<?xml version="1.0" encoding="UTF-8"?>
  <?xml-stylesheet type="text/xsl" href="rfc2629.xslt" ?>
  <!-- generated by https://github.com/cabo/kramdown-rfc version 1.7.32 (Ruby 3.3.0) -->


<!DOCTYPE rfc  [
  <!ENTITY nbsp    "&#160;">
  <!ENTITY zwsp   "&#8203;">
  <!ENTITY nbhy   "&#8209;">
  <!ENTITY wj     "&#8288;">

]>


<rfc ipr="trust200902" docName="draft-ietf-jmap-object-history-04" category="std" consensus="true" submissionType="IETF" tocInclude="true" sortRefs="true" symRefs="true">
  <front>
    <title abbrev="JMAP Object History">JMAP Object History</title>

    <author initials="B." surname="Gondwana" fullname="Bron Gondwana">
      <organization>Fastmail</organization>
      <address>
        <email>brong@fastmailteam.com</email>
      </address>
    </author>
    <author initials="J." surname="Coffey" fullname="John Coffey">
      <organization>Coffey Labs</organization>
      <address>
        <email>john@coffeylabs.org</email>
      </address>
    </author>

    <date year="2026" month="November" day="15"/>

    
    
    <keyword>Internet-Draft</keyword>

    <abstract>


<?line 47?>

<t>The JMAP base protocol (RFC8620) provides methods for synchronizing
the current state of data objects between client and server.  This
extension adds the ability to retrieve historical versions of objects,
including objects that have been destroyed, by extending the standard
Foo/get method, to find destroyed objects with Foo/query, and to
restore them with Foo/set.</t>



    </abstract>



  </front>

  <middle>


<?line 56?>

<section anchor="introduction"><name>Introduction</name>

<t>JMAP (<xref target="JMAP-CORE"/> — JSON Meta Application Protocol) provides a
standard set of methods for each data type: Foo/get, Foo/changes,
Foo/set, Foo/query, and Foo/queryChanges.  These methods operate
on the current state of objects.  The Foo/changes method tells a
client which objects have changed, but does not provide the
previous values.</t>

<t>In many applications it is useful to access historical versions
of objects.  For example, a user may want to see previous versions
of a contact record (<xref target="JMAP-CONTACTS"/>) or recover a deleted
email (<xref target="JMAP-MAIL"/>).</t>

<t>This extension adds history support to the standard Foo/get
method.  When the <spanx style="verb">urn:ietf:params:jmap:object-history</spanx> capability
is included in the request's <spanx style="verb">using</spanx> array, Foo/get gains
additional parameters that allow the client to request previous
versions of objects, including objects that have been destroyed,
Foo/query can search the destroyed objects, and Foo/set can
restore them.</t>

<section anchor="conventions-used-in-this-document"><name>Conventions Used in This Document</name>

<t>The key words "<bcp14>MUST</bcp14>", "<bcp14>MUST NOT</bcp14>", "<bcp14>REQUIRED</bcp14>", "<bcp14>SHALL</bcp14>", "<bcp14>SHALL
NOT</bcp14>", "<bcp14>SHOULD</bcp14>", "<bcp14>SHOULD NOT</bcp14>", "<bcp14>RECOMMENDED</bcp14>", "<bcp14>NOT RECOMMENDED</bcp14>",
"<bcp14>MAY</bcp14>", and "<bcp14>OPTIONAL</bcp14>" in this document are to be interpreted as
described in BCP 14 <xref target="RFC2119"/> <xref target="RFC8174"/> when, and only when, they
appear in all capitals, as shown here.</t>

<?line -18?>

</section>
</section>
<section anchor="addition-to-the-capabilities-object"><name>Addition to the Capabilities Object</name>

<t>The capabilities object is returned as part of the JMAP Session
object; see <xref target="JMAP-CORE"/>, Section 2.</t>

<t>This document defines an additional capability URI.</t>

<section anchor="urnietfparamsjmapobject-history"><name>urn:ietf:params:jmap:object-history</name>

<t>The capability <spanx style="verb">urn:ietf:params:jmap:object-history</spanx> being present
in the "accountCapabilities" property of an account represents
support for retrieving historical object versions in that account.
Servers that include the capability in one or more
"accountCapabilities" properties <bcp14>MUST</bcp14> also include the property in
the "capabilities" property.</t>

<t>The value of this property in the JMAP session "capabilities"
property <bcp14>MUST</bcp14> be an empty object.</t>

<t>The value of this property in an account's "accountCapabilities"
property is an object that <bcp14>MUST</bcp14> contain the following information
on server capabilities for that account:</t>

<t><list style="symbols">
  <t>maxHistoryDuration: "UnsignedInt|null"  <vspace blankLines='1'/>
The number of seconds after replacement for which the server
  intends to retain old versions, or null if the server does
  not have a time-based retention policy.  This is a statement
  of server policy, not a guarantee: retention is best effort,
  and the server <bcp14>MAY</bcp14> discard versions sooner (for example, due
  to storage limits).  A version whose <spanx style="verb">replaced</spanx> timestamp is
  more than this many seconds ago has probably been discarded.
  Clients that need to be certain an object remains available
  should not destroy it.</t>
  <t>supportedTypes: "String[]"  <vspace blankLines='1'/>
The names of the data types, from the "JMAP Data Types"
  registry <xref target="IANA-JMAP"/> or private types, for which the server
  supports the history extensions in this account.  A server
  <bcp14>MAY</bcp14> support history for only a subset of the types it
  supports, and <bcp14>MAY</bcp14> retain history for only some objects of
  a supported type (for example, only for some folders).
  Clients <bcp14>MUST</bcp14> ignore type names they do not recognise.</t>
</list></t>

</section>
</section>
<section anchor="extensions-to-fooget"><name>Extensions to Foo/get</name>

<t>When this capability is included in the request's <spanx style="verb">using</spanx> array,
the standard Foo/get method (<xref target="JMAP-CORE"/>, Section 5.1) accepts
the following additional request arguments:</t>

