>
  </reference>

  <reference anchor="USASCII">
    <front>
      <title>Coded Character Set -- 7-bit American Standard Code for Information Interchange</title>
      <author>
        <organization>American National Standards Institute</organization>
      </author>
      <date year="1986"/>
    </front>
    <seriesInfo name="ANSI" value="X3.4"/>
  </reference>

</references>

<references title="Informative References">

  <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="RFC7234">
    <front>
      <title>Hypertext Transfer Protocol (HTTP): Caching</title>
      <author initials="R." surname="Fielding" fullname="Roy T. Fielding" role="editor">
        <organization>Adobe</organization>
        <address><email>fielding@gbiv.com</email></address>
      </author>
      <author initials="M." surname="Nottingham" fullname="Mark Nottingham" role="editor">
        <organization>Akamai</organization>
        <address><email>mnot@mnot.net</email></address>
      </author>
      <author initials="J. F." surname="Reschke" fullname="Julian F. Reschke" role="editor">
        <organization abbrev="greenbytes">greenbytes GmbH</organization>
        <address><email>julian.reschke@greenbytes.de</email></address>
      </author>
      <date month="June" year="2014"/>
    </front>
    <seriesInfo name="RFC" value="7234"/>
  </reference>
  
  <reference anchor='RFC5861'>
    <front>
      <title abbrev="HTTP stale controls">HTTP Cache-Control Extensions for Stale Content</title>
      <author initials="M." surname="Nottingham" fullname="Mark Nottingham">
        <organization>Yahoo! Inc.</organization>
        <address><email>mnot@yahoo-inc.com</email></address>
      </author>
      <date month="April" year="2010"/>
    </front>
    <seriesInfo name='RFC' value='5861' />
  </reference>

  <reference anchor='RFC5905'>
    <front>
      <title>Network Time Protocol Version 4: Protocol and Algorithms Specification</title>
      <author initials='D.' surname='Mills' fullname='David L. Mills'/>
      <author initials='J.' surname='Martin' fullname='Jim Martin' role="editor"/>
      <author initials='J.' surname='Burbank' fullname='Jack Burbank'/>
      <author initials='W.' surname='Kasch' fullname='William Kasch'/>
      <date year='2010' month='June' />
    </front>
    <seriesInfo name='RFC' value='5905' />
  </reference>

  <reference anchor="RFC6265">
    <front>
      <title>HTTP State Management Mechanism</title>
      <author initials="A." surname="Barth" fullname="Adam Barth">
        <organization abbrev="U.C. Berkeley">
          University of California, Berkeley
        </organization>
        <address><email>abarth@eecs.berkeley.edu</email></address>
      </author>
      <date year="2011" month="April" />
    </front>
    <seriesInfo name="RFC" value="6265"/>
  </reference>

  <reference anchor="RFC8126">
    <front>
      <title>Guidelines for Writing an IANA Considerations Section in RFCs</title>
      <author initials="M." surname="Cotton" fullname="M. Cotton"/>
      <author initials="B." surname="Leiba" fullname="B. Leiba"/>
      <author initials="T." surname="Narten" fullname="T. Narten"/>
      <date year="2017" month="June" />
    </front>
    <seriesInfo name="BCP" value="26"/>
    <seriesInfo name="RFC" value="8126"/>
  </reference>
  
</references>

<?BEGININC build/draft-ietf-httpbis-cache-latest.abnf-appendix ?>
<section xmlns:x="http://purl.org/net/xml2rfc/ext" title="Collected ABNF" anchor="collected.abnf"><t>In the collected ABNF below, list rules are expanded as per <xref target="Semantics" x:rel="#abnf.extension"/>.</t><sourcecode type="abnf" name="draft-ietf-httpbis-cache-latest.parsed-abnf">
<x:ref>Age</x:ref> = delta-seconds

<x:ref>Cache-Control</x:ref> = [ cache-directive ] *( OWS "," OWS [ cache-directive
 ] )

<x:ref>Expires</x:ref> = HTTP-date

<x:ref>HTTP-date</x:ref> = &lt;HTTP-date, see <xref target="Semantics" x:fmt="," x:sec="10.1.1.1"/>&gt;

<x:ref>OWS</x:ref> = &lt;OWS, see <xref target="Semantics" x:fmt="," x:sec="4.3"/>&gt;

<x:ref>cache-directive</x:ref> = token [ "=" ( token / quoted-string ) ]

<x:ref>delta-seconds</x:ref> = 1*DIGIT

<x:ref>field-name</x:ref> = &lt;field-name, see <xref target="Semantics" x:fmt="," x:sec="4.2"/>&gt;

<x:ref>quoted-string</x:ref> = &lt;quoted-string, see <xref target="Semantics" x:fmt="," x:sec="4.2.3"/>&gt;

<x:ref>token</x:ref> = &lt;token, see <xref target="Semantics" x:fmt="," x:sec="4.2.3"/>&gt;
</sourcecode>
</section>
<?ENDINC build/draft-ietf-httpbis-cache-latest.abnf-appendix ?>

<section title="Changes from RFC 7234" anchor="changes.from.rfc.7234">
<t>
   The Warning response header was obsoleted. Much of the information
   supported by Warning could be gleaned by examining the response), and the
   remaining warn-codes &mdash; although potentially useful &mdash; were entirely
   advisory, and in practice were not added by caches or intermediaries.
   (<xref target="header.warning"/>)
</t>
</section>

