If the original fetch attempt did not want tags, we should continue to
honor that when doing a retry fetch with depth set to None. This seems
to match the intent of the retry based on the inline comment and results
in a significant performance improvement when the original fetch-by-sha1
fails due to the server not allowing requests for unadvertised objects.
Change-Id: Ia26bb31ea9aecc4ba2d3e87fc0c5412472cd98c4
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/374918
Reviewed-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Kaushik Lingarkar <kaushik.lingarkar@linaro.org>
Tested-by: Kaushik Lingarkar <kaushik.lingarkar@linaro.org>
Apply rules set by https://gerrit-review.googlesource.com/c/git-repo/+/362954/ across the codebase and fix any lingering errors caught
by flake8. Also check black formatting in run_tests (and CQ).
Bug: b/267675342
Change-Id: I972d77649dac351150dcfeb1cd1ad0ea2efc1956
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/363474
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: Gavin Mak <gavinmak@google.com>
Commit-Queue: Gavin Mak <gavinmak@google.com>
Superproject update failures on single-manifest checkouts had an extra
space.
Bug: b/254523816
Change-Id: I6f71e42337e324a6975c5d6bba487f83abaf054f
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/357056
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Xin Li <delphij@google.com>
If this is a project that is not using object sharing (there is only one
copy of the remote project) then clear preciousObjects.
To override this for a project, run:
git config --replace-all repo.preservePreciousObjects true
Change-Id: If3ea061c631c5ecd44ead84f68576012e2c7405c
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/350235
Reviewed-by: Jonathan Nieder <jrn@google.com>
Tested-by: LaMont Jones <lamontjones@google.com>
If a file that is copied using a <copyfile> tag is modified and not
committed or if it is committed to a detached head, then running `repo
sync` would update the target file as expected. However, if the
modified file is committed to a local branch, then running `repo sync'
would not update the target file as expected.
Change-Id: Ic98e37d1c2e51fd1bf15abf149c7d06190cfd6d2
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/344475
Reviewed-by: Mike Frysinger <vapier@google.com>
Replace tuple returns with namedtuples, to simplify adding new fields.
Extend the Sync_NetworkHalf return value to:
- success: True if successful (the former return value)
- remote_fetched: True if we called `git fetch`
Change-Id: If63c24c2f849523f77fa19c05bbf23a5e9a20ba9
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/344534
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: LaMont Jones <lamontjones@google.com>
This reverts commit 07d21e6bde.
Reason for revert: crbug.com/gerrit/16230, b/244467766 - breaks aosp-master-with-phones case
Change-Id: Id967d92f8622c2c13356b09e46ece9f20040aabc
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/344314
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Xin Li <delphij@google.com>
If initializing the manifest fails for any reason, don't leave it in
a half complete state. This can cause problems if/when the user tries
to reinit because different codepaths will be taken. For example, if
we initialize manifests.git and don't finish probing the remote to see
what default branch it uses, we end up always using "master" even if
that isn't what the remote uses.
To avoid all of this, use .tmp dirs when initializing, and rename to
the final path only after we complete all the right steps.
We should roll this out to all projects we clone, but start with the
manifest project for now.
Bug: https://crbug.com/gerrit/13526
Bug: https://crbug.com/gerrit/15805
Change-Id: I0214338de69ee11e090285c6b0b211052804af06
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/343539
Reviewed-by: LaMont Jones <lamontjones@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Normally git produces no output when creating or switching branches.
If there's a problem though, we want to show that to the user. So
switch from capturing all output to running in quiet mode.
Bug: https://crbug.com/gerrit/15819
Change-Id: I7873ecc7c3bacce591899cc9471cb0244eb74541
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/343454
Reviewed-by: LaMont Jones <lamontjones@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
We almost always use self.remote.name when calling self.GetRemote,
so make that the default to simplify the code a bit.
Change-Id: Ifdf6e1370d6b8963b44e6d384b0fac8fa5c4f2ba
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/343184
Tested-by: Mike Frysinger <vapier@google.com>
Reviewed-by: LaMont Jones <lamontjones@google.com>
The `--wip` allow to bulk push changed as work-in-progress. This CL
intend to allow the opposite opperation by removing the wip mark on the
CL and set it to be ready to review
Change-Id: If0743c5b14829f77be2def5a8547060d06a5648c
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/342214
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: William Escande <wescande@google.com>
Display the project path relative to the outermost manifest by default,
and relative to the sub manifest only when --this-manifest-only is
specified.
