##// END OF EJS Templates
upstream » mercurial-mirror
RSS

Clone from https://www.mercurial-scm.org/repo/hg-all

wireproto: add streams to frame-based protocol...
wireproto: add streams to frame-based protocol Previously, the frame-based protocol was just a series of frames, with each frame associated with a request ID. In order to scale the protocol, we'll want to enable the use of compression. While it is possible to enable compression at the socket/pipe level, this has its disadvantages. The big one is it undermines the point of frames being standalone, atomic units that can be read and written: if you add compression above the framing protocol, you are back to having a stream-based protocol as opposed to something frame-based. So in order to preserve frames, compression needs to occur at the frame payload level. Compressing each frame's payload individually will limit compression ratios because the window size of the compressor will be limited by the max frame size, which is 32-64kb as currently defined. It will also add CPU overhead, as it is more efficient for compressors to operate on fewer, larger blocks of data than more, smaller blocks. So compressing each frame independently is out. This means we need to compress each frame's payload as if it is part of a larger stream. The simplest approach is to have 1 stream per connection. This could certainly work. However, it has disadvantages (documented below). We could also have 1 stream per RPC/command invocation. (This is the model HTTP/2 goes with.) This also has disadvantages. The main disadvantage to one global stream is that it has the very real potential to create CPU bottlenecks doing compression. Networks are only getting faster and the performance of single CPU cores has been relatively flat. Newer compression formats like zstandard offer better CPU cycle efficiency than predecessors like zlib. But it still all too common to saturate your CPU with compression overhead long before you saturate the network pipe. The main disadvantage with streams per request is that you can't reap the benefits of the compression context for multiple requests. For example, if you send 1000 RPC requests (or HTTP/2 requests for that matter), the response to each would have its own compression context. The overall size of the raw responses would be larger because compression contexts wouldn't be able to reference data from another request or response. The approach for streams as implemented in this commit is to support N streams per connection and for streams to potentially span requests and responses. As explained by the added internals docs, this facilitates servers and clients delegating independent streams and compression to independent threads / CPU cores. This helps alleviate the CPU bottleneck of compression. This design also allows compression contexts to be reused across requests/responses. This can result in improved compression ratios and less overhead for compressors and decompressors having to build new contexts. Another feature that was defined was the ability for individual frames within a stream to declare whether that individual frame's payload uses the content encoding (read: compression) defined by the stream. The idea here is that some servers may serve data from a combination of caches and dynamic resolution. Data coming from caches may be pre-compressed. We want to facilitate servers being able to essentially stream bytes from caches to the wire with minimal overhead. Being able to mix and match with frames are compressed within a stream enables these types of advanced server functionality. This commit defines the new streams mechanism. Basic code for supporting streams in frames has been added. But that code is seriously lacking and doesn't fully conform to the defined protocol. For example, we don't close any streams. And support for content encoding within streams is not yet implemented. The change was rather invasive and I didn't think it would be reasonable to implement the entire feature in a single commit. For the record, I would have loved to reuse an existing multiplexing protocol to build the new wire protocol on top of. However, I couldn't find a protocol that offers the performance and scaling characteristics that I desired. Namely, it should support multiple compression contexts to facilitate scaling out to multiple CPU cores and compression contexts should be able to live longer than single RPC requests. HTTP/2 *almost* fits the bill. But the semantics of HTTP message exchange state that streams can only live for a single request-response. We /could/ tunnel on top of HTTP/2 streams and frames with HEADER and DATA frames. But there's no guarantee that HTTP/2 libraries and proxies would allow us to use HTTP/2 streams and frames without the HTTP message exchange semantics defined in RFC 7540 Section 8. Other RPC protocols like gRPC tunnel are built on top of HTTP/2 and thus preserve its semantics of stream per RPC invocation. Even QUIC does this. We could attempt to invent a higher-level stream that spans HTTP/2 streams. But this would be violating HTTP/2 because there is no guarantee that HTTP/2 streams are routed to the same server. The best we can do - which is what this protocol does - is shoehorn all request and response data into a single HTTP message and create streams within. At that point, we've defined a Content-Type in HTTP parlance. It just so happens our media type can also work as a standalone, stream-based protocol, without leaning on HTTP or similar protocol. Differential Revision: https://phab.mercurial-scm.org/D2907
Yuya Nishihara - - Load All Authors

File last commit:

r35166:27196b7f stable
r37304:9bfcbe4f default
Show More
subrepos.txt
169 lines | 7.1 KiB | text/plain | TextLexer
/ mercurial / help / subrepos.txt
History | Annotation | Raw |Copy content |Copy permalink
Subrepositories let you nest external repositories or projects into a
parent Mercurial repository, and make commands operate on them as a
group.
Mercurial currently supports Mercurial, Git, and Subversion
subrepositories.
Subrepositories are made of three components:
1. Nested repository checkouts. They can appear anywhere in the
parent working directory.
2. Nested repository references. They are defined in ``.hgsub``, which
should be placed in the root of working directory, and
tell where the subrepository checkouts come from. Mercurial
subrepositories are referenced like::
path/to/nested = https://example.com/nested/repo/path
Git and Subversion subrepos are also supported::
path/to/nested = [git]git://example.com/nested/repo/path
path/to/nested = [svn]https://example.com/nested/trunk/path
where ``path/to/nested`` is the checkout location relatively to the
parent Mercurial root, and ``https://example.com/nested/repo/path``
is the source repository path. The source can also reference a
filesystem path.
Note that ``.hgsub`` does not exist by default in Mercurial
repositories, you have to create and add it to the parent
repository before using subrepositories.
3. Nested repository states. They are defined in ``.hgsubstate``, which
is placed in the root of working directory, and
capture whatever information is required to restore the
subrepositories to the state they were committed in a parent
repository changeset. Mercurial automatically record the nested
repositories states when committing in the parent repository.
.. note::
The ``.hgsubstate`` file should not be edited manually.
Adding a Subrepository
======================
If ``.hgsub`` does not exist, create it and add it to the parent
repository. Clone or checkout the external projects where you want it
to live in the parent repository. Edit ``.hgsub`` and add the
subrepository entry as described above. At this point, the
subrepository is tracked and the next commit will record its state in
``.hgsubstate`` and bind it to the committed changeset.
Synchronizing a Subrepository
=============================
Subrepos do not automatically track the latest changeset of their
sources. Instead, they are updated to the changeset that corresponds
with the changeset checked out in the top-level changeset. This is so
developers always get a consistent set of compatible code and
libraries when they update.
Thus, updating subrepos is a manual process. Simply check out target
subrepo at the desired revision, test in the top-level repo, then
commit in the parent repository to record the new combination.
Deleting a Subrepository
========================
To remove a subrepository from the parent repository, delete its
reference from ``.hgsub``, then remove its files.
Interaction with Mercurial Commands
===================================
:add: add does not recurse in subrepos unless -S/--subrepos is
specified. However, if you specify the full path of a file in a
subrepo, it will be added even without -S/--subrepos specified.
Subversion subrepositories are currently silently
ignored.
:addremove: addremove does not recurse into subrepos unless
-S/--subrepos is specified. However, if you specify the full
path of a directory in a subrepo, addremove will be performed on
it even without -S/--subrepos being specified. Git and
Subversion subrepositories will print a warning and continue.
:archive: archive does not recurse in subrepositories unless
-S/--subrepos is specified.
:cat: Git subrepositories only support exact file matches.
Subversion subrepositories are currently ignored.
:commit: commit creates a consistent snapshot of the state of the
entire project and its subrepositories. If any subrepositories
have been modified, Mercurial will abort. Mercurial can be made
to instead commit all modified subrepositories by specifying
-S/--subrepos, or setting "ui.commitsubrepos=True" in a
configuration file (see :hg:`help config`). After there are no
longer any modified subrepositories, it records their state and
finally commits it in the parent repository. The --addremove
option also honors the -S/--subrepos option. However, Git and
Subversion subrepositories will print a warning and abort.
:diff: diff does not recurse in subrepos unless -S/--subrepos is
specified. Changes are displayed as usual, on the subrepositories
elements. Subversion subrepositories are currently silently ignored.
:files: files does not recurse into subrepos unless -S/--subrepos is
specified. However, if you specify the full path of a file or
directory in a subrepo, it will be displayed even without
-S/--subrepos being specified. Git and Subversion subrepositories
are currently silently ignored.
:forget: forget currently only handles exact file matches in subrepos.
Git and Subversion subrepositories are currently silently ignored.
:incoming: incoming does not recurse in subrepos unless -S/--subrepos
is specified. Git and Subversion subrepositories are currently
silently ignored.
:outgoing: outgoing does not recurse in subrepos unless -S/--subrepos
is specified. Git and Subversion subrepositories are currently
silently ignored.
:pull: pull is not recursive since it is not clear what to pull prior
to running :hg:`update`. Listing and retrieving all
subrepositories changes referenced by the parent repository pulled
changesets is expensive at best, impossible in the Subversion
case.
:push: Mercurial will automatically push all subrepositories first
when the parent repository is being pushed. This ensures new
subrepository changes are available when referenced by top-level
repositories. Push is a no-op for Subversion subrepositories.
:serve: serve does not recurse into subrepositories unless
-S/--subrepos is specified. Git and Subversion subrepositories
are currently silently ignored.
:status: status does not recurse into subrepositories unless
-S/--subrepos is specified. Subrepository changes are displayed as
regular Mercurial changes on the subrepository
elements. Subversion subrepositories are currently silently
ignored.
:remove: remove does not recurse into subrepositories unless
-S/--subrepos is specified. However, if you specify a file or
directory path in a subrepo, it will be removed even without
-S/--subrepos. Git and Subversion subrepositories are currently
silently ignored.
:update: update restores the subrepos in the state they were
originally committed in target changeset. If the recorded
changeset is not available in the current subrepository, Mercurial
will pull it in first before updating. This means that updating
can require network access when using subrepositories.
Remapping Subrepositories Sources
=================================
A subrepository source location may change during a project life,
invalidating references stored in the parent repository history. To
fix this, rewriting rules can be defined in parent repository ``hgrc``
file or in Mercurial configuration. See the ``[subpaths]`` section in
hgrc(5) for more details.