Bundle File Formats#
Mercurial supports generating standalone “bundle” files that hold repository data. These “bundles” are typically saved locally and used later or exchanged between different repositories, possibly on different machines. Example commands using bundles are `hg bundle` and `hg unbundle`.
Generation of bundle files is controlled by a “bundle specification” (“bundlespec”) string. This string tells the bundle generation process how to create the bundle.
A “bundlespec” string is composed of the following elements:
- type
A string denoting the bundle format to use.
- compression
Denotes the compression engine to use compressing the raw bundle data.
- parameters
Arbitrary key-value parameters to further control bundle generation.
A “bundlespec” string has the following formats:
- <type>
The literal bundle format string is used.
- <compression>-<type>
The compression engine and format are delimited by a hyphen (
-
).
Optional parameters follow the <type>
. Parameters are URI escaped
key=value
pairs. Each pair is delimited by a semicolon (;
). The
first parameter begins after a ;
immediately following the <type>
value.
Available Types#
The following bundle <type> strings are available:
- v1
Produces a legacy “changegroup” version 1 bundle.
This format is compatible with nearly all Mercurial clients because it is the oldest. However, it has some limitations, which is why it is no longer the default for new repositories.
v1
bundles can be used with modern repositories using the “generaldelta” storage format. However, it may take longer to produce the bundle and the resulting bundle may be significantly larger than av2
bundle.v1
bundles can only use thegzip
,bzip2
, andnone
compression formats.- v2
Produces a version 2 bundle.
Version 2 bundles are an extensible format that can store additional repository data (such as bookmarks and phases information) and they can store data more efficiently, resulting in smaller bundles.
Version 2 bundles can also use modern compression engines, such as
zstd
, making them faster to compress and often smaller.
Available Compression Engines#
The following bundle <compression> engines can be used:
The compression engines can be prepended with stream
to create a streaming bundle.
These are bundles that are extremely efficient to produce and consume,
but do not have guaranteed compatibility with older clients.
Available Options#
The following options exist:
- changegroup
Include the changegroup data in the bundle (default to True).
- cg.version
Select the version of the changegroup to use. Available options are : 01, 02 or 03. By default it will be automatically selected according to the current repository format.
- obsolescence
Include obsolescence-markers relevant to the bundled changesets.
- phases
Include phase information relevant to the bundled changesets.
- revbranchcache
Include the “tags-fnodes” cache inside the bundle.
- tagsfnodescache
Include the “tags-fnodes” cache inside the bundle.
Examples#
v2
Produce a
v2
bundle using default options, including compression.none-v1
Produce a
v1
bundle with no compression.zstd-v2
Produce a
v2
bundle with zstandard compression using default settings.zstd-v1
This errors because
zstd
is not supported forv1
types.none-streamv2
Produce a
v2
streaming bundle with no compression.zstd-v2;obsolescence=true;phases=true
Produce a
v2
bundle with zstandard compression which includes obsolescence markers and phases.