For project-related diagnostic messages, use the outermost manifest for
messages.
Change-Id: I4537d7dd412a2c182e77d6720e95c1b0ef70eb0e
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/340754
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: LaMont Jones <lamontjones@google.com>
This option will suppress the
Uncommitted changes in ... (did you forget to amend?)
prompt when there are untracked (unknown) files in the working copy.
The prompt is still shown if tracked files are modified.
Change-Id: Ia3fcc82989b7fad09b69214eda31e2d0dfc14600
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/340456
Tested-by: Martin Geisler <mgeisler@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
If the user says `--no-use-superproject`, then do not bother syncing the
superproject.
Also add/update docstrings and comments throughout.
Change-Id: I9cdad706130501bab9a22d3099a1dae605e9c194
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/338975
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
With this change, partial syncs (sync with a project list) are again
supported.
If the updated manifest includes new sub manifests, download them
inheriting options from the parent manifestProject.
Change-Id: Id952f85df2e26d34e38b251973be26434443ff56
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/334819
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: LaMont Jones <lamontjones@google.com>
For now, this is opt-in via environment variables:
- export REPO_USE_ALTERNATES=1
The shared project logic that shares the internal .git/objects/ dir
directly between multiple projects via the project-objects/ tree has
a lot of KI with random corruption. It all boils down to projects
sharing objects/ but not refs/. Git operations that use refs to see
what objects are reachable and discard the rest can easily discard
objects that are used by other projects.
Consider this project layout:
<show fs layout>
There are unique refs in each of these trees that are not visible in
the others. This means it's not safe to run basic operations like
git prune or git gc.
Since we can't share refs (each project needs to have unique refs
like HEAD in order to function), let's change how we share objects.
The old way involved symlinking .git/objects/ to the project-objects
tree. The new way shares objects using git's info/alternates.
This means project-objects/ will only contain objects that exist in
the remote project. Local per-project objects (like when creating
branches and making changes) will never be shared. When running a
prune or gc operation in the per-project state, it will only ever
repack or discard those per-project objects. The common shared
objects would only be cleaned up when running a common operation
(i.e. by repo itself).
One downside to this for users is if they try blending unrelated
upstream projects. For example, in CrOS we have multiple kernel
projects (for diff versions) checked out. If a dev fetched the
upstream Linus tree into one of them, the objects & tags would
not be shared with the others, so they would have to fetch the
upstream state for each project. Annoying, but better than the
current corruption situation we're in now.
Also if the dev runs a manual `git fetch` in the per-project to
sync it up to newer state than the last `repo sync` they ran,
the objects would get duplicated. However, git operations later
on should eventually dedupe this.
Bug: https://crbug.com/gerrit/15553
Change-Id: I313a9b8962f9d439ef98ac0ed37ecfb9e0b3864e
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/328101
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: LaMont Jones <lamontjones@google.com>
This way we know we don't need to encode the labels.
Change-Id: Ib83ed8f4ed05f00b9d2d06a9dd3f304e4443430e
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/337518
Tested-by: Mike Frysinger <vapier@google.com>
Reviewed-by: LaMont Jones <lamontjones@google.com>
If init was run with --use-superproject, init failed.
If init was run without --{no,}use-superproject option then manifests
with <superproject/> elements were mishandled.
Bug: b/233226285
Test: manual
Change-Id: I737e71c89d2d7c324114f58bf2dc82b40e5beba7
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/337534
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: LaMont Jones <lamontjones@google.com>
When the user does not specify any manifest groups, this allows the
parent manifest to indicate which manifest groups should be used for
syncing the submanifest.
Change-Id: I88806ed35013d13dd2ab3cd245fcd4f9061112c4
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/335474
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
When a manifest file is overridden, remember that and keep using the
override for the remainder of the process. If we need to revert it,
make the override name evaluate False.
Change-Id: I1eee05fec6988c1ee4a3c751c4b540d5b5d11797
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/335136
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
And fix most of the other attributes to return the value instead of
None.
Change-Id: Iddcbbeb56238ee082bb1cae30adbd27a2f551f3d
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/335134
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Reviewed-by: Xin Li <delphij@google.com>
Always create Superproject when there is a <superproject> tag, and have
it hold the XML content, similar to how other manifest elements are
handled.
This also adds SetQuiet and SetPrintMessages to Superproject
consistent with manifest.SetUseLocalManifests.