<section title="Change Log" anchor="change.log" removeInRFC="true">

<section title="Between RFC7234 and draft 00" anchor="changes.since.publication.as.rfc">
<t>
  The changes were purely editorial:
</t>
<ul>
  <li>Change boilerplate and abstract to indicate the "draft" status, and update references to ancestor specifications.</li>
  <li>Remove version "1.1" from document title, indicating that this specification applies to all HTTP versions.</li>
  <li>Adjust historical notes.</li>
  <li>Update links to sibling specifications.</li>
  <li>Replace sections listing changes from RFC 2616 by new empty sections referring to RFC 723x.</li>
  <li>Remove acknowledgements specific to RFC 723x.</li>
  <li>Move "Acknowledgements" to the very end and make them unnumbered.</li>
</ul>
</section>

<section title="Since draft-ietf-httpbis-cache-00" anchor="changes.since.00">
<t>
  The changes are purely editorial:
</t>
<ul>
  <li>Moved all extensibility tips, registration procedures, and registry
      tables from the IANA considerations to normative sections, reducing the
      IANA considerations to just instructions that will be removed prior to
      publication as an RFC.</li>
</ul>
</section>

<section title="Since draft-ietf-httpbis-cache-01" anchor="changes.since.01">
<ul>
  <li>Cite RFC 8126 instead of RFC 5226 (<eref target="https://github.com/httpwg/http-core/issues/75"/>)</li>
  <li>In <xref target="header.pragma"/>, misleading statement about the relation between Pragma and Cache-Control (<eref target="https://github.com/httpwg/http-core/issues/92"/>, <eref target="https://www.rfc-editor.org/errata/eid4674"/>)</li>
</ul>
</section>

<section title="Since draft-ietf-httpbis-cache-02" anchor="changes.since.02">
<ul>
   <li>In <xref target="response.cacheability"/>, explain that only final responses are cacheable (<eref target="https://github.com/httpwg/http-core/issues/29"/>)</li>
   <li>In <xref target="cache-response-directive"/>, clarify what responses various directives apply to (<eref target="https://github.com/httpwg/http-core/issues/52"/>)</li>
   <li>In <xref target="validation.sent"/>, clarify the source of validators in conditional requests (<eref target="https://github.com/httpwg/http-core/issues/110"/>)</li>
   <li>Revise <xref target="history.lists"/> to apply to more than just History Lists (<eref target="https://github.com/httpwg/http-core/issues/126"/>)</li>
   <li>In <xref target="header.warning"/>, deprecated "Warning" header field (<eref target="https://github.com/httpwg/http-core/issues/139"/>)</li>
   <li>In <xref target="caching.authenticated.responses"/>, remove a spurious note (<eref target="https://github.com/httpwg/http-core/issues/141"/>)</li>
</ul>
</section>

<section title="Since draft-ietf-httpbis-cache-03" anchor="changes.since.03">
<ul>
   <li>In <xref target="caching.overview"/>, define what a disconnected cache is (<eref target="https://github.com/httpwg/http-core/issues/5"/>)</li>
   <li>In <xref target="constructing.responses.from.caches"/>, clarify language around how to select a response when more than one matches (<eref target="https://github.com/httpwg/http-core/issues/23"/>)</li>
   <li>in <xref target="serving.stale.responses"/>, mention stale-while-revalidate and stale-if-error (<eref target="https://github.com/httpwg/http-core/issues/122"/>)</li>
   <li>Remove requirements around cache request directives (<eref target="https://github.com/httpwg/http-core/issues/129"/>)</li>
   <li>Deprecate Pragma (<eref target="https://github.com/httpwg/http-core/issues/140"/>)</li>
   <li>In <xref target="caching.authenticated.responses"/> and <xref target="cache-response-directive"/>, note effect of some directives on authenticated requests (<eref target="https://github.com/httpwg/http-core/issues/161"/>)</li>
</ul>
</section>

<section title="Since draft-ietf-httpbis-cache-04" anchor="changes.since.04">
<ul>
   <li>In <xref target="header.cache-control"/>, remove the registrations for stale-if-error and stale-while-revalidate which happened in RFC 7234 (<eref target="https://github.com/httpwg/http-core/issues/207"/>)</li>
</ul>
</section>

<section title="Since draft-ietf-httpbis-cache-05" anchor="changes.since.05">
<ul x:when-empty="None yet.">
   <li>In <xref target="incomplete.responses"/>, clarify how weakly framed content is considered for purposes of completeness (<eref target="https://github.com/httpwg/http-core/issues/25"/>)</li>
   <li>Throughout, describe Vary and cache key operations more clearly (<eref target="https://github.com/httpwg/http-core/issues/28"/>)</li>
   <li>In <xref target="response.cacheability"/>, remove concept of "cacheable methods" in favor of prose (<eref target="https://github.com/httpwg/http-core/issues/54"/>, <eref target="https://www.rfc-editor.org/errata/eid5300"/>)</li>
   <li>Refactored <xref target="security.considerations"/>, and added a section on timing attacks (<eref target="https://github.com/httpwg/http-core/issues/233"/>)</li>
   <li>Changed "cacheable by default" to "heuristically cacheable" throughout (<eref target="https://github.com/httpwg/http-core/issues/242"/>)</li>
</ul>
</section>

<section title="Since draft-ietf-httpbis-cache-06" anchor="changes.since.06">
<ul x:when-empty="None yet.">
   <li>Fix typo in <xref target="cache-response-directive.s-maxage"/> (<eref target="https://github.com/httpwg/http-core/issues/264"/>)</li>
</ul>
</section>
</section>

<section title="Acknowledgments" anchor="acks" numbered="false">
<t>
  See <xref target="Semantics-acks"/>.
</t>
</section>
</back>
</rfc>