<t><list style="symbols">
  <t>includeReplaced: "Boolean" (default: false)  <vspace blankLines='1'/>
If true, the server returns previous versions of the
  requested objects in addition to the current version.
  Each previous version appears as a separate entry in the
  <spanx style="verb">list</spanx> array, with the same <spanx style="verb">id</spanx> as the current version.
  Every entry includes an <spanx style="verb">objectHistory</spanx> property (see
  below).  Entries with the same <spanx style="verb">id</spanx> <bcp14>MUST</bcp14> be ordered by
  <spanx style="verb">objectHistory.version</spanx> (ascending).  Entries for
  different objects <bcp14>MAY</bcp14> be interleaved.</t>
  <t>includeDestroyed: "Boolean" (default: false)  <vspace blankLines='1'/>
If true, the server also returns objects that have been
  destroyed.  If <spanx style="verb">includeReplaced</spanx> is false, only the final
  version before destruction is returned.  If <spanx style="verb">includeReplaced</spanx>
  is true, all retained historical versions of destroyed
  objects are returned.  For destroyed objects, every entry
  has a non-null <spanx style="verb">objectHistory.replaced</spanx> value, since every
  version has been superseded (the final version was replaced
  by destruction).  <vspace blankLines='1'/>
An object counts as destroyed however it was removed: by
  Foo/set, through another protocol (for example an IMAP EXPUNGE
  <xref target="IMAP"/> or a POP3 DELE <xref target="POP3"/>), or by the server itself, for
  example when it empties a Trash mailbox automatically or destroys
  a Mailbox with <spanx style="verb">onDestroyRemoveEmails</spanx> (<xref target="JMAP-MAIL"/>,
  Section 2.5).  Likewise, changes made through other protocols can
  create versions just as changes made through JMAP do.</t>
  <t>historyAfter: "UTCDate|null" (default: null)  <vspace blankLines='1'/>
If supplied, only versions with an <spanx style="verb">objectHistory.replaced</spanx>
  value strictly after this date are returned, plus the
  current (live) version which has a null <spanx style="verb">replaced</spanx> value.
  If null, no time-based filter is applied.</t>
  <t>historyBefore: "UTCDate|null" (default: null)  <vspace blankLines='1'/>
If supplied, only versions with an <spanx style="verb">objectHistory.replaced</spanx>
  value strictly before this date are returned.  The current
  (live) version, which has a null <spanx style="verb">replaced</spanx> value, is not
  returned.  If null, no time-based filter is applied.  <vspace blankLines='1'/>
A client can use this to page backwards through history: it
  passes the oldest <spanx style="verb">replaced</spanx> value from the previous response
  as <spanx style="verb">historyBefore</spanx> in the next request.  Since <spanx style="verb">historyBefore</spanx>
  applies to every id in the call, a client paging through the
  history of several ids makes a separate Foo/get call for each
  id.  Clients <bcp14>SHOULD</bcp14> pass the <spanx style="verb">replaced</spanx>
  value exactly as received, rather than parsing and
  re-serialising it, since re-serialising at a lower precision
  could cause versions to be skipped.</t>
  <t>historySkip: "UnsignedInt" (default: 0)  <vspace blankLines='1'/>
The number of entries to skip for each id, counting from the
  most recent (the highest version number), before any entries
  are returned.  The current version, if it would otherwise be
  returned, counts as one entry.  For example, a client that
  wants only versions older than the 20 most recent passes
  <spanx style="verb">historySkip: 20</spanx>; any id with 20 or fewer matching entries
  is returned in <spanx style="verb">notFound</spanx>.  <vspace blankLines='1'/>
A client can also use this to page backwards through history,
  by adding the number of entries returned for an id to
  <spanx style="verb">historySkip</spanx> in the next request.  This is less robust than
  <spanx style="verb">historyBefore</spanx>: if a new version is recorded between
  requests, the window shifts and one entry is returned again.</t>
  <t>historyLimit: "UnsignedInt|null" (default: null)  <vspace blankLines='1'/>
If supplied, the maximum number of entries to return in the
  <spanx style="verb">list</spanx> array for each id, including the current version if
  any.  For each id, the server <bcp14>MUST</bcp14> return the most recent
  entries (those with the highest version numbers) remaining
  after <spanx style="verb">historySkip</spanx> is applied, not the oldest.  If null, the
  server returns all remaining retained versions, subject to
  its own limits.</t>
</list></t>

<t>The server applies <spanx style="verb">includeReplaced</spanx>, <spanx style="verb">includeDestroyed</spanx>,
<spanx style="verb">historyAfter</spanx>, and <spanx style="verb">historyBefore</spanx> to select the matching
versions of each id, then skips <spanx style="verb">historySkip</spanx> entries, then
returns at most <spanx style="verb">historyLimit</spanx> of the remaining entries.</t>

<t>If <spanx style="verb">includeReplaced</spanx> and <spanx style="verb">includeDestroyed</spanx> are both false, the
server ignores <spanx style="verb">historyAfter</spanx>, <spanx style="verb">historyBefore</spanx>, <spanx style="verb">historySkip</spanx>, and
<spanx style="verb">historyLimit</spanx>.</t>

<t>If the <spanx style="verb">ids</spanx> argument is null, the server returns all objects of
the type as per <xref target="JMAP-CORE"/>, Section 5.1.  If <spanx style="verb">includeDestroyed</spanx>
is true, this includes every destroyed object for which the server
still retains any version.</t>

<t>The "maxObjectsInGet" limit and the "requestTooLarge" error
(<xref target="JMAP-CORE"/>, Section 5.1) apply to the number of distinct ids,
not to the number of entries in the <spanx style="verb">list</spanx> array.  Each id can
have up to <spanx style="verb">historyLimit</spanx> entries.</t>

<t>If, after applying <spanx style="verb">includeReplaced</spanx>, <spanx style="verb">includeDestroyed</spanx>,
<spanx style="verb">historyAfter</spanx>, <spanx style="verb">historyBefore</spanx>, and <spanx style="verb">historySkip</spanx>, no versions of a
requested id remain to be returned, that id is included in the
<spanx style="verb">notFound</spanx> array.</t>

<section anchor="the-objecthistory-property"><name>The objectHistory property</name>

<t>When <spanx style="verb">includeReplaced</spanx> or <spanx style="verb">includeDestroyed</spanx> is true, each object
in the <spanx style="verb">list</spanx> array includes an additional property:</t>

<t><list style="symbols">
  <t>objectHistory: "ObjectHistoryInfo"  <vspace blankLines='1'/>
An ObjectHistoryInfo object with the following properties:  <list style="symbols">
      <t>version: "UnsignedInt"
A per-object monotonically increasing number that
provides ordering among the versions of a single object
within a response.  Version numbers <bcp14>MAY</bcp14> have gaps (e.g.,
a server may use internal sequence numbers such as modseq
values).  The current (live) version of an object always
has the highest version number.  Servers <bcp14>SHOULD</bcp14> return
consistent version numbers across requests, but clients
<bcp14>MUST NOT</bcp14> rely on this — version numbers exist primarily
to provide ordering within a single response.</t>
      <t>replaced: "UTCDate|null"
The date and time at which this version was superseded,
either by a subsequent update or by destruction.  If null,
this is the current live version of the object.  For
destroyed objects, every version has a non-null <spanx style="verb">replaced</spanx>
value — the <spanx style="verb">replaced</spanx> of the version that was current when
the object was destroyed is the time of destruction.  The <spanx style="verb">replaced</spanx> values of different versions
of the same object <bcp14>MUST</bcp14> be distinct, so that <spanx style="verb">historyBefore</spanx>
and <spanx style="verb">historyAfter</spanx> partition the history without ambiguity;
a server that records two versions with the same timestamp
can collapse them into a single version.  The value <bcp14>MAY</bcp14>
include fractional seconds (<xref target="JMAP-CORE"/>, Section 1.4),
but servers <bcp14>SHOULD NOT</bcp14> use more than millisecond precision,
as many client date libraries cannot represent more.</t>
      <t>hasOlder: "Boolean"
If true, the server retains at least one older version of
this object than this one, whether or not that version was
returned in this response.  A client can check <spanx style="verb">hasOlder</spanx>
on the oldest entry returned for an id to learn whether
more history is available, and retrieve it using
<spanx style="verb">historyBefore</spanx> or <spanx style="verb">historySkip</spanx>.  The server <bcp14>MAY</bcp14> return
false even if older versions exist (e.g., if it cannot
efficiently determine whether older versions remain).  If
the server returns true, the client <bcp14>MUST NOT</bcp14> assume the
older versions will still be available in a subsequent
request, as the server may purge history at any time.</t>
      <t>expires: "UTCDate|null"
The time after which the server expects to discard this
version, or null if it has no planned time.  The current
(live) version of an object always has a null <spanx style="verb">expires</spanx>.
Like <spanx style="verb">maxHistoryDuration</spanx>, this is a best-effort statement
and not a guarantee: the server <bcp14>MAY</bcp14> discard the version
sooner, for example when storage runs short, and <bcp14>MAY</bcp14> return a
different value in a later response.  A client can use this
to tell the user how long a destroyed object can still be
recovered, but <bcp14>MUST NOT</bcp14> rely on the version remaining
available until then.</t>
    </list></t>
</list></t>

<t>When <spanx style="verb">includeReplaced</spanx> and <spanx style="verb">includeDestroyed</spanx> are both false
(the default), the <spanx style="verb">objectHistory</spanx> property is not included in
the response, and the behaviour is identical to standard Foo/get.</t>

<t>The client <bcp14>MAY</bcp14> include "objectHistory" in the <spanx style="verb">properties</spanx> argument.
It is returned if and only if <spanx style="verb">includeReplaced</spanx> or <spanx style="verb">includeDestroyed</spanx>
is true, regardless of whether it is listed in <spanx style="verb">properties</spanx>.</t>

