Fix gclient to correctly check out remote refs during update.

Fix gclient to correctly check out remote refs during update.

Currently, any bot run with "--revision=project@origin/master" fails, because
it fetches the remote ref into refs/remotes/origin/master, but then checks out
refs/heads/master, which hasn't been rolled forward. This is very similar to
the bug fixed by https://codereview.chromium.org/85473007

However, I am very wary of this patch, because I am worried about changing
something commented with " For compatibility...". Marc-Antoine, please comment.

R=maruel@chromium.org, smut@chromium.org, stip@chromium.org

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=266251

[ghost stip (do not use), 2014-04-21 21:26:57]

https://chromiumcodereview.appspot.com/244253007/diff/1/gclient_scm.py
File gclient_scm.py (right):

https://chromiumcodereview.appspot.com/244253007/diff/1/gclient_scm.py#newcod...
gclient_scm.py:344: # For compatibility with old naming, translate 'origin' to
'refs/heads'
you should change this comment to reflect the new logic

[smut, 2014-04-21 21:28:43]

lgtm

https://codereview.chromium.org/244253007/diff/1/gclient_scm.py
File gclient_scm.py (right):

https://codereview.chromium.org/244253007/diff/1/gclient_scm.py#newcode344
gclient_scm.py:344: # For compatibility with old naming, translate 'origin' to
'refs/heads'
This comment is no longer applicable.

[agable, 2014-04-21 21:31:22]

https://chromiumcodereview.appspot.com/244253007/diff/1/gclient_scm.py
File gclient_scm.py (right):

https://chromiumcodereview.appspot.com/244253007/diff/1/gclient_scm.py#newcod...
gclient_scm.py:344: # For compatibility with old naming, translate 'origin' to
'refs/heads'
On 2014/04/21 21:26:57, stip wrote:
> you should change this comment to reflect the new logic

On 2014/04/21 21:28:43, smut wrote:
> This comment is no longer applicable.

Done.

[Marc-Antoine Ruel (Google), 2014-04-22 13:24:07]

I miss background details here. Why is specifying a remote ref useful for
--revision?

[pgervais, 2014-04-23 00:10:46]

On 2014/04/22 13:24:07, Marc-Antoine Ruel (Google) wrote:
> I miss background details here. Why is specifying a remote ref useful for
> --revision?

This is how some trybots use it, at least.

[M-A Ruel, 2014-04-23 00:24:27]

On 2014/04/23 00:10:46, pgervais wrote:
> On 2014/04/22 13:24:07, Marc-Antoine Ruel (Google) wrote:
> > I miss background details here. Why is specifying a remote ref useful for
> > --revision?
> 
> This is how some trybots use it, at least.

Ok but that's definitely the code of change that should have a proper test.
That's easy for this to fail silently.

[agable, 2014-04-23 19:52:33]

On 2014/04/23 00:24:27, M-A Ruel wrote:
> On 2014/04/23 00:10:46, pgervais wrote:
> > On 2014/04/22 13:24:07, Marc-Antoine Ruel (Google) wrote:
> > > I miss background details here. Why is specifying a remote ref useful for
> > > --revision?
> > 
> > This is how some trybots use it, at least.
> 
> Ok but that's definitely the code of change that should have a proper test.
> That's easy for this to fail silently.

Specifying a remote ref in git is exactly the same as specifying
chrome/branches/foo/src@HEAD as the url for svn. So this is definitely useful.
I'm happy to add a test (though the gclient tests are so convoluted it's not
clear to me how) but it won't actually catch if this change itself is breaking
anything.

[agable, 2014-04-24 22:04:44]

On 2014/04/23 19:52:33, agable wrote:
> On 2014/04/23 00:24:27, M-A Ruel wrote:
> > On 2014/04/23 00:10:46, pgervais wrote:
> > > On 2014/04/22 13:24:07, Marc-Antoine Ruel (Google) wrote:
> > > > I miss background details here. Why is specifying a remote ref useful
for
> > > > --revision?
> > > 
> > > This is how some trybots use it, at least.
> > 
> > Ok but that's definitely the code of change that should have a proper test.
> > That's easy for this to fail silently.
> 
> Specifying a remote ref in git is exactly the same as specifying
> chrome/branches/foo/src@HEAD as the url for svn. So this is definitely useful.
> I'm happy to add a test (though the gclient tests are so convoluted it's not
> clear to me how) but it won't actually catch if this change itself is breaking
> anything.

Ping. Alternatively, this can not land in favor of
https://codereview.chromium.org/250523004/

[M-A Ruel, 2014-04-25 12:57:54]

On