Revert 186598

Revert 186598 ("Add option to automatically fetch git "branch-heads".")

Some SVN-only trybots get errors when running gclient revert with this
change. I assume it's because the flag is not present for the gclient
revert command, and that the code checking if options.with_branch_heads
is set is common to both revert and sync.

BUG=none
TEST=reverted locally and gclient revert worked again.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=186967

[kjellander_chromium, 2013-03-08 15:13:33]

This change broke some of our WebRTC try bots. 
These bots have two solutions in the .gclient file, if that helps.

Here's an example trace from the 'gclient revert --nohooks' command:

Traceback (most recent call last):
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient.py", line
1765, in <module>
    sys.exit(Main(sys.argv[1:]))
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient.py", line
1754, in Main
    return command(parser, argv[1:])
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient.py", line
1569, in CMDrevert
    return client.RunOnDeps('revert', args)
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient.py", line
1112, in RunOnDeps
    work_queue.flush(revision_overrides, command, args, options=self._options)
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient_utils.py",
line 679, in run
    self.item.run(*self.args, **self.kwargs)
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient.py", line
604, in run
    scm.RunCommand(command, options, args, file_list)
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient_scm.py",
line 142, in RunCommand
    return getattr(self, command)(options, args, file_list)
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient_scm.py",
line 532, in revert
    return self.update(options, [], file_list)
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient_scm.py",
line 279, in update
    self._Clone(revision, url, options)
  File "/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient_scm.py",
line 658, in _Clone
    if options.with_branch_heads:
AttributeError: Values instance has no attribute 'with_branch_heads'
Sending crash report ...
  args: ['/usr/local/google/buildbot/webrtc-cb/cb/depot_tools/gclient.py',
'revert', '--nohooks']
  cwd: /usr/local/google/buildbot/webrtc-cb/cb/build/slav
  exception: Values instance has no attribute 'with_branch_head
  host: xxx
  stack:   File "/usr/local/google/buildbot/webrtc-cb/cb/de
  user: webrtc-cb
  version: 2.6.5 (r265:79063, Oct  1 2012, 22:04:36) 
[GCC 4.
A stack trace has been sent to the maintainers.

[M-A Ruel, 2013-03-08 15:16:55]

lgtm

[commit-bot: I haz the power, 2013-03-08 15:17:00]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/kjellander@chromium.org/12616006/1

[commit-bot: I haz the power, 2013-03-08 15:19:40]

Change committed as 186967

[Michael Moss, 2013-03-08 16:30:17]

On 2013/03/08 15:19:40, I haz the power (commit-bot) wrote:
> Change committed as 186967

Is anyone concerned that 'revert' flows through update() like this? Can we
really claim that it's properly reverting local modifications if it's
potentially updating/cloning projects without using the parameters that were
originally passed to 'sync'? Missing flags like --revision or --head could cause
revert to checkout wildly different sources than originally specified.

[Michael Moss, 2013-03-08 16:44:32]

In fact, it seems like a better "nuclear option to get back to a 'clean'
state" for git would be to run the normal 'sync' command, but with the
"--reset --force --delete_unversioned_trees" flags. That's actually what
the official builders do. Maybe 'gclient revert' needs to be svn-only?


On Fri, Mar 8, 2013 at 8:30 AM, <mmoss@chromium.org> wrote:

> On 2013/03/08 15:19:40, I haz the power (commit-bot) wrote:
>
>> Change committed as 186967
>>
>
> Is anyone concerned that 'revert' flows through update() like this? Can we
> really claim that it's properly reverting local modifications if it's
> potentially updating/cloning projects without using the parameters that
> were
> originally passed to 'sync'? Missing flags like --revision or --head could
> cause
> revert to checkout wildly different sources than originally specified.
>
>
https://codereview.chromium.**org/12616006/<https://codereview.chromium.org/1...
>

[M-A Ruel, 2013-03-08 16:51:48]

I wasn't happy with --reset being added either. I'm fine with replacing
gclient revert with gclient sync --reset.

M-A


2013/3/8 Michael Moss <mmoss@chromium.org>

> In fact, it seems like a better "nuclear option to get back to a 'clean'
> state" for git would be to run the normal 'sync' command, but with the
> "--reset --force --delete_unversioned_trees" flags. That's actually what
> the official builders do. Maybe 'gclient revert' needs to be svn-only?
>
>
> On Fri, Mar 8, 2013 at 8:30 AM, <mmoss@chromium.org> wrote:
>
>> On 2013/03/08 15:19:40, I haz the power (commit-bot) wrote:
>>
>>> Change committed as 186967
>>>
>>
>> Is anyone concerned that 'revert' flows through update() like this? Can we
>> really claim that it's properly reverting local modifications if it's
>> potentially updating/cloning projects without using the parameters that
>> were
>> originally passed to 'sync'? Missing flags like --revision or --head
>> could cause
>> revert to checkout wildly different sources than originally specified.
>>
>>
https://codereview.chromium.**org/12616006/<https://codereview.chromium.org/1...
>>
>
>

[iannucci1, 2013-03-08 19:44:27]

Don't forget about git clean -xf :-)


On Fri, Mar 8, 2013 at 8:51 AM, Marc-Antoine Ruel <maruel@chromium.org>wrote:

> I wasn't happy with --reset being added either. I'm fine with replacing
> gclient revert with gclient sync --reset.
>
> M-A
>
>
>
> 2013/3/8 Michael Moss <mmoss@chromium.org>
>
>> In fact, it seems like a better "nuclear option to get back to a 'clean'
>> state" for git would be to run the normal 'sync' command, but with the
>> "--reset --force --delete_unversioned_trees" flags. That's actually what
>> the official builders do. Maybe 'gclient revert' needs to be svn-only?
>>
>>
>> On Fri, Mar 8, 2013 at 8:30 AM, <mmoss@chromium.org> wrote:
>>
>>> On 2013/03/08 15:19:40, I haz the power (commit-bot) wrote:
>>>
>>>> Change committed as 186967
>>>>
>>>
>>> Is anyone concerned that 'revert' flows through update() like this? Can
>>> we
>>> really claim that it's properly reverting local modifications if it's
>>> potentially updating/cloning projects without using the parameters that
>>> were
>>> originally passed to 'sync'? Missing flags like --revision or --head
>>> could cause
>>> revert to checkout wildly different sources than originally specified.
>>>
>>>
https://codereview.chromium.**org/12616006/<https://codereview.chromium.org/1...
>>>
>>
>>
>