Patchwork [2,of,6,zstd-wireproto,V3] internals: document compression negotiation

mail settings
Submitter Gregory Szorc
Date Dec. 24, 2016, 10:33 p.m.
Message ID <b60c2cbd013a31ca33de.1482618786@gps-mbp.local>
Download mbox | patch
Permalink /patch/18032/
State Accepted
Headers show


Gregory Szorc - Dec. 24, 2016, 10:33 p.m.
# HG changeset patch
# User Gregory Szorc <>
# Date 1482612996 25200
#      Sat Dec 24 13:56:36 2016 -0700
# Node ID b60c2cbd013a31ca33defab813388950024db5be
# Parent  13890ebe4322e735a589409b5d62da9555fae7d2
internals: document compression negotiation

As part of adding zstd support to all of the things, we'll need
to teach the wire protocol to support non-zlib compression formats.

This commit documents how we'll implement that.

To understand how we arrived at this proposal, let's look at how
things are done today.

The wire protocol today doesn't have a unified format. Instead,
there is a limited facility for differentiating replies as successful
or not. And, each command essentially defines its own response format.

A significant deficiency in the current protocol is the lack of
payload framing over the SSH transport. In the HTTP transport,
chunked transfer is used and the end of an HTTP response body (and
the end of a Mercurial command response) can be identified by a 0
length chunk. This is how HTTP chunked transfer works. But in the
SSH transport, there is no such framing, at least for certain
responses (notably the response to "getbundle" requests). Clients
can't simply read until end of stream because the socket is
persistent and reused for multiple requests. Clients need to know
when they've encountered the end of a request but there is nothing
simple for them to key off of to detect this. So what happens is
the client must decode the payload (as opposed to being dumb and
forwarding frames/packets). This means the payload itself needs
to support identifying end of stream. In some cases (bundle2), it
also means the payload can encode "error" or "interrupt" events
telling the client to e.g. abort processing. The lack of framing
on the SSH transport and the transfer of its responsibilities to
e.g. bundle2 is a massive layering violation and a wart on the
protocol architecture. It needs to be fixed someday by inventing a
proper framing protocol.

So about compression.

The client transport abstractions have a "_callcompressable()"
API. This API is called to invoke a remote command that will
send a compressible response. The response is essentially a
"streaming" response (no framing data at the Mercurial layer)
that is fed into a decompressor.

On the HTTP transport, the decompressor is zlib and only zlib.
There is currently no mechanism for the client to specify an
alternate compression format. And, clients don't advertise what
compression formats they support or ask the server to send a
specific compression format. Instead, it is assumed that non-error
responses to "compressible" commands are zlib compressed.

On the SSH transport, there is no compression at the Mercurial
protocol layer. Instead, compression must be handled by SSH
itself (e.g. `ssh -C`) or within the payload data (e.g. bundle

For the HTTP transport, adding new compression formats is pretty
straightforward. Once you know what decompressor to use, you can
stream data into the decompressor until you reach a 0 size HTTP
chunk, at which point you are at end of stream.

So our wire protocol changes for the HTTP transport are pretty
straightforward: the client and server advertise what compression
formats they support and an appropriate compression format is
chosen. We introduce a new HTTP media type to hold compressed
payloads. The header of the payload defines the compression format
being used. Whoever is on the receiving end can sniff the first few
bytes route to an appropriate decompressor.

Support for multiple compression formats is advertised on both
server and client. The server advertises a "compression" capability
saying which compression formats it supports and in what order they
are preferred. Clients advertise their support for multiple
compression formats and media types via the introduced "X-HgProto"
request header.

Strictly speaking, servers don't need to advertise which compression
formats they support. But doing so allows clients to fail fast if
they don't support any of the formats the server does. This is useful
in situations like sending bundles, where the client may have to
perform expensive computation before sending data to the server.

Rather than simply advertise a list of supported compression formats,
we introduce an additional "httpmediatype" server capability
advertising which media types the server supports. This means servers
are explicit about what formats they exchange. IMO, this is superior
to inferring support from other capabilities (like "compression").

By advertising compression support on each request in the "X-HgProto"
header and media type and direction at the server level, we are able
to gradually transition existing commands/responses to the new media
type and possibly compression. Contrast with the old world, where we
only supported a single media type and the use of compression was
built-in to the semantics of the command on both client and server.
In the new world, if "application/mercurial-0.2" is supported,
compression is supported. It's that simple.