<t>Which changes produce a version is server dependent: a version is
a snapshot of the object that the server recorded at some point.
The server is not required to create a new version for every
individual change to an object.  For example, rapid successive
updates may be collapsed into a single version, and a server may
choose not to record a version for changes that only affect
server-computed properties.  The server <bcp14>MAY</bcp14> also prune old versions
in any order, so there can be gaps in the version history available
to the client.  Clients <bcp14>MUST NOT</bcp14> assume that the version history is
complete or contiguous.</t>

<t>The final version of a destroyed object is the most recent version
of it that the server retains.  This is normally the version that
was current when the object was destroyed.  If the server has
discarded that version and retains an older one, the older one is
the final version wherever this document refers to it, and its
<spanx style="verb">replaced</spanx> value, which is then earlier than the time of
destruction, is used as the time the object was destroyed.</t>

<t>The values of server-computed properties (for example, the counts
on a Mailbox) in a historical version are also server dependent.
The server <bcp14>SHOULD</bcp14> return the value the property had at the time of
that version if it has that information.</t>

<t>If a historical version contains a reference to a blob (for
example, the <spanx style="verb">blobId</spanx> of an Email), that blob <bcp14>MUST</bcp14> be downloadable
for as long as the server returns that version.  A version that is
not current does not count as a reference to the blob, for garbage
collection (<xref target="JMAP-CORE"/>, Section 6) or for any other purpose, such
as the "blobHasReference" error and Blob/lookup method of
<xref target="JMAP-BLOBEXT"/>.  A server that discards a blob referenced by a
version <bcp14>MUST NOT</bcp14> return that version.</t>

<t>If <spanx style="verb">includeReplaced</spanx> or <spanx style="verb">includeDestroyed</spanx> is true and the data type
is not listed in the <spanx style="verb">supportedTypes</spanx> capability property, the
server <bcp14>MUST</bcp14> reject the call with an "invalidArguments" error.</t>

<t>If the data type is supported but the server does not retain
history for a particular object, it <bcp14>MUST</bcp14> still return the current
version with an <spanx style="verb">objectHistory</spanx> property when requested.  In this
case, the server returns only the current version, with a <spanx style="verb">replaced</spanx>
of null.</t>

</section>
<section anchor="examples"><name>Examples</name>

<section anchor="contact-history"><name>Contact history</name>

<t>Requesting the history of a contact, including all versions:</t>

<figure><sourcecode type="json"><![CDATA[
[["ContactCard/get", {
  "accountId": "abc",
  "ids": ["contact1"],
  "properties": ["name", "emails"],
  "includeReplaced": true,
  "historyLimit": 10
}, "0"]]
]]></sourcecode></figure>

<t>The response includes three versions: the original creation,
a name update, and the current version with an additional email
address:</t>

<figure><sourcecode type="json"><![CDATA[
[["ContactCard/get", {
  "accountId": "abc",
  "state": "abc123",
  "list": [
    {
      "id": "contact1",
      "name": {"full": "Robert Smith"},
      "emails": {
        "e1": {"address": "bob@example.com"}
      },
      "objectHistory": {
        "version": 1,
        "replaced": "2026-02-20T14:15:00Z",
        "hasOlder": false,
        "expires": "2026-03-22T14:15:00Z"
      }
    },
    {
      "id": "contact1",
      "name": {"full": "Bob Smith"},
      "emails": {
        "e1": {"address": "bob@example.com"}
      },
      "objectHistory": {
        "version": 2,
        "replaced": "2026-03-01T09:00:00Z",
        "hasOlder": true,
        "expires": "2026-03-31T09:00:00Z"
      }
    },
    {
      "id": "contact1",
      "name": {"full": "Bob Smith"},
      "emails": {
        "e1": {"address": "bob@example.com"},
        "e2": {"address": "bob@personal.example"}
      },
      "objectHistory": {
        "version": 3,
        "replaced": null,
        "hasOlder": true,
        "expires": null
      }
    }
  ],
  "notFound": []
}, "0"]]
]]></sourcecode></figure>

</section>
<section anchor="recovering-destroyed-emails"><name>Recovering destroyed emails</name>

<t>A client can use Email/changes to discover which emails have been
destroyed, then use a backreference to fetch their final state.
This is useful for displaying a "recently deleted" view:</t>

<figure><sourcecode type="json"><![CDATA[
[
  ["Email/changes", {
    "accountId": "abc",
    "sinceState": "state42"
  }, "0"],
  ["Email/get", {
    "accountId": "abc",
    "#ids": {
      "resultOf": "0",
      "name": "Email/changes",
      "path": "/destroyed"
    },
    "properties": ["subject", "from", "receivedAt"],
    "includeDestroyed": true
  }, "1"]
]
]]></sourcecode></figure>

<t>The Email/get response returns the final state of each
destroyed email.  The <spanx style="verb">objectHistory.replaced</spanx> value is the
time of destruction:</t>

<figure><sourcecode type="json"><![CDATA[
[["Email/get", {
  "accountId": "abc",
  "state": "state43",
  "list": [
    {
      "id": "email42",
      "subject": "Meeting notes",
      "from": [{"name": "Alice",
        "email": "alice@example.com"}],
      "receivedAt": "2026-03-10T09:00:00Z",
      "objectHistory": {
        "version": 1,
        "replaced": "2026-03-15T16:30:00Z",
        "hasOlder": false,
        "expires": "2026-04-14T16:30:00Z"
      }
    },
    {
      "id": "email57",
      "subject": "Project update",
      "from": [{"name": "Bob",
        "email": "bob@example.com"}],
      "receivedAt": "2026-03-12T14:00:00Z",
      "objectHistory": {
        "version": 1,
        "replaced": "2026-03-15T16:31:00Z",
        "hasOlder": false,
        "expires": "2026-04-14T16:31:00Z"
      }
    }
  ],
  "notFound": []
}, "1"]]
]]></sourcecode></figure>

<t>Since <spanx style="verb">includeReplaced</spanx> was not set, only the final version of
each destroyed email is returned.  The non-null <spanx style="verb">replaced</spanx>
value indicates when each email was destroyed.  The server in
this example keeps history for 30 days.</t>

</section>
</section>
</section>
<section anchor="query"><name>Extensions to Foo/query</name>

<t>The Foo/changes approach in <xref target="recovering-destroyed-emails"/> only
works for a client that holds a state string from before the
objects were destroyed, and only while the server can still
calculate changes from that state.  Foo/get with a null <spanx style="verb">ids</spanx>
argument and <spanx style="verb">includeDestroyed</spanx> returns every retained object of
the type at once, with no filtering, sorting or paging.</t>

<t>When this capability is included in the request's <spanx style="verb">using</spanx> array,
the standard Foo/query method (<xref target="JMAP-CORE"/>, Section 5.5) accepts
the following additional request arguments:</t>

<t><list style="symbols">
  <t>onlyDestroyed: "Boolean" (default: false)  <vspace blankLines='1'/>
If true, the query runs over the destroyed objects of this
  type for which the server still retains at least one version,
  instead of over current objects.  An object that has been
  restored (<xref target="restoring"/>) is a current object again, and is not
  included.</t>
  <t>destroyedAfter: "UTCDate|null" (default: null)  <vspace blankLines='1'/>
If supplied, only objects destroyed strictly after this date
  are included.</t>
  <t>destroyedBefore: "UTCDate|null" (default: null)  <vspace blankLines='1'/>
If supplied, only objects destroyed strictly before this date
  are included.</t>
</list></t>

<t>The time an object was destroyed is the <spanx style="verb">replaced</spanx> value of its
final version.</t>

<t>The server <bcp14>MUST</bcp14> reject the call with an "invalidArguments" error if
<spanx style="verb">onlyDestroyed</spanx> is true and the type is not listed in
<spanx style="verb">supportedTypes</spanx>, or if <spanx style="verb">destroyedAfter</spanx> or <spanx style="verb">destroyedBefore</spanx> is
supplied and <spanx style="verb">onlyDestroyed</spanx> is false.</t>

