Running repo with Python 3.5 fails due to the use of the encoding
parameter to subprocess.run(). There are also f-strings being used in
some of the tests.
This drops support for these systems:
* Ubuntu Xenial: released Apr 2016, EOS Apr 2021, EOL Apr 2024
* Debian Stretch: released Jun 2017, EOL Jun 2022
So the minimum required distros now are:
* Ubuntu Bionic: released Apr 2018 w/Python 3.6
* Debian Buster: released Jul 2019 w/Python 3.7
Change-Id: I1144f7ab6f882b10cac0131982df081fe4ac44f9
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303363
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Before Python 3.8, xml.dom.minidom sorted the attributes of an element
when writing it to a file, while later versions output the attributes
in the order they were created. Avoid these differences by sorting the
attributes for each element before comparing the generated manifests
with the expected ones.
Bug: https://crbug.com/gerrit/14382
Change-Id: Ie2597727afcc48f9063a7261ad970e8a549f0587
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303326
Reviewed-by: Mike Frysinger <vapier@google.com>
Commit b4429439 (sync: refactor main fetch loop) broke syncing of
submodules with sync-s="true". The first `repo sync` would just fetch
the superrepo, while the second `repo sync` would fetch the submodules.
This was due to the new _FetchMain() function not passing back the
all_projects variable that it had modified.
Change-Id: Ie8187cde7bb894e4e9a6b76c2aed83873d9f69a4
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/307065
Reviewed-by: Mike Frysinger <vapier@google.com>
We always pass in output_redir when syncing, but that's the common case:
there are a few situations (like `repo init`) where we don't pass in a
buffer, and if any errors show up in that case, we'd crash. Rely on the
print function to handle this logic for us.
Bug: https://crbug.com/gerrit/14568
Change-Id: I8cd47e82329797ffc42534418a3dfbd8429205be
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/307222
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
+ In XmlManifest._Unload set 'bugurl' to Wrapper().BUG_URL.
+ contactinfo returns a namedtuple.
+ bug_url can be accessed as self._manifest.contactinfo.bugurl.
Tested the code with the following commands.
$ ./run_tests -v
Added contactinfo tag to default.xml and verified that bugurl is used.
Bug: [google internal] b/186220520.
Change-Id: Iaafd6465e072b2e47a0a0b548bf6cb608a0b0a04
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/306342
Tested-by: Raman Tenneti <rtenneti@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
We changed sync to use multiprocessing for parallel work. This broke
the ssh proxy code as it's all based on threads. Rewrite the logic to
be multiprocessing safe.
Now instead of the module acting as a stateful object, callers have to
instantiate a new ProxyManager class that holds all the state, an pass
that down to any users.
Bug: https://crbug.com/gerrit/12389
Change-Id: I4b1af116f7306b91e825d3c56fb4274c9b033562
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305486
Tested-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Chris Mcdonald <cjmcdonald@google.com>
The only time we really need ssh proxies is when we want to run many
connections and reuse them. That only happens when running sync.
Every other command makes at most two connections, and even then it's
only one or none. So the effort of setting up & tearing down ssh
proxies isn't worth it most of the time.
The big reason we want to move this logic to sync is that it's now
using multiprocessing for parallel work. The current ssh proxy code
is all based on threads, which means none of the logic is working
correctly. The current ssh design makes it hard to fix when all of
the state lives in the global/module scope.
So the first step to fixing this is top move the setup & teardown to
the one place that really needs it: sync. No other commands will use
proxies anymore, just direct connections.
Bug: https://crbug.com/gerrit/12389
Change-Id: Ibd351acdec39a87562b3013637c5df4ea34e03c6
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305485
Reviewed-by: Chris Mcdonald <cjmcdonald@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Gitlab returns a 400 error when trying to fetch clone.bundle
from a repository containing the git-repo tool. The repo
launcher doesn't then fall back to not using a clone.bundle
file and the repo init fails.
Change-Id: Ia3390d0638ef9a39fb2fab84625b269d28caf1cc
Signed-off-by: Craig Northway <cnorthway@codeaurora.org>
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305382
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested the code with the following commands.
$ ./run_tests -v
$ repo_dev sync -c -j8 --no-use-superproject
Fetching: 100% (1041/1041), done in 1m22.743s
$ repo_dev sync -c -j8 --use-superproject
WARNING: --use-superproject is experimental and not for general use
..
Bug: [google internal] b/187459275
Change-Id: I3f4269df38cd24a21723e8b2be5a1f013e7b5a91
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305682
Tested-by: Raman Tenneti <rtenneti@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
If superproject was not available for a branch, then the next
repo sync would also fail because --use-superproject is
remembered across repo init. In such cases, hoping the hint to
to use --no-use-superproject will help.