Change-Id: I522bf3da542006575799f0640c67f7052704f266
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/334641
Reviewed-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: LaMont Jones <lamontjones@google.com>
- do not call the internal method from subcmds/sync.py.
- use the correct default groups for submanifests.
- only sync the superproject when we are told to.
Change-Id: I81e4025058f1ee564732b9e17aecc522f6b5f626
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/334639
Reviewed-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: LaMont Jones <lamontjones@google.com>
Also fixes some typos
Change-Id: Id2ba5834ba3a74ed3f29c36d2c0030737dc63e35
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/334579
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
This moves more of the manifest project handling into ManifestProject.
Change-Id: Iecdafbec18cccdfd8e625753c3bd1bcddf2b227f
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/334520
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
Use ManifestProject properties for config values.
Change-Id: Ib4ad90b0d9a089916e35615b8058942e6d01dc04
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/334519
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
Move the logic to sync a ManifestProject out of subcmds/init.py
Change-Id: Ia9d00f3da1dc3c5dada84c4d19cf9802c2346cb0
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/334140
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
Create RepoProject and ManifestProject, inheriting from MetaProject,
with methods separated for isolation and clarity.
Change-Id: Ic1d6efc65c99470290fea612e2abaf8670d199f4
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/334139
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
To be addressed in another change:
- a partial `repo sync` (with a list of projects/paths to sync)
requires `--this-tree-only`.
Change-Id: I6c7400bf001540e9d7694fa70934f8f204cb5f57
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/322657
Tested-by: LaMont Jones <lamontjones@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
If submodules is False, explicitly pass '=no'. Uninitialized submodules
may cause the default option to fail.
Change-Id: Ia00bcba5b69c4b65195f4c469c686a3ef9a4a3ad
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/330159
Reviewed-by: Xin Li <delphij@google.com>
Tested-by: LaMont Jones <lamontjones@google.com>
This is just a log file that, while useful for humans when gc aborts,
doesn't contain any data, so it's safe to throw away.
Bug: https://crbug.com/gerrit/15619
Change-Id: Ia95e0e281f52260668f7a80b5d5f990e32a8597a
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/328999
Tested-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
It was reported that git-lfs did not work with git-repo. Specifically,
`git read-tree -u` run by `repo sync` would fail git-lfs's smudge
filter. See https://github.com/github/git-lfs/issues/1422.
In fact, by the time `git read-tree -u` is run, the repository is not
bare. It is just that, the working directory is not the same as the
.git directory. git-lfs's filter should work. No one seems to have
delved into that issue.
Today, with newer versions of git-repo and git-lfs, that issue will
not reproduce. Tested with
- git 2.33, git-lfs 2.13 on macOS
- git 2.17, git-lfs 2.3 on ubuntu
So, it seems fine to add an option --enable-git-lfs-filter, default to
false, and stat that it may not work with older versions of git and
git-lfs in the help doc.
Bug: https://crbug.com/gerrit/14516
Change-Id: I8d21854eeeea541e072f63d6b10ad1253b1a9826
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/328359
Tested-by: XD Trol <milestonejxd@gmail.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
Removing the sample hooks is just clean up, so if repo cannot remove a
sample hook that should not cause it to fail.
Change-Id: I716b977da091c22b8f53e134f4fbc114116f9a65
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/328635
Reviewed-by: Mike Frysinger <vapier@google.com>
In order to stop sharing objects/ directly between shared projects,
we have to fetch the remote objects into project-objects/ manually.
So instead of running git operations in the individual project dirs
and relying on .git/objects being symlinked to project-objects/,
tell git to store any objects it fetches in project-objects/.
We do this by leveraging the GIT_OBJECT_DIRECTORY override. This
has been in git forever, or at least since v1.7.2 which is what we
already hard require. This tells git to save new objects to the
specified path no matter where it's being run otherwise.
We still otherwise run git in the project-specific dir so that it
can find the right set of refs that it wants to compare against,
including local refs. For that reason, we also have to leverage
GIT_ALTERNATE_OBJECT_DIRECTORIES to tell git where to find objects
that are not in the upstream remote. This way git doesn't blow up
when it can't find objects only associated with local commits.
As it stands right now, the practical result is the same: since we
symlink the project objects/ dir to the project-objects/ tree, the
default objects dir, the one we set $GIT_OBJECT_DIRECTORY to, and
the one we set $GIT_ALTERNATE_OBJECT_DIRECTORIES to are actually
all the same. So this commit by itself should be safe. But in a
follow up commit, we can replace the symlink with a separate dir
and git will keep working.