<t>When <spanx style="verb">onlyDestroyed</spanx> is true:</t>

<t><list style="symbols">
  <t>The <spanx style="verb">filter</spanx> and <spanx style="verb">sort</spanx> arguments apply to the final version of
each destroyed object.  The server need not support every filter
condition or comparator for destroyed objects that it supports
for current objects.  It rejects one it does not support with
"unsupportedFilter" or "unsupportedSort", as usual.</t>
  <t>If <spanx style="verb">sort</spanx> is null or empty, the results are ordered by time of
destruction, most recent first.</t>
  <t>Arguments that apply only to current objects, such as
<spanx style="verb">collapseThreads</spanx> for Email, are ignored.</t>
  <t>Foo/queryChanges (<xref target="JMAP-CORE"/>, Section 5.6) also accepts the
<spanx style="verb">onlyDestroyed</spanx>, <spanx style="verb">destroyedAfter</spanx>, and <spanx style="verb">destroyedBefore</spanx>
arguments when this capability is included in the request's
<spanx style="verb">using</spanx> array.  The client <bcp14>MUST</bcp14> pass the same values that it
passed to Foo/query, as it does for <spanx style="verb">filter</spanx> and <spanx style="verb">sort</spanx>, and the
server rejects them with "invalidArguments" in the same cases as
Foo/query does.  An object leaves the results when it is restored
or discarded, and enters them when it is destroyed.  A server
that cannot track these changes <bcp14>MAY</bcp14> return a
"cannotCalculateChanges" error.</t>
</list></t>

<t>The ids returned can then be passed to Foo/get with
<spanx style="verb">includeDestroyed</spanx> set, usually with a backreference.</t>

<section anchor="example"><name>Example</name>

<t>A client builds a "recently deleted" view of the 50 emails
destroyed most recently since 1 March:</t>

<figure><sourcecode type="json"><![CDATA[
[
  ["Email/query", {
    "accountId": "abc",
    "onlyDestroyed": true,
    "destroyedAfter": "2026-03-01T00:00:00Z",
    "limit": 50
  }, "0"],
  ["Email/get", {
    "accountId": "abc",
    "#ids": {
      "resultOf": "0",
      "name": "Email/query",
      "path": "/ids"
    },
    "properties": ["subject", "from", "receivedAt"],
    "includeDestroyed": true
  }, "1"]
]
]]></sourcecode></figure>

<t>The Email/query response lists the destroyed emails, most recently
destroyed first:</t>

<figure><sourcecode type="json"><![CDATA[
[["Email/query", {
  "accountId": "abc",
  "queryState": "dq7",
  "canCalculateChanges": true,
  "position": 0,
  "ids": ["email57", "email42"]
}, "0"]]
]]></sourcecode></figure>

<t>The Email/get response is the same as in
<xref target="recovering-destroyed-emails"/>.</t>

</section>
</section>
<section anchor="restoring"><name>Restoring Objects</name>

<t>To restore an earlier version of an existing object, the client
uses a normal Foo/set update with the values from that version.</t>

<t>To restore a destroyed object, the client uses the <spanx style="verb">restore</spanx>
argument of Foo/set.</t>

<section anchor="extensions-to-fooset"><name>Extensions to Foo/set</name>

<t>When this capability is included in the request's <spanx style="verb">using</spanx> array,
the standard Foo/set method (<xref target="JMAP-CORE"/>, Section 5.3) accepts
the following additional request argument:</t>

<t><list style="symbols">
  <t>restore: "Id[PatchObject|null]|null" (default: null)  <vspace blankLines='1'/>
A map of the ids of destroyed objects to restore.  Each value
  is null, or a PatchObject to apply to the restored object.  An
  empty PatchObject means the same as null.</t>
</list></t>

<t>The response has the following additional properties:</t>

<t><list style="symbols">
  <t>restored: "Id[Foo|null]|null"  <vspace blankLines='1'/>
A map of the ids of objects that were successfully restored.
  The value for each is an object containing any properties of the
  restored object whose values differ from those in the final
  version with the patch applied, or null if there are none.  This
  is null if no objects were restored.</t>
  <t>notRestored: "Id[SetError]|null"  <vspace blankLines='1'/>
A map of the ids of objects that could not be restored to a
  SetError object explaining why.  This is null if every restore
  succeeded or none was requested.</t>
</list></t>

<t>The server <bcp14>MUST</bcp14> reject the call with an "invalidArguments" error if
<spanx style="verb">restore</spanx> is non-null and the type is not listed in
<spanx style="verb">supportedTypes</spanx>.</t>

<t>The server processes each restore before any create, update, or
destroy in the same call, so these can refer to a restored object.
Where restored objects refer to other objects of the same type, the
server <bcp14>MUST</bcp14> order the restores within a single call so that an
object is restored before any restore that refers to it, as for
creates in <xref target="JMAP-CORE"/>, Section 5.3; for example, a Mailbox is
restored before a child Mailbox whose <spanx style="verb">parentId</spanx> refers to it.</t>

<t>The restored object takes the id of the destroyed object, and the
values of the final version of that object
(<xref target="the-objecthistory-property"/>), including server-set properties
such as the <spanx style="verb">threadId</spanx> and <spanx style="verb">size</spanx> of an Email, with the PatchObject
applied.  The patch follows the same rules as a PatchObject in an
update in Foo/set: it <bcp14>MAY</bcp14> include a server-set property only if the
value is the same as in the final version, and the server otherwise
rejects the restore with "invalidProperties".  So a client can send
a whole version of the object as the patch.</t>

<t>Server-set properties keep their values from the final version.  The
server <bcp14>MUST NOT</bcp14> change a property of the object through a destroy and
restore in a way that it could not when updating the object if it had
never been destroyed.  In particular, an immutable property, such as
the <spanx style="verb">threadId</spanx> of an Email, keeps its value; if no other Email is in
that Thread, the Thread exists again with the same id.  Where the
server would also change a server-set property on an update, such as
a modification time, it can change it on a restore, and returns the
new values in <spanx style="verb">restored</spanx>.  If the server cannot restore the object
without a change that is not allowed, it <bcp14>MUST</bcp14> reject the restore.</t>

<t>If the server has discarded the version that was current when the
object was destroyed, the final version is an older one
(<xref target="the-objecthistory-property"/>), and the restore takes its values.
If these no longer make a valid object, the server rejects the
restore with "invalidProperties", as for any other restore.</t>

<t>The history of the restored object continues from that of the
destroyed object, so the destroyed versions become previous versions
of the restored object.  To restore a version other than the final
one, the client includes the values of that version in the
PatchObject.</t>

<t>The server rejects a restore with a SetError as follows:</t>

<t><list style="symbols">
  <t>notFound: The id is not the id of a destroyed object of this type
for which the server retains any history.  This includes the id of
a current object.</t>
  <t>invalidPatch: The PatchObject is not valid.</t>
  <t>invalidProperties: The restored object would not be valid, as for
an update in Foo/set.</t>
  <t>forbidden: The server does not allow this object to be restored,
for example because of its content (<xref target="harmful-content"/>), or
because it cannot restore it without changing an immutable
property.</t>
  <t>overQuota: The restored object would exceed a quota.</t>
</list></t>

<t>The server <bcp14>MAY</bcp14> also return any other SetError that Foo/set returns
for an update or create of this type.</t>

<t>The server <bcp14>MUST NOT</bcp14> reuse the id of a destroyed object for which it
retains any history, other than by restoring that object.</t>

<t>A restore is a change that the Foo/set requests, as a create, update,
or destroy is.  Each entry in <spanx style="verb">restore</spanx> counts as one object towards
the "maxObjectsInSet" limit (<xref target="JMAP-CORE"/>, Section 2): the limit, and
the "requestTooLarge" error (<xref target="JMAP-CORE"/>, Section 5.3), apply to the
total number of objects that the call creates, updates, destroys, and
restores.</t>