Tested the code with the following commands and by forcing
a failure.
$ ./run_tests -v
Bug: [google internal] b/187459275
Change-Id: Ie250812b7ba83afc230b5b1d154ba11f245f8b8a
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305622
Reviewed-by: Xin Li <delphij@google.com>
Tested-by: Raman Tenneti <rtenneti@google.com>
This is a large chunk of code that is largely isolated. Move it into
a class method to make it easier to manage & reason about, and in a
follow up CL, easier to scope.
Bug: https://crbug.com/gerrit/12389
Change-Id: I0c69d95a9e03478d347b761580b2343bffa012d5
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305484
Tested-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Chris Mcdonald <cjmcdonald@google.com>
We had ssh logic sprinkled between two git modules, and neither was
quite the right home for it. This largely moves the logic as-is to
its new home. We'll leave major refactoring to followup commits.
Bug: https://crbug.com/gerrit/12389
Change-Id: I300a8f7dba74f2bd132232a5eb1e856a8490e0e9
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305483
Reviewed-by: Chris Mcdonald <cjmcdonald@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Simplifies the code a bit to use the stdlib cache helper.
Change-Id: I778e90100ce748a71cc3a5a5d67dda403334315e
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305482
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Commit b2fa30a2b8 ("sync: switch network
fetch to multiprocessing") accidentally changed the variable passed to
the 2nd fetch call from |missing| to |to_fetch| due to a copy & paste
of the earlier changed logic. Undo that to fix git submodule fetching.
Bug: https://crbug.com/gerrit/14489
Change-Id: I627954f80fd2e80d9d5809b530aa6b0ef9260abb
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305262
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
The ssh master logic has never worked under Windows which is why this
code always returned False when running there (including cygwin). But
the OS check was still done while holding the threading lock. While
it might be a little slower than necessary, it still worked.
The switch from the threading module to the multiprocessing module
changed global behavior subtly under Windows and broke things: the
globals previously would stay valid, but now they get cleared. So
the lock is reset to None in children workers.
We could tweak the logic to pass the lock through, but there isn't
much point when the rest of the code is still disabled in Windows.
So perform the platform check before we grab the lock. This fixes
the crash, and probably speeds things up a few nanoseconds.
This shouldn't be a problem on Linux systems as the platform fork
will duplicate the existing process memory (including globals).
Bug: https://crbug.com/gerrit/14480
Change-Id: I1d1da82c6d7bd6b8cdc1f03f640a520ecd047063
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305149
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
It will be used to let manifest authors self-register contact info.
This element can be repeated, and any later entries will clobber
earlier ones. This would allow manifest authors who extend
manifests to specify their own contact info.
It would have 1 required attribute: bugurl.
"bugurl" specifies the URL to file a bug against the manifest owner.
<contactinfo bugurl="bug-url"/>
TODO: This CL only implements the parsing logic and further work
will be in followup CLs.
Tested the code with the following commands.
$ ./run_tests tests/test_manifest_xml.py
$ ./run_tests -v
Bug: [google internal] b/186220520.
Change-Id: I47e765ba2dab5cdf850191129f4d4cd6b803f451
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305203
Tested-by: Raman Tenneti <rtenneti@google.com>
Reviewed-by: Mike Frysinger <vapier@google.com>
Commit fb527e3f52 ("sync: create dedicated
manifest project update func") refactored code from the main body into a
dedicated method. The manifest_name was passed as an argument, but never
used it, and instead reaches back out to the command line options. This
ignores the logic in the main loop where manifest_name might have changed
(like when using smart sync).
Change-Id: I4b84638fbb10c2b6f8f4b555e1475b0669c2daf4
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305148
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
We've switched most of this command over to multiprocessing and off
of _threading, so do the Event object too. The APIs are the same
between the modules, so we shouldn't need to update anything else.
Change-Id: I52d31f1c6ef2bcbe7bbc1dd1add79a8d5d08784a
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305147
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
We check for updates only once per day, so clarify the docs.
Change-Id: Ib669ca6ebc67bc13204996fa40e1a3a82012295e
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305145
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
We're inconsistent with help text as to whether it uses title case and
whether it ends in a period. Add a test to enforce a standard, and use
the style that Python optparse & argparse use themselves (e.g. with the
--help option): always lowercase, and never trailing period.
Change-Id: Ic1defae23daeac0ac9116aaf487427f50b34050d
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305144
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
When converting this logic from print() to the output buffer, this
error codepath should have dropped the use of the file= redirect.