It's worth noting that we explicitly don't use "Accept,"
"Accept-Encoding," "Content-Encoding," or "Transfer-Encoding" for
content negotiation and compression. People knowledgeable of the HTTP
specifications will say that we should use these because that's
what they are designed to be used for. They have a point and I
sympathize with the argument. Earlier versions of this commit even
defined supported media types in the "Accept" header. However, my
years of experience rolling out services leveraging HTTP has taught
me to not trust the HTTP layer, especially if you are going outside
the normal spec (such as using a custom "Content-Encoding" value to
represent zstd streams). I've seen load balancers, proxies, and other
network devices do very bad and unexpected things to HTTP messages
(like insisting zlib compressed content is decoded and then re-encoded
at a different compression level or even stripping compression
completely). I've found that the best way to avoid surprises when
writing protocols on top of HTTP is to use HTTP as a dumb transport as
much as possible to minimize the chances that an "intelligent" agent
between endpoints will muck with your data. While the widespread use of
TLS is mitigating many intermediate network agents interfering with
HTTP, there are still problems at the edges, with e.g. the origin HTTP
server needing to convert HTTP to and from WSGI and buggy or
feature-lacking HTTP client implementations. I've found the best way to
avoid these problems is to avoid using headers like "Content-Encoding"
and to bake as much logic as possible into media types and HTTP message
bodies. The protocol changes in this commit do rely on a custom HTTP
request header and the "Content-Type" headers. But we used them before,
so we shouldn't be increasing our exposure to "bad" HTTP agents.

For the SSH transport, we can't easily implement content negotiation
to determine compression formats because the SSH transport has no
content negotiation capabilities today. And without a framing protocol,
we don't know how much data to feed into a decompressor. So in order
to implement compression support on the SSH transport, we'd need to
invent a mechanism to represent content types and an outer framing
protocol to stream data robustly. While I'm fully capable of doing
that, it is a lot of work and not something that should be undertaken
lightly. My opinion is that if we're going to change the SSH transport
protocol, we should take a long hard look at implementing a grand
unified protocol that attempts to address all the deficiencies with
the existing protocol. While I want this to happen, that would be
massive scope bloat standing in the way of zstd support. So, I've
decided to take the easy solution: the SSH transport will not gain
support for multiple compression formats. Keep in mind it doesn't
support *any* compression today. So essentially nothing is changing
on the SSH front.