<t>When the "atomic" argument of <xref target="JMAP-CONDITIONAL"/> is
true, the restores are part of the single unit: no restore takes
effect unless every change in the method does, and a restore that
fails causes an "atomicFailure" error, which <bcp14>MAY</bcp14> carry a
<spanx style="verb">notRestored</spanx> map.  The order of processing given above then
determines only how references to a restored object are resolved.
The server checks a constraint, such as the uniqueness of a name,
against the state that results from all of the changes, so a restore
can depend on an update or destroy in the same call, for example one
that frees the name that the restored object takes.</t>

<t>The server evaluates an "ifUnchangedBy" precondition
(<xref target="JMAP-CONDITIONAL"/>) against the state at the start of the method,
before any restore.  A destroyed object that the method restores is
therefore an object that does not exist.</t>

<t>Foo/changes (<xref target="JMAP-CORE"/>, Section 5.2) reports a restored object
by comparing <spanx style="verb">sinceState</spanx> with the new state, exactly as if the object
had never been destroyed between them:</t>

<t><list style="symbols">
  <t>If the object exists in both states, the id is reported in
<spanx style="verb">updated</spanx>, even if it was destroyed and restored in between.</t>
  <t>If it exists only in the new state, the id is reported in
<spanx style="verb">created</spanx>.</t>
  <t>If it exists only at <spanx style="verb">sinceState</spanx>, for example because it was
restored before <spanx style="verb">sinceState</spanx> and destroyed again after it, the id
is reported in <spanx style="verb">destroyed</spanx>.</t>
  <t>If it exists in neither, because it was destroyed at <spanx style="verb">sinceState</spanx>
and was restored and destroyed again after it, the server <bcp14>SHOULD NOT</bcp14>
report the id at all, but <bcp14>MAY</bcp14> report it as for an object created
and destroyed since <spanx style="verb">sinceState</spanx> in <xref target="JMAP-CORE"/>, Section 5.2.</t>
</list></t>

</section>
<section anchor="type-specific-notes"><name>Type-Specific Notes</name>

<t>A restored Email is the original message: it has the same <spanx style="verb">blobId</spanx>,
and so the same content, as the final version.  A client can
restore an Email with Email/set without fetching its content first.</t>

<t>A SieveScript <xref target="JMAP-SIEVE"/> is deactivated before it is destroyed
(<xref target="JMAP-SIEVE"/>, Section 2.4), so a restored SieveScript is not
active.  The client can activate it with <spanx style="verb">onSuccessActivateScript</spanx>,
using its id.</t>

<t>The final version may refer to an object that no longer exists,
such as a Mailbox in an Email's <spanx style="verb">mailboxIds</spanx>.  The server then
rejects the restore as it would an update that refers to a
nonexistent object.  The client can restore that object first, in
an earlier method call in the same request or in an earlier request,
since Mailbox, AddressBook, and Calendar can have history too.  Or
it can choose another one in the PatchObject, for example the Mailbox with the "inbox" role.  The
server can also reject a restore that conflicts with a current
object, using the error that Foo/set would normally return; for
example, a SieveScript whose name is now used by another script is
rejected with "alreadyExists" (<xref target="JMAP-SIEVE"/>, Section 2.4).</t>

</section>
<section anchor="example-1"><name>Example</name>

<t>Having found email42 with Email/query (see <xref target="query"/>), the client
restores it.  The Mailbox it was in has since been destroyed, so
the client puts it in the Inbox (mb1):</t>

<figure><sourcecode type="json"><![CDATA[
[["Email/set", {
  "accountId": "abc",
  "restore": {
    "email42": {
      "mailboxIds": {"mb1": true}
    }
  }
}, "0"]]
]]></sourcecode></figure>

<t>The server restores the Email, with the same <spanx style="verb">threadId</spanx> as before.
No server-set property has changed, so the value in <spanx style="verb">restored</spanx> is
null:</t>

<figure><sourcecode type="json"><![CDATA[
[["Email/set", {
  "accountId": "abc",
  "oldState": "s110",
  "newState": "s111",
  "created": null,
  "updated": null,
  "destroyed": null,
  "notCreated": null,
  "notUpdated": null,
  "notDestroyed": null,
  "restored": {"email42": null},
  "notRestored": null
}, "0"]]
]]></sourcecode></figure>

</section>
</section>
<section anchor="design-rationale"><name>Design Rationale</name>

<t>This section is not normative.</t>

<t>History is returned through the existing Foo/get method rather than
a new method.  The entries in the <spanx style="verb">list</spanx> array are ordinary objects
of the data type, so a client can feed them to the same parser it
uses for current objects, and the existing <spanx style="verb">properties</spanx> argument
works unchanged.  It also allows a client to fetch the final state
of many destroyed objects in a single call, using a backreference
to the <spanx style="verb">destroyed</spanx> list from Foo/changes, as shown in the examples
above.</t>

<t>The history metadata is embedded in each object as the
<spanx style="verb">objectHistory</spanx> property, rather than wrapping each object in a
separate container, for the same reason: the entries remain
ordinary objects.  This follows the existing pattern of JMAP-specific
properties such as <spanx style="verb">id</spanx> being added to objects whose format is
defined elsewhere; the <spanx style="verb">objectHistory</spanx> name is reserved in the
relevant property registries (see <xref target="iana-considerations"/>).</t>

<t>Each entry is a full copy of the object rather than a patch
against the previous version.  Patches would save space, but the
client would need every intermediate version to reconstruct any
one of them, which conflicts with the server's freedom to prune
versions, and the final version of a destroyed object would need
to be returned in full anyway.</t>

<t>Every superseded version carries a <spanx style="verb">replaced</spanx> timestamp.  The
server always knows, at least approximately, when it replaced a
snapshot it was keeping, and the timestamp is what a client needs
to present history to a user and to filter it with <spanx style="verb">historyAfter</spanx>.</t>

<t>Destroyed objects are restored with a new <spanx style="verb">restore</spanx> argument to
Foo/set rather than by creating an object with the old id.  A
create in Foo/set cannot set the id or server-set properties, so the
restored object would differ from the original; an Email created
with Email/set is a new message, not the original one.  Restoring
the final version, with a patch, brings back the object exactly,
and the client need not send the whole object back to the server.</t>

<t>Destroyed objects are found with Foo/query rather than a new
method, for the same reasons: the ids it returns go straight into
Foo/get with <spanx style="verb">includeDestroyed</spanx>.  The query runs over destroyed
objects only, rather than over current and destroyed objects
together, because "recently deleted" is normally a separate view.
It also lets a server keep destroyed objects in a separate index.</t>

</section>
<section anchor="iana-considerations"><name>IANA Considerations</name>

<section anchor="jmap-capability-registration-for-urnietfparamsjmapobject-history"><name>JMAP Capability Registration for urn:ietf:params:jmap:object-history</name>

<t>IANA is requested to register the "Object History" Capability
as follows:</t>

<t>Capability Name: urn:ietf:params:jmap:object-history</t>

<t>Intended use: common</t>

<t>Change Controller: IETF</t>

<t>Specification document: this document</t>

<t>Security and privacy considerations: this document,
Security Considerations</t>

</section>
<section anchor="update-to-the-jmap-data-types-registry"><name>Update to the JMAP Data Types Registry</name>

<t>This document adds a "Supports History" column to the "JMAP Data
Types" registry <xref target="IANA-JMAP"/>.  The value
is "Yes" or "No", indicating whether the data type supports the
history extensions defined in this document.</t>

