Hi all,
I've finally had some time to write an asciidoc man page for git
subtree, which is included below for your convenience.
I added a new "--squash" option to the subtree merge command, which
lets you do cool stuff like switch your subtree branch back and forth
between upstream versions without actually losing any of your local
customizations. It also avoids the problem of polluting your
project's history: it pulls in only individual subproject trees (by
creating a synthetic commit for the desired tree, with its parent
being the *previous* synthetic commit). This saves repository size
(you don't slurp in all the intermediate trees), plus you don't get
all the commit messages from the subproject, which are probably
irrelevant to you anyway.
As always, comments welcome. Quite a few people have emailed me about
git subtree lately; I think the --squash mode will probably resolve
the problems for almost everyone who had them. Assuming it works
correctly, which is of course 100% guaranteed or your money back.
You can always find the latest git-subtree at:
git://github.com/apenwarr/git-subtree
Docs below. Check out the EXAMPLES section for a quick start.
Have fun,
Avery
git-subtree(1)
==============
NAME
----
git-subtree - add, merge, and split subprojects stored in subtrees
SYNOPSIS
--------
[verse]
'git subtree' add --prefix=<prefix> <commit>
'git subtree' merge --prefix=<prefix> <commit>
'git subtree' pull --prefix=<prefix> <repository> <refspec...>
'git subtree' split --prefix=<prefix> <commit...>
DESCRIPTION
-----------
git subtree allows you to include an subproject in your
own repository as a subdirectory, optionally including the
subproject's entire history. For example, you could
include the source code for a library as a subdirectory of your
application.
You can also extract the entire history of a subdirectory from
your project and make it into a standalone project. For
example, if a library you made for one application ends up being
useful elsewhere, you can extract its entire history and publish
that as its own git repository, without accidentally
intermingling the history of your application project.
Most importantly, you can alternate back and forth between these
two operations. If the standalone library gets updated, you can
automatically merge the changes into your project; if you
update the library inside your project, you can "split" the
changes back out again and merge them back into the library
project.
Unlike the 'git submodule' command, git subtree doesn't produce
any special constructions (like .gitmodule files or gitlinks) in
your repository, and doesn't require end-users of your
repository to do anything special or to understand how subtrees
work. A subtree is just another subdirectory and can be
committed to, branched, and merged along with your project in
any way you want.
In order to keep your commit messages clean, we recommend that
people split their commits between the subtrees and the main
project as much as possible. That is, if you make a change that
affects both the library and the main application, commit it in
two pieces. That way, when you split the library commits out
later, their descriptions will still make sense. But if this
isn't important to you, it's not *necessary*. git subtree will
simply leave out the non-library-related parts of the commit
when it splits it out into the subproject later.
COMMANDS
--------
add::
Create the <prefix> subtree by importing its contents
from the given commit. A new commit is created
automatically, joining the imported project's history
with your own. With '--squash', imports only a single
commit from the subproject, rather than its entire
history.
merge::
Merge recent changes up to <commit> into the <prefix>
subtree. As with normal 'git merge', this doesn't
remove your own local changes; it just merges those
changes into the latest <commit>. With '--squash',
creates only one commit that contains all the changes,
rather than merging in the entire history.
If you use '--squash', the merge direction doesn't
always have to be forward; you can use this command to
go back in time from v2.5 to v2.4, for example. If your
merge introduces a conflict, you can resolve it in the
usual ways.
pull::
Exactly like 'merge', but parallels 'git pull' in that
it fetches the given commit from the specified remote
repository.
split::
Extract a new, synthetic project history from the
history of the <prefix> subtree. The new history
includes only the commits (including merges) that
affected <prefix>, and each of those commits now has the
contents of <prefix> at the root of the project instead
of in a subdirectory. Thus, the newly created history
is suitable for export as a separate git repository.
After splitting successfully, a single commit id is
printed to stdout. This corresponds to the HEAD of the
newly created tree, which you can manipulate however you
want.
Repeated splits of exactly the same history are
guaranteed to be identical (ie. to produce the same
commit ids). Because of this, if you add new commits
and then re-split, the new commits will be attached as
commits on top of the history you generated last time,
so 'git merge' and friends will work as expected.
Note that if you use '--squash' when you merge, you
should usually not just '--rejoin' when you split.
OPTIONS
-------
-q::
--quiet::
Suppress unnecessary output messages on stderr.
-d::
--debug::
Produce even more unnecessary output messages on stderr.
--prefix=<prefix>::
Specify the path in the repository to the subtree you
want to manipulate. This option is currently mandatory
for all commands.
OPTIONS FOR add, merge, AND pull
--------------------------------
--squash::
Instead of merging the entire history from the subtree
project, produce only a single commit that contains all
the differences you want to merge, and then merge that
new commit into your project.
