comparison mercurial/help/internals/wireprotocolv2.txt @ 40178:46a40bce3ae0

wireprotov2: define and implement "filesdata" command

Previously, the only way to access file revision data was the
"filedata" command. This command is useful to have. But, it only
allowed resolving revision data for a single file. This meant that
clients needed to send 1 command for each tracked path they were
seeking data on. Furthermore, those commands would need to enumerate
the exact file nodes they wanted data for.

This approach meant that clients were sending a lot of data to
remotes in order to request file data. e.g. if there were 1M
file revisions, we'd need at least 20,000,000 bytes just to encode
file nodes! Many clients on the internet don't have that kind of
upload capacity.

In order to limit the amount of data that clients must send, we'll
need more efficient ways to request repository data.

This commit defines and implements a new "filesdata" command. This
command allows the retrieval of data for multiple files by specifying
changeset revisions and optional file patterns. The command figures
out what file revisions are "relevant" and sends them in bulk.

The logic around choosing which file revisions to send in the case of
haveparents not being set is overly simple and will over-send files. We
will need more smarts here eventually. (Specifically, the client will
need to tell the server which revisions it knows about.) This work
is deferred until a later time.

Differential Revision: https://phab.mercurial-scm.org/D4981

40177:41e2633bcd00

When ``revision`` data is requested, the server chooses to emit either fulltext
revision data or a delta. What the server decides can be inferred by looking
for the presence of the ``delta`` or ``revision`` keys in the
``fieldsfollowing`` array.

heads
-----

Obtain DAG heads in the repository.

   The DAG range between ``roots`` and ``heads`` will be resolved and all
   revisions between will be used. Nodes in ``roots`` are not part of the
   resolved set. Nodes in ``heads`` are. The ``roots`` array may be empty.
   The ``heads`` array MUST be defined.

40178:46a40bce3ae0

When ``revision`` data is requested, the server chooses to emit either fulltext
revision data or a delta. What the server decides can be inferred by looking
for the presence of the ``delta`` or ``revision`` keys in the
``fieldsfollowing`` array.

filesdata
---------

Obtain various data related to multiple tracked files for specific changesets.

This command is similar to ``filedata`` with the main difference being that
individual requests operate on multiple file paths. This allows clients to
request data for multiple paths by issuing a single command.

The command accepts the following arguments:

fields
   (set of bytestring) Which data associated with a file to fetch.
   The following values are recognized:

   parents
      Parent nodes for the revision.

   revision
      The raw revision data for a file.

haveparents
   (bool) Whether the client has the parent revisions of all requested
   nodes.

pathfilter
   (map) Defines a filter that determines what file paths are relevant.

   See the *Path Filters* section for more.

   If the argument is omitted, it is assumed that all paths are relevant.

revisions
   (array of maps) Specifies revisions whose data is being requested. Each value
   in the array is a map describing revisions. See the *Revisions Specifiers*
   section below for the format of this map.

   Data will be sent for the union of all revisions resolved by all revision
   specifiers.

   Only revision specifiers operating on changeset revisions are allowed.

The response bytestream starts with a CBOR map describing the data that
follows. This map has the following bytestring keys:

totalpaths
   (unsigned integer) Total number of paths whose data is being transferred.

totalitems
   (unsigned integer) Total number of file revisions whose data is being
   transferred.

Following the map header are 0 or more sequences of CBOR values. Each sequence
represents data for a specific tracked path. Each sequence begins with a CBOR
map describing the file data that follows. Following that map is N CBOR values
describing file revision data. The format of this data is identical to that
returned by the ``filedata`` command.

Each sequence's map header has the following bytestring keys:

path
   (bytestring) The tracked file path whose data follows.

totalitems
   (unsigned integer) Total number of file revisions whose data is being
   transferred.

The ``haveparents`` argument has significant implications on the data
transferred.

When ``haveparents`` is true, the command MAY only emit data for file
revisions introduced by the set of changeset revisions whose data is being
requested. In other words, the command may assume that all file revisions
for all relevant paths for ancestors of the requested changeset revisions
are present on the receiver.

When ``haveparents`` is false, the command MUST assume that the receiver
has no file revisions data. This means that all referenced file revisions
in the queried set of changeset revisions will be sent.

TODO we'll probably want a more complicated mechanism for the client to
specify which ancestor revisions are known.

heads
-----

Obtain DAG heads in the repository.

   The DAG range between ``roots`` and ``heads`` will be resolved and all
   revisions between will be used. Nodes in ``roots`` are not part of the
   resolved set. Nodes in ``heads`` are. The ``roots`` array may be empty.
   The ``heads`` array MUST be defined.

Path Filters
============

Various commands accept a *path filter* argument that defines the set of file
paths relevant to the request.

A *path filter* is defined as a map with the bytestring keys ``include`` and
``exclude``. Each is an array of bytestring values. Each value defines a pattern
rule (see :hg:`help patterns`) that is used to match file paths.

A path matches the path filter if it is matched by a rule in the ``include``
set but doesn't match a rule in the ``exclude`` set. In other words, a path
matcher takes the union of all ``include`` patterns and then substracts the
union of all ``exclude`` patterns.

Patterns MUST be prefixed with their pattern type. Only the following pattern
types are allowed: ``path:``, ``rootfilesin:``.

If the ``include`` key is omitted, it is assumed that all paths are
relevant. The patterns from ``exclude`` will still be used, if defined.

An example value is ``path:tests/foo``, which would match a file named
``tests/foo`` or a directory ``tests/foo`` and all files under it.