<t>The initial values for existing registrations are:</t>

<texttable>
      <ttcol align='left'>Type Name</ttcol>
      <ttcol align='left'>Supports History</ttcol>
      <c>Core</c>
      <c>No</c>
      <c>PushSubscription</c>
      <c>No</c>
      <c>Mailbox</c>
      <c>Yes</c>
      <c>Thread</c>
      <c>No</c>
      <c>Email</c>
      <c>Yes</c>
      <c>EmailDelivery</c>
      <c>No</c>
      <c>SearchSnippet</c>
      <c>No</c>
      <c>Identity</c>
      <c>Yes</c>
      <c>EmailSubmission</c>
      <c>No</c>
      <c>VacationResponse</c>
      <c>Yes</c>
      <c>MDN</c>
      <c>No</c>
      <c>Quota</c>
      <c>No</c>
      <c>SieveScript</c>
      <c>Yes</c>
      <c>Principal</c>
      <c>No</c>
      <c>ShareNotification</c>
      <c>No</c>
      <c>AddressBook</c>
      <c>Yes</c>
      <c>ContactCard</c>
      <c>Yes</c>
      <c>Calendar</c>
      <c>Yes</c>
      <c>CalendarEvent</c>
      <c>Yes</c>
      <c>CalendarEventNotification</c>
      <c>No</c>
      <c>ParticipantIdentity</c>
      <c>Yes</c>
      <c>FileNode</c>
      <c>Yes</c>
</texttable>

<t>FileNode is registered by <xref target="JMAP-FILENODE"/>.</t>

<t>The default value for this column is "No".  New registrations
that do not explicitly include a value are assumed not to
support history.</t>

<t>This column records whether history is meaningful for a data type.
Whether a particular server supports history for a type is
advertised in the <spanx style="verb">supportedTypes</spanx> capability property.</t>

</section>
<section anchor="jscontact-property-registration"><name>JSContact Property Registration</name>

<t>IANA is requested to register the following property in the
"JSContact Properties" registry defined in <xref target="JSCONTACT"/>,
Section 3, so that the name cannot be used for a property of a
JSContact Card.</t>

<t>Property Name:  objectHistory</t>

<t>Property Type:  not applicable</t>

<t>Property Context:  Card</t>

<t>Intended Usage:  reserved</t>

<t>Since Version:  1.0</t>

<t>Change Controller:  IETF</t>

<t>Reference:  this document</t>

</section>
<section anchor="jscalendar-property-registration"><name>JSCalendar Property Registration</name>

<t>IANA is requested to register the following property in the
"JSCalendar Properties" registry defined in <xref target="JSCALENDAR"/>,
Section 8.2, so that the name cannot be used for a property of a
JSCalendar object.</t>

<t>Property Name:  objectHistory</t>

<t>Property Type:  not applicable</t>

<t>Property Context:  Event, Task, Group</t>

<t>Intended Usage:  reserved</t>

<t>Change Controller:  IETF</t>

<t>Reference:  this document</t>

<t>Specifications that define a JMAP data type whose object format is
also governed by a separate property registry <bcp14>SHOULD</bcp14> reserve the
<spanx style="verb">objectHistory</spanx> property name in that registry.</t>

</section>
</section>
<section anchor="security-considerations"><name>Security Considerations</name>

<t>All security considerations from <xref target="JMAP-CORE"/> apply to this
document.</t>

<section anchor="storage-and-resource-consumption"><name>Storage and Resource Consumption</name>

<t>Retaining historical versions of objects can consume significant
server storage.  Servers <bcp14>SHOULD</bcp14> impose reasonable limits on
history retention, either by duration (advertised via
<spanx style="verb">maxHistoryDuration</spanx>) or by total storage.  Servers <bcp14>MAY</bcp14>
silently discard history entries when storage limits are
reached.</t>

<t>Counting retained history towards a user's quota (<xref target="JMAP-QUOTAS"/>)
is <bcp14>NOT RECOMMENDED</bcp14>.  If it counts, a user who destroys objects to
free space sees no change until the history expires.  A restored
object counts towards quota as any newly created object does.</t>

</section>
<section anchor="access-control"><name>Access Control</name>

<t>Historical versions of an object <bcp14>MUST</bcp14> respect the same access
controls as the current version.  A user who does not have
permission to read an object <bcp14>MUST NOT</bcp14> be able to read its
history.</t>

<t>Access rights can change over time, for example when an object is
shared with a new user, or moved into a shared container.  The
server <bcp14>MUST NOT</bcp14> return a historical version to a user who could not
have read that version at the time it was current.  This requires
the server to track changes to access rights.  A server that does
not keep a full history of access rights can take a conservative
approach: record the point at which the access rights affecting an
object (or its containing object, such as a Mailbox or AddressBook)
last changed, for example as a modseq, and not return any version
that was replaced before that point.</t>

<t>Similarly, for destroyed objects, the server <bcp14>MUST NOT</bcp14> reveal the
existence or content of a destroyed object to a user who could not
read it at the time it was destroyed.</t>

</section>
<section anchor="information-disclosure"><name>Information Disclosure</name>

<t>Object history may reveal information that the user or an
administrator intended to remove.  For example, a contact
record that was updated to remove a phone number will still
have that phone number visible in its history.  Servers
<bcp14>SHOULD</bcp14> provide administrators with the ability to purge
history for specific objects when required for privacy or
compliance reasons.</t>

</section>
<section anchor="privacy-considerations"><name>Privacy Considerations</name>

<t>When a server supports this extension, updating or destroying an
object does not immediately remove the previous content from the
server; it remains retrievable by anyone with read access to the
object until the server discards it under its retention policy.
Users may not expect this.  This is not new in kind (servers
commonly retain deleted data in backups, and many mail systems
offer a "recently deleted" view), but this extension makes such
data retrievable through the protocol.</t>

<t>Servers <bcp14>SHOULD</bcp14> make their retention policy visible to users, for
example through the <spanx style="verb">maxHistoryDuration</spanx> capability property and
the server's user interface.  Clients that present a "destroy"
operation to users <bcp14>SHOULD NOT</bcp14> describe it as irreversible or
permanent when the server advertises this capability for that data
type.  Whether, and how, a user can request that history be purged
before the end of the retention period is a matter of server
policy and is outside the scope of this document.</t>

</section>
<section anchor="harmful-content"><name>Harmful Content</name>

<t>An administrator may remove content from an account because it is
harmful, for example a phishing message or malware.  The server <bcp14>MAY</bcp14>
decline to retain history for such an object, and <bcp14>MAY</bcp14> reject an
attempt to restore it with a "forbidden" SetError, so that the user
cannot bring the content back.</t>

</section>
</section>
<section anchor="changes"><name>Changes</name>

<t>EDITOR: please remove this section before publication.</t>

<t>The source of this document exists on github at: https://github.com/brong/draft-gondwana-jmap-object-history/</t>

<t><strong>draft-ietf-jmap-object-history-04</strong></t>

<t><list style="symbols">
  <t>Restores of objects of the same type that refer to each other,
such as a Mailbox and its child, are ordered by the server, as
for creates in Foo/set.</t>
  <t>Clarified that a restored object takes the server-set properties
of the final version; the patch may include them only with the
same values, as for any update.</t>
  <t>A restore <bcp14>MUST NOT</bcp14> change a property in a way an update could
not, so an Email keeps its <spanx style="verb">threadId</spanx>; a server that cannot do
this rejects the restore with "forbidden".  Removed the example
of a new <spanx style="verb">threadId</spanx> when the Thread no longer exists.</t>
  <t>A missing container is restored in an earlier method call or
request, not the same call.</t>
  <t>Foo/changes reports a restored id by comparing <spanx style="verb">sinceState</spanx> with
the new state, as if the object had never been destroyed; it can
now be reported as destroyed, or not at all.</t>
  <t>Thanks to John Coffey for review.</t>
  <t>Restores count towards "maxObjectsInSet", which limits the total
number of objects a call creates, updates, destroys, and restores.</t>
  <t>Foo/queryChanges takes <spanx style="verb">onlyDestroyed</spanx>, <spanx style="verb">destroyedAfter</spanx>, and
<spanx style="verb">destroyedBefore</spanx>, with the values passed to Foo/query.</t>
  <t>If the server has discarded the version that was current when an
object was destroyed, the most recent version it retains is the
final version.</t>
  <t>A restore is part of an atomic Foo/set <xref target="JMAP-CONDITIONAL"/>: a
failed restore causes "atomicFailure", which may carry
<spanx style="verb">notRestored</spanx>, and "ifUnchangedBy" is evaluated before any restore.
Thanks to Mark Atwood for raising this.</t>
</list></t>