Using this option helps to reduce log clutter. People
rarely want to see every change that happened between
v1.0 and v1.1 of the library they're using, since none of the
interim versions were ever included in their application.
Using '--squash' also helps avoid problems when the same
subproject is included multiple times in the same
project, or is removed and then re-added. In such a
case, it doesn't make sense to combine the histories
anyway, since it's unclear which part of the history
belongs to which subtree.
Furthermore, with '--squash', you can switch back and
forth between different versions of a subtree, rather
than strictly forward. 'git subtree merge --squash'
always adjusts the subtree to match the exactly
specified commit, even if getting to that commit would
require undoing some changes that were added earlier.
Whether or not you use '--squash', changes made in your
local repository remain intact and can be later split
and send upstream to the subproject.
OPTIONS FOR split
-----------------
--annotate=<annotation>::
When generating synthetic history, add <annotation> as a
prefix to each commit message. Since we're creating new
commits with the same commit message, but possibly
different content, from the original commits, this can help
to differentiate them and avoid confusion.
Whenever you split, you need to use the same
<annotation>, or else you don't have a guarantee that
the new re-created history will be identical to the old
one. That will prevent merging from working correctly.
git subtree tries to make it work anyway, particularly
if you use --rejoin, but it may not always be effective.
-b <branch>::
--branch=<branch>::
After generating the synthetic history, create a new
branch called <branch> that contains the new history.
This is suitable for immediate pushing upstream.
<branch> must not already exist.
--ignore-joins::
If you use '--rejoin', git subtree attempts to optimize
its history reconstruction to generate only the new
commits since the last '--rejoin'. '--ignore-join'
disables this behaviour, forcing it to regenerate the
entire history. In a large project, this can take a
long time.
--onto=<onto>::
If your subtree was originally imported using something
other than git subtree, its history may not match what
git subtree is expecting. In that case, you can specify
the commit id <onto> that corresponds to the first
revision of the subproject's history that was imported
into your project, and git subtree will attempt to build
its history from there.
If you used 'git subtree add', you should never need
this option.
--rejoin::
After splitting, merge the newly created synthetic
history back into your main project. That way, future
splits can search only the part of history that has
been added since the most recent --rejoin.
If your split commits end up merged into the upstream
subproject, and then you want to get the latest upstream
version, this will allow git's merge algorithm to more
intelligently avoid conflicts (since it knows these
synthetic commits are already part of the upstream
repository).
Unfortunately, using this option results in 'git log'
showing an extra copy of every new commit that was
created (the original, and the synthetic one).
If you do all your merges with '--squash', don't use
'--rejoin' when you split, because you don't want the
subproject's history to be part of your project anyway.
EXAMPLES
--------
Let's use the repository for the git source code as an example.
First, get your own copy of the git.git repository:
$ git clone git://git.kernel.org/pub/scm/git/git.git test-git
$ cd test-git
gitweb (commit 1130ef3) was merged into git as of commit
0a8f4f0, after which it was no longer maintained separately.
But imagine it had been maintained separately, and we wanted to
extract git's changes to gitweb since that time, to share with
the upstream. You could do this:
$ git subtree split --prefix=gitweb --annotate='(split) ' \
0a8f4f0^.. --onto=1130ef3 --rejoin \
--branch gitweb-latest
$ gitk gitweb-latest
$ git push git@xxxxxxxxxx:whatever/gitweb gitweb-latest:master
(We use '0a8f4f0^..' because that means "all the changes from
0a8f4f0 to the current version, including 0a8f4f0 itself.")
If gitweb had originally been merged using 'git subtree add' (or
a previous split had already been done with --rejoin specified)
then you can do all your splits without having to remember any
weird commit ids:
$ git subtree split --prefix=gitweb --annotate='(split) ' --rejoin \
--branch gitweb-latest2
And you can merge changes back in from the upstream project just
as easily:
$ git subtree pull --prefix=gitweb \
git@xxxxxxxxxx:whatever/gitweb gitweb-latest:master
Or, using '--squash', you can actually rewind to an earlier
version of gitweb:
$ git subtree merge --prefix=gitweb --squash gitweb-latest~10
Then make some changes:
$ date >gitweb/myfile
$ git add gitweb/myfile
$ git commit -m 'created myfile'
And fast forward again:
$ git subtree merge --prefix=gitweb --squash gitweb-latest
And notice that your change is still intact:
$ ls -l gitweb/myfile
And you can split it out and look at your changes versus
the standard gitweb:
git log gitweb-latest..$(git subtree split --prefix=gitweb)
AUTHOR
------
Written by Avery Pennarun <apenwarr@xxxxxxxxx>
GIT
---
Part of the linkgit:git[1] suite
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html