Bug: https://crbug.com/gerrit/15553
Change-Id: Ie4e654aec3e1ee307eee925a54908a2db6a5869f
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/328100
Reviewed-by: Jack Neus <jackneus@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
When using --reference, the path is written to objects/info/alternates.
The path is accessed inconsistently -- sometimes through projects/ (via
self.gitdir) and sometimes through project-objects/ (via self.objdir).
This works because projects/.../objects is a symlink to the objects dir
under project-objects/. Change all accesses to go through self.objdir.
This will allow us to stop symlinking projects/.../objects without the
reference dir logic breaking. The projects/ path is going to use its
alternates file for its own needs.
Bug: https://crbug.com/gerrit/15553
Change-Id: I6b452ad1aaffec74ecb7ac1bb9baa3a3a52e076c
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/328099
Tested-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Jack Neus <jackneus@google.com>
These hooks are never used and often get stale, so just trim them.
Users rarely look in these dirs to begin with.
Change-Id: Ic785aa55fb7ec84a61376df101127d0018882030
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/327538
Reviewed-by: Jack Neus <jackneus@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Since we don't do this anymore, and there prob won't be a need to
bring it back, drop support for it.
Bug: https://crbug.com/gerrit/15460
Change-Id: I7d86706f108c797a5c7962cb1578693d49430367
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/327537
Reviewed-by: Jack Neus <jackneus@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Verify all the .git/ paths will be handled by the migration logic before
starting the migration. This way we still abort & log an error, but the
user gets to see it before we put the tree into a state that they have to
manually recover. Also add a few more known-safe-to-clobber paths.
Bug: https://crbug.com/gerrit/15273
Change-Id: If49d69b341bc960ddcafa30da333fb5ec7145b51
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/327557
Reviewed-by: Colin Cross <ccross@android.com>
Tested-by: Mike Frysinger <vapier@google.com>
Unsharing this directory shouldn't be a problem. The current repo code
treated it as a file, and while that's actually incorrect, files & dirs
are basically treated the same, so it's practically the same.
Let's enumerate each subpath since there aren't that many.
info/refs:
Only used when the project is exported over git dumb transports (i.e.
a http:// server). Repo never does this, and it's extremely unlikely
any user has ever done this. Plus, this proposal talks about unsharing
project refs, so this file should get unshared too.
info/grafts:
A user-configurable file that repo never touches. Might be useful to
share across projects, but probably rarely (if ever) used by developers,
and forcing them to configure it for each project isn't that big of a
deal.
info/exclude:
info/attributes:
User-configurable files that repo never touches. Doesn't seem like
most users ever touch these, and if they do, having them do it for
each shared project isn't a big deal.
info/sparse-checkout:
Repo doesn't use sparse checkouts, and it's extremely unlikely to even
work if a user tried doing something themselves.
Bug: https://crbug.com/gerrit/15460
Change-Id: I53e44d73a6d7a92da615b46600d8ea51cb46e3ac
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/327519
Reviewed-by: Jonathan Nieder <jrn@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Nothing uses this path. It’s only for exporting git dirs e.g. for
online gitweb use which probably no one does. It is not the same
description file as exists on servers we cloned from. Leaving it
as the default plain text file will simplify code.
We don't undo any existing symlinks if they exist since repo does
not care about them, and their existence doesn't hurt.
Bug: https://crbug.com/gerrit/15460
Change-Id: Ic34fe7c3cfb8f6da844de5be30158f59382b1cc8
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/327518
Reviewed-by: Jonathan Nieder <jrn@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
This path only matters to users of `git svn` who manually run it in
local projects after they get a full repo client checkout. With svn
usage falling in general, and with the fact that the source checkout
now symlinks its .git/ state to the internal projects/ path, we don't
need to manage this anymore.
It means the path won't be shared among multiple local projects that
have the same remote, but so it goes. It was an optimization only,
not functionality required for correctness. We want to simplify the
internals to stop messing with git state, and this particular path
doesn't seem worth the effort to maintain.
We don't undo any existing svn symlinks if they exist since repo does
not care about them, and their existence doesn't hurt anything.
Bug: https://crbug.com/gerrit/15460
Change-Id: Ie8496b275bcc589771aa9f4ee874ed2ee6d5241d
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/327517
Reviewed-by: Jonathan Nieder <jrn@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