<t><strong>draft-ietf-jmap-object-history-03</strong></t>

<t><list style="symbols">
  <t>Added John Coffey as co-author; folded in material from
draft-coffey-jmap-undelete.</t>
  <t>Added Foo/query arguments <spanx style="verb">onlyDestroyed</spanx>, <spanx style="verb">destroyedAfter</spanx>, and
<spanx style="verb">destroyedBefore</spanx>, for finding destroyed objects without an old
state string.</t>
  <t>Replaced restoring by creating with the old id (and Email/import
with the old id) with a new Foo/set <spanx style="verb">restore</spanx> argument, which
restores the final version with an optional patch, keeping
server-set properties.  Notes on Email, SieveScript, missing
containers, and conflicts.</t>
  <t>Added a per-version <spanx style="verb">expires</spanx> to <spanx style="verb">objectHistory</spanx>.</t>
  <t>Destruction through other protocols and by the server counts as
destruction.</t>
  <t>Security: quota, harmful content.</t>
  <t>Paging several ids with <spanx style="verb">historyBefore</spanx> takes one Foo/get call per
id.</t>
  <t>The history arguments are ignored unless <spanx style="verb">includeReplaced</spanx> or
<spanx style="verb">includeDestroyed</spanx> is true.</t>
  <t>Foo/changes reports a restored id as updated or created, never as
destroyed.</t>
  <t>Versions that are not current do not count as blob references.</t>
  <t>Added FileNode to the "Supports History" table.</t>
  <t>Thanks to Mark Atwood for an implementation review.</t>
</list></t>

<t><strong>draft-ietf-jmap-object-history-02</strong></t>

<t><list style="symbols">
  <t>Replaced the top-level <spanx style="verb">hasMoreHistory</spanx> response property with a
per-entry <spanx style="verb">objectHistory.hasOlder</spanx> property, so the client can
tell which ids have more history.</t>
  <t>Added <spanx style="verb">historySkip</spanx>, to skip the N most recent versions of each
id; ids with N or fewer versions are returned in <spanx style="verb">notFound</spanx>.</t>
</list></t>

<t><strong>draft-ietf-jmap-object-history-01</strong></t>

<t><list style="symbols">
  <t>Removed "Updates: 8620"; the JMAP Data Types registry is not
defined by RFC 8620.  The registry is now referenced by name
rather than by its defining document.</t>
  <t>Corrected the registry table to match the current registry.</t>
  <t><spanx style="verb">historyLimit</spanx> is now per id.</t>
  <t>Defined behaviour for <spanx style="verb">ids: null</spanx> with <spanx style="verb">includeDestroyed</spanx>, and
clarified that "maxObjectsInGet" counts ids, not entries.</t>
  <t>Ids with no matching versions are returned in <spanx style="verb">notFound</spanx>.</t>
  <t>Which changes create a version, and the values of server-computed
properties, are server dependent.</t>
  <t>Blobs referenced by returned versions must be downloadable.</t>
  <t>Expanded access control considerations for changing access rights.</t>
  <t>Added <spanx style="verb">supportedTypes</spanx> to the account capability; requesting
history for an unsupported type is an error.</t>
  <t>Reserved the <spanx style="verb">objectHistory</spanx> property name in the JSContact and
JSCalendar property registries.</t>
  <t>Added acknowledgements.</t>
  <t>Added a Design Rationale section (Foo/get rather than a new method,
embedded metadata, full objects rather than patches, timestamps).</t>
  <t>Added a Restoring Objects section: a client may ask to recreate a
destroyed object with its old id, and the server may accept or
reject; ids of destroyed objects are otherwise not reused.</t>
  <t><spanx style="verb">maxHistoryDuration</spanx> is now a best-effort statement of server policy.</t>
  <t>Clarified that "objectHistory" may appear in <spanx style="verb">properties</spanx>.</t>
  <t>Added Privacy Considerations.</t>
  <t>Added <spanx style="verb">historyBefore</spanx> for paging backwards through history;
versions of the same object have distinct <spanx style="verb">replaced</spanx> values, with
at most millisecond precision.</t>
  <t>Removed the audit use case from the introduction.</t>
</list></t>

<t><strong>draft-ietf-jmap-object-history-00</strong></t>

<t><list style="symbols">
  <t>Uploaded with working-group adopted name.</t>
</list></t>

<t><strong>draft-gondwana-jmap-object-history-00</strong></t>

<t><list style="symbols">
  <t>initial proposal</t>
</list></t>

</section>
<section anchor="acknowledgements"><name>Acknowledgements</name>

<t>The authors would like to thank Mark Atwood, Ben Bucksch, Mauro De
Gennaro, Arnt Gulbrandsen, Neil Jenkins, and Robert Stepanek for their
feedback on this document.</t>

</section>


  </middle>

  <back>


<references title='References' anchor="sec-combined-references">

    <references title='Normative References' anchor="sec-normative-references">



<reference anchor="JMAP-CORE">
  <front>
    <title>The JSON Meta Application Protocol (JMAP)</title>
    <author fullname="N. Jenkins" initials="N." surname="Jenkins"/>
    <author fullname="C. Newman" initials="C." surname="Newman"/>
    <date month="July" year="2019"/>
    <abstract>
      <t>This document specifies a protocol for clients to efficiently query, fetch, and modify JSON-based data objects, with support for push notification of changes and fast resynchronisation and for out-of- band binary data upload/download.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="8620"/>
  <seriesInfo name="DOI" value="10.17487/RFC8620"/>
</reference>

<reference anchor="RFC2119">
  <front>
    <title>Key words for use in RFCs to Indicate Requirement Levels</title>
    <author fullname="S. Bradner" initials="S." surname="Bradner"/>
    <date month="March" year="1997"/>
    <abstract>
      <t>In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements.</t>
    </abstract>
  </front>
  <seriesInfo name="BCP" value="14"/>
  <seriesInfo name="RFC" value="2119"/>
  <seriesInfo name="DOI" value="10.17487/RFC2119"/>
</reference>

<reference anchor="RFC8174">
  <front>
    <title>Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words</title>
    <author fullname="B. Leiba" initials="B." surname="Leiba"/>
    <date month="May" year="2017"/>
    <abstract>
      <t>RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings.</t>
    </abstract>
  </front>
  <seriesInfo name="BCP" value="14"/>
  <seriesInfo name="RFC" value="8174"/>
  <seriesInfo name="DOI" value="10.17487/RFC8174"/>
</reference>




    </references>

    <references title='Informative References' anchor="sec-informative-references">



<reference anchor="JMAP-MAIL">
  <front>
    <title>The JSON Meta Application Protocol (JMAP) for Mail</title>
    <author fullname="N. Jenkins" initials="N." surname="Jenkins"/>
    <author fullname="C. Newman" initials="C." surname="Newman"/>
    <date month="August" year="2019"/>
    <abstract>
      <t>This document specifies a data model for synchronising email data with a server using the JSON Meta Application Protocol (JMAP). Clients can use this to efficiently search, access, organise, and send messages, and to get push notifications for fast resynchronisation when new messages are delivered or a change is made in another client.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="8621"/>
  <seriesInfo name="DOI" value="10.17487/RFC8621"/>
</reference>

<reference anchor="JMAP-CONTACTS">
  <front>
    <title>JSON Meta Application Protocol (JMAP) for Contacts</title>
    <author fullname="N. Jenkins" initials="N." role="editor" surname="Jenkins"/>
    <date month="December" year="2024"/>
    <abstract>
      <t>This document specifies a data model for synchronising contact data with a server using the JSON Meta Application Protocol (JMAP).</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9610"/>
  <seriesInfo name="DOI" value="10.17487/RFC9610"/>
</reference>


<reference anchor="JMAP-BLOBEXT">
   <front>
      <title>JMAP Blob Management</title>
      <author fullname="Bron Gondwana" initials="B." surname="Gondwana">
         <organization>Fastmail</organization>
      </author>
      <date day="2" month="November" year="2026"/>
      <abstract>
	 <t>   The JSON Meta Application Protocol (JMAP) base protocol ([JMAP-CORE])
   provides the ability to upload and download arbitrary binary data via
   HTTP POST and GET on a defined endpoint.  This binary data is called
   a &quot;blob&quot;.