Bug: https://crbug.com/gerrit/14482
Change-Id: Ib484924a2031ba3295c1c1a5b9a2d816b9912279
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/305142
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
While this provides a way to undo earlier command line options (e.g.
`repo sync --tags --no-tags`) which can be helpful for scripting &
automation, this more importantly allows the user to override the
manifest settings for syncing tags from a project.
Bug: https://crbug.com/gerrit/12401
Change-Id: Id4c36cd82e6ca7cb073b5d63a09f6c7ccdebba83
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/304904
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
For most commands, this is more about providing a way to undo earlier
command line options (e.g. `repo info -c --no-current-branch`) which
can be helpful for scripting & automation. But for the sync command,
this is helpful to undo the setting that exists in the manifest itself.
With this in place, tweak the sync current_branch_only logic to only
apply the manifest settings when the user hasn't specified a command
line option.
Bug: https://crbug.com/gerrit/12401
Change-Id: I21e2384624680cc740d1b5d1e49c50589d2fe6a0
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/304903
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
We're inconsistent with the short option for this flag:
* gitc-init: <none as -c is already used>
* info: -b
* init: -c
* overview: -b
* sync: -c
* upload: --cbr
Since info & overview are not as heavily used as the others, switch
them from -b to -c. We leave -b in as a hidden alias for now.
Similarly, switch upload from --cbr to just -c. A lot of people
use --cbr, so we leave this as a hidden alias for now too.
Ideally gitc-init wouldn't use -c, but that ship has sailed, and
we're more likely to deprecate gitc entirely at this point.
This provides a consistent set of options across subcommands.
Bug: https://crbug.com/gerrit/12401
Change-Id: Iec249729223866fe1ea0ebabed12ca851cc38b35
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/304902
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Search for project branches to upload in parallel. This can cut the
lookup time in half for large projects. We still run the actual hooks
in serial once we have the list of projects to process, but we would
need to rethink things quite a bit before we could handle running them
in parallel too.
Change-Id: I8da0cbc5010566aa860e1a158f3dc07f0709dcff
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/304842
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
We were updating the per-checkout m/ pseudo ref when syncing, but we
only created the common m/ redirect when initializing a project for
the first time. This is fine unless the user switches the manifest
branch in an existing project, then we never create that redirect.
Bug: https://crbug.com/gerrit/14468
Change-Id: I5325e7e602dcb4ce150bef258901ba5e9fdea461
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/304822
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
When running upload across multiple projects, include the project in
any error messages that come up. This lets users figure out where
the problem might be.
Change-Id: I09470c9a1b512baf910d6d97b747816d1a6f3a87
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/304783
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Currently, copyfiles and linkfiles which marked by
"<copyfile/>" and "<linkfile/>" in manifest will
be created by first exec 'repo sync'.
But if some "<copyfile/>" or "<linkfile/>" are removed
in manifest, then 'repo sync', these removed item
dest can not be removed in the sourcecode workspace.
This patch is intent to fix this issue, by save a
'copy-link-files.json' in .repo and then compared with
new dest path when next sync. If any "<copyfile/>" or
"<linkfile/>" were removed, the dest path will be
removed in sourcecode at the same time.
Bug: https://crbug.com/gerrit/11008
Change-Id: I6b7b41e94df0f9e6e52801ec755951a4c572d05d
Signed-off-by: jiajia tang <tangjiajia@xiaomi.com>
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/304202
Reviewed-by: Mike Frysinger <vapier@google.com>
When depth is used, we would fetch only SHA1 when superproject is
used, as the result, only the manifest branch is being recorded,
and commands like repo start would fail.
Fix this by saving the upstream branch value in the overlay
manifest and add the upstream branch to fetch list.
Bug: [google internal] b/185951360
Change-Id: Ib36f56067723f2572ed817785b31cc928ddfec0a
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/304562
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Reviewed-by: Jonathan Nieder <jrn@google.com>
Tested-by: Xin Li <delphij@google.com>
Make it a bit easier for people to locate bug reporting info.
Change-Id: If9c8939c84ebd52eb96b353c1797afa25868bb85
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303943
Tested-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Default the partial_clone_exclude argument to an empty set.
Fixes the following report by Emil Medve.