diff --git a/mercurial/help/internals/wireprotocol.txt b/mercurial/help/internals/wireprotocol.txt
--- a/mercurial/help/internals/wireprotocol.txt
+++ b/mercurial/help/internals/wireprotocol.txt
@@ -64,13 +64,29 @@  Example HTTP requests::
     GET /repo?cmd=capabilities
     X-HgArg-1: foo=bar&baz=hello%20world
+The request media type should be chosen based on server support. If the
+``httpmediatype`` server capability is present, the client should send
+the newest mutually supported media type. If this capability is absent,
+the client must assume the server only supports the
+``application/mercurial-0.1`` media type.
 The ``Content-Type`` HTTP response header identifies the response as coming
 from Mercurial and can also be used to signal an error has occurred.
-The ``application/mercurial-0.1`` media type indicates a generic Mercurial
-response. It matches the media type sent by the client.
+The ``application/mercurial-*`` media types indicate a generic Mercurial
+data type.
+The ``application/mercurial-0.1`` media type is raw Mercurial data. It is the
+predecessor of the format below.
+The ``application/mercurial-0.2`` media type is compression framed Mercurial
+data. The first byte of the payload indicates the length of the compression
+format identifier that follows. Next are N bytes indicating the compression
+format. e.g. ``zlib``. The remaining bytes are compressed according to that
+compression format. The decompressed data behaves the same as with
 The ``application/hg-error`` media type indicates a generic error occurred.
 The content of the HTTP response body typically holds text describing the
@@ -80,17 +96,21 @@  type.
 Clients also accept the ``text/plain`` media type. All other media
 types should cause the client to error.
+Behavior of media types is further described in the ``Content Negotiation``
+section below.
 Clients should issue a ``User-Agent`` request header that identifies the client.
 The server should not use the ``User-Agent`` for feature detection.
-A command returning a ``string`` response issues the
-``application/mercurial-0.1`` media type and the HTTP response body contains
-the raw string value. A ``Content-Length`` header is typically issued.
+A command returning a ``string`` response issues a
+``application/mercurial-0.*`` media type and the HTTP response body contains
+the raw string value (after compression decoding, if used). A
+``Content-Length`` header is typically issued, but not required.
-A command returning a ``stream`` response issues the
-``application/mercurial-0.1`` media type and the HTTP response is typically
+A command returning a ``stream`` response issues a
+``application/mercurial-0.*`` media type and the HTTP response is typically
 using *chunked transfer* (``Transfer-Encoding: chunked``).
 SSH Transport
@@ -232,8 +252,26 @@  2006).
 This capability was introduced at the same time as the ``lookup``
+Declares support for negotiating compression formats.
+Presence of this capability indicates the server supports dynamic selection
+of compression formats based on the client request.
+Servers advertising this capability are required to support the
+``application/mercurial-0.2`` media type in response to commands returning
+streams. Servers may support this media type on any command.
+The value of the capability is a comma-delimited list of strings declaring
+supported compression formats. The order of the compression formats is in
+server-preferred order, most preferred first.
+This capability was introduced in Mercurial 4.1 (released February 2017).
 Whether the server supports the ``getbundle`` command.
@@ -251,8 +289,53 @@  length that clients should send. Clients
 comma in the value, as this is reserved for future use.
 This capability was introduced in Mercurial 1.9 (released July 2011).
+Indicates which HTTP media types (``Content-Type`` header) the server is
+capable of receiving and sending.
+The value of the capability is a comma-delimited list of strings identifying
+support for media type and transmission direction. The following strings may
+be present:
+   Indicates server support for receiving ``application/mercurial-0.1`` media
+   types.
+   Indicates server support for sending ``application/mercurial-0.1`` media
+   types.
+   Indicates server support for receiving ``application/mercurial-0.2`` media
+   types.
+   Indicates server support for sending ``application/mercurial-0.2`` media
+   types.
+   Minimum media type version the server is capable of receiving. Value is a
+   string like ``0.2``.
+   This capability can be used by servers to limit connections from legacy
+   clients not using the latest supported media type. However, only clients
+   with knowledge of this capability will know to consult this value. This
+   capability is present so the client may issue a more user-friendly error
+   when the server has locked out a legacy client.
+   Minimum media type version the server is capable of sending. Value is a
+   string like ``0.1``.
+Servers advertising support for the ``application/mercurial-0.2`` media type
+should also advertise the ``compression`` capability.
+This capability was introduced in Mercurial 4.1 (released February 2017).
@@ -415,8 +498,59 @@  announcements. Clients assume servers wi
 Mercurial server replies to the client-issued commands. So any server output
 not conforming to the expected command responses is assumed to be not related
 to Mercurial and can be ignored.
+Content Negotiation
+The wire protocol has some mechanisms to help peers determine what content
+types and encoding the other side will accept. Historically, these mechanisms
+have been built into commands themselves because most commands only send a
+well-defined response type and only certain commands needed to support
+functionality like compression.
+Currently, only the HTTP transport supports content negotiation at the protocol
+HTTP requests advertise supported response formats via the ``X-HgProto-<N>``
+request header, where ``<N>`` is an integer starting at 1 allowing the logical
+value to span multiple headers. This value consists of a list of
+space-delimited parameters. Each parameter denotes a feature or capability.
+The following parameters are defined:
+   Indicates the client supports receiving ``application/mercurial-0.1``
+   responses.
+   Indicates the client supports receiving ``application/mercurial-0.2``
+   responses.
+   Indicates compression formats the client can decode. Value is a list of
+   comma delimited strings identifying compression formats ordered from
+   most preferential to least preferential. e.g. ``comp=zstd,zlib,none``.
+   This parameter does not have an effect if only the ``0.1`` parameter
+   is defined, as support for ``application/mercurial-0.2`` or greater is
+   required to use arbitrary compression formats.
+   If this parameter is not advertised, the server interprets this as
+   equivalent to ``zlib,none``.
+Clients may choose to only send this header if the ``httpmediatype``
+server capability is present, as currently all server-side features
+consulting this header require the client to opt in to new protocol features
+advertised via the ``httpmediatype`` capability.
+A server that doesn't receive an ``X-HgProto-<N>`` header should infer a
+value of ``0.1``. This is compatible with legacy clients.
+A server receiving a request indicating support for multiple media type
+versions may respond with any of the supported media types. Not all servers
+may support all media types on all commands.
 This section contains a list of all wire protocol commands implemented by