   This extension adds additional ways to create and access blobs by
   making inline method calls within a standard JMAP request.  It also
   adds a reverse lookup mechanism to discover where blobs are
   referenced within other data types, support for large blobs via
   chunked construction with server-side optimisation, and server-side
   blob conversion operations including image format conversion, archive
   creation and extraction, compression and decompression, and delta/
   patch operations.

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-ietf-jmap-blobext-03"/>
   
</reference>


<reference anchor="JMAP-CONDITIONAL">
   <front>
      <title>JMAP Conditional Set</title>
      <author fullname="Bron Gondwana" initials="B." surname="Gondwana">
         <organization>Fastmail</organization>
      </author>
      <date day="26" month="October" year="2026"/>
      <abstract>
	 <t>   The JMAP base protocol ([JMAP-CORE]) provides the Foo/set method for
   creating, updating, and destroying objects.  It offers a single
   concurrency control, the &quot;ifInState&quot; argument, which guards an entire
   object type: if any object of that type has changed, the whole method
   is rejected.

   This extension adds a finer, per-object conditional mechanism.  A
   client may require that an individual update or destroy proceed only
   if the target object still matches a set of expected property values,
   expressed using the JMAP PatchObject already defined for updates.
   This provides optimistic concurrency control scoped to a single
   object — the equivalent of an HTTP &quot;If-Match&quot; precondition — for any
   JMAP data type.

   This extension also defines an optional &quot;atomic&quot; argument that
   applies an entire Foo/set as a single unit: either every change it
   requests takes effect, or none does.  Combined with the per-object
   precondition, this lets a client express a multi-object change that
   is safe only when applied together — such as an atomic rename that
   exchanges two names.

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-ietf-jmap-conditional-02"/>
   
</reference>


<reference anchor="JMAP-FILENODE">
   <front>
      <title>JMAP File Storage extension</title>
      <author fullname="Bron Gondwana" initials="B." surname="Gondwana">
         <organization>Fastmail</organization>
      </author>
      <date day="2" month="November" year="2026"/>
      <abstract>
	 <t>   The JMAP base protocol (RFC 8620) provides the ability to upload and
   download arbitrary binary data.  This binary data is called a &quot;blob&quot;,
   and can be used in all other JMAP extensions.

   This extension adds a FileNode data type and methods to expose blobs
   as a filesystem, along with the types of metadata that are provided
   by other remote filesystem protocols.

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-ietf-jmap-filenode-20"/>
   
</reference>

<reference anchor="JMAP-QUOTAS">
  <front>
    <title>JSON Meta Application Protocol (JMAP) for Quotas</title>
    <author fullname="R. Cordier" initials="R." role="editor" surname="Cordier"/>
    <date month="June" year="2023"/>
    <abstract>
      <t>This document specifies a data model for handling quotas on accounts with a server using the JSON Meta Application Protocol (JMAP).</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9425"/>
  <seriesInfo name="DOI" value="10.17487/RFC9425"/>
</reference>

<reference anchor="JMAP-SIEVE">
  <front>
    <title>The JSON Meta Application Protocol (JMAP) for Sieve Scripts</title>
    <author fullname="K. Murchison" initials="K." surname="Murchison"/>
    <date month="September" year="2024"/>
    <abstract>
      <t>This document specifies a data model for managing Sieve scripts on a server using the JSON Meta Application Protocol (JMAP). Clients can use this protocol to efficiently search, access, organize, and validate Sieve scripts.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9661"/>
  <seriesInfo name="DOI" value="10.17487/RFC9661"/>
</reference>

<reference anchor="IMAP">
  <front>
    <title>Internet Message Access Protocol (IMAP) - Version 4rev2</title>
    <author fullname="A. Melnikov" initials="A." role="editor" surname="Melnikov"/>
    <author fullname="B. Leiba" initials="B." role="editor" surname="Leiba"/>
    <date month="August" year="2021"/>
    <abstract>
      <t>The Internet Message Access Protocol Version 4rev2 (IMAP4rev2) allows a client to access and manipulate electronic mail messages on a server. IMAP4rev2 permits manipulation of mailboxes (remote message folders) in a way that is functionally equivalent to local folders. IMAP4rev2 also provides the capability for an offline client to resynchronize with the server.</t>
      <t>IMAP4rev2 includes operations for creating, deleting, and renaming mailboxes; checking for new messages; removing messages permanently; setting and clearing flags; parsing per RFCs 5322, 2045, and 2231; searching; and selective fetching of message attributes, texts, and portions thereof. Messages in IMAP4rev2 are accessed by the use of numbers. These numbers are either message sequence numbers or unique identifiers.</t>
      <t>IMAP4rev2 does not specify a means of posting mail; this function is handled by a mail submission protocol such as the one specified in RFC 6409.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9051"/>
  <seriesInfo name="DOI" value="10.17487/RFC9051"/>
</reference>

<reference anchor="POP3">
  <front>
    <title>Post Office Protocol - Version 3</title>
    <author fullname="J. Myers" initials="J." surname="Myers"/>
    <author fullname="M. Rose" initials="M." surname="Rose"/>
    <date month="May" year="1996"/>
    <abstract>
      <t>The Post Office Protocol - Version 3 (POP3) is intended to permit a workstation to dynamically access a maildrop on a server host in a useful fashion. [STANDARDS-TRACK]</t>
    </abstract>
  </front>
  <seriesInfo name="STD" value="53"/>
  <seriesInfo name="RFC" value="1939"/>
  <seriesInfo name="DOI" value="10.17487/RFC1939"/>
</reference>

<reference anchor="JSCONTACT">
  <front>
    <title>JSContact: A JSON Representation of Contact Data</title>
    <author fullname="R. Stepanek" initials="R." surname="Stepanek"/>
    <author fullname="M. Loffredo" initials="M." surname="Loffredo"/>
    <date month="May" year="2024"/>
    <abstract>
      <t>This specification defines a data model and JavaScript Object Notation (JSON) representation of contact card information that can be used for data storage and exchange in address book or directory applications. It aims to be an alternative to the vCard data format and to be unambiguous, extendable, and simple to process. In contrast to the JSON-based jCard format, it is not a direct mapping from the vCard data model and expands semantics where appropriate. Two additional specifications define new vCard elements and how to convert between JSContact and vCard.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="9553"/>
  <seriesInfo name="DOI" value="10.17487/RFC9553"/>
</reference>

<reference anchor="JSCALENDAR">
  <front>
    <title>JSCalendar: A JSON Representation of Calendar Data</title>
    <author fullname="N. Jenkins" initials="N." surname="Jenkins"/>
    <author fullname="R. Stepanek" initials="R." surname="Stepanek"/>
    <date month="July" year="2021"/>
    <abstract>
      <t>This specification defines a data model and JSON representation of calendar data that can be used for storage and data exchange in a calendaring and scheduling environment. It aims to be an alternative and, over time, successor to the widely deployed iCalendar data format. It also aims to be unambiguous, extendable, and simple to process. In contrast to the jCal format, which is also based on JSON, JSCalendar is not a direct mapping from iCalendar but defines the data model independently and expands semantics where appropriate.</t>
    </abstract>
  </front>
  <seriesInfo name="RFC" value="8984"/>
  <seriesInfo name="DOI" value="10.17487/RFC8984"/>
</reference>


<reference anchor="IANA-JMAP" target="https://www.iana.org/assignments/jmap/">
  <front>
    <title>JSON Meta Application Protocol (JMAP)</title>
    <author >
      <organization>IANA</organization>
    </author>
    <date year="n.d."/>
  </front>
</reference>


    </references>

</references>



  </back>

<!-- ##markdown-source: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-->

</rfc>