With this change (up to v2.14.1), on an existing "normal" clone (without partial-clone options) I'm seeing this traceback during `repo selfupdate`:
Traceback (most recent call last):
File ".../.repo/repo/main.py", line 630, in <module>
_Main(sys.argv[1:])
File ".../.repo/repo/main.py", line 604, in _Main
result = run()
File ".../.repo/repo/main.py", line 597, in <lambda>
run = lambda: repo._Run(name, gopts, argv) or 0
File ".../.repo/repo/main.py", line 261, in _Run
result = cmd.Execute(copts, cargs)
File ".../.repo/repo/subcmds/selfupdate.py", line 54, in Execute
if not rp.Sync_NetworkHalf():
File ".../.repo/repo/project.py", line 1091, in Sync_NetworkHalf
if self.name in partial_clone_exclude:
TypeError: argument of type 'NoneType' is not iterable
$ ./run_tests -v
Change-Id: I71e744e4ef2a37b13aa9ba42eba3935e78c4e40a
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/304082
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: Raman Tenneti <rtenneti@google.com>
Some refactors during review dropped this import when it was reworked,
but it's still needed when using the --quiet setting.
Change-Id: I6d9302ef5a056e52415ea63f35bad592b9dfa75d
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303942
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
The merge of the repo & init parser missed this default.
When running `repo init ...` in an existing checkout but w/out the -m
option, then repo would error out complaining that -m is required when
it didn't do this before.
Change-Id: I58035d48cc413b5d373702b9dc3b9ecd3fd1e900
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303945
Tested-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Jonathan Nieder <jrn@google.com>
Make sure we print a message whenever we retry so it's clear to the
user why repo is pausing for a long time, and why repo might have
passed even though it displayed some errors earlier.
Also unify the sleep logic so we don't have two independent methods.
This makes it easier to reason about.
Also don't sleep if we're in the last iteration of the for loop. It
doesn't make sense to and needlessly slows things down when there are
real errors.
Bug: https://crbug.com/gerrit/12494
Change-Id: Ifceace5b2dde75c2dac39ea5388527dd37376336
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303402
Reviewed-by: Sam Saccone 🐐 <samccone@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
In _FetchOne & _CheckOne, only print error.GitError exception,
but other exceptions are still thrown
Fixes the GitError exceptions from /usr/lib/python3.8/multiprocessing/pool.py
exiting the repo sync.
Tested the code with the following commands and verified repo sync
continues after fetch error because of an invalid SHA1.
$ ./run_tests -v
$ python3 ~/work/repo/git-repo/repo sync -m manifest_P21623846.xml -j32
...
error.GitError: Cannot fetch platform/vendor/google_devices/redbull/proprietary update-ref: fatal: d5a99e518f09d6abb0c0dfa899594e1ea6232459^0: not a valid SHA1
....
An error like the following when jobs=1
error.GitError: Cannot checkout platform/vendor/qcom/sdm845/proprietary/qcrilOemHook: Cannot initialize work tree for platform/vendor/qcom/sdm845/proprietary/qcrilOemHook
Bug: https://crbug.com/gerrit/14392
Change-Id: I8922ad6c07c733125419f5698b0f7e32d70c7905
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303544
Reviewed-by: Mike Frysinger <vapier@google.com>
Tested-by: Raman Tenneti <rtenneti@google.com>
Now that we have a bunch of subcommands doing parallel execution, a
common pattern arises that we can factor out for most of them. We
leave forall alone as it's a bit too complicated atm to cut over.
Change-Id: I3617a4f7c66142bcd1ab030cb4cca698a65010ac
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/301942
Tested-by: Mike Frysinger <vapier@google.com>
Reviewed-by: Chris Mcdonald <cjmcdonald@google.com>
Since Python has deprecated the formatter module, switch to the textwrap
module instead for reflowing text. We weren't really using any other
feature anyways.
Verified by diffing the output before & after the change and making sure
it was the same.
Then made a few tweaks to tighten up the output.
Change-Id: I0be1bc2a6661a311b1a4693c80d0f8366320ba55
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303282
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Currently, list only shows projects that exist in the checkout, and
doesn't offer any way to list all projects in the manifest (based on
the current settings, or on the options passed to list). This seems
to be the opposite of what (at least some) users expect, so let's
add an option to show all of them regardless of checkout state.
Change-Id: I94bbdc5bd0ff2a411704fa215e7fc2b60fa3360e
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/301263
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
We want progress bars in the default output mode, but not when the
user specifies --quiet. Add a setting to the Progress bar class so
it takes care of not displaying anything itself rather than having
to update every subcommand to conditionally setup & call the object.
Change-Id: I1134993bffc5437bc22e26be11a512125f10597f
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303225
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
Add new CommonOptions entry points to move the existing --jobs to,
and relocate all --verbose/--quiet options to that. This provides
both a consistent interface for users as well as for code.
Change-Id: Ifaf83b88872421f4749b073c472b4a67ca6c0437
Reviewed-on: https://gerrit-review.googlesource.com/c/git-repo/+/303224
Reviewed-by: Raman Tenneti <rtenneti@google.com>
Tested-by: Mike Frysinger <vapier@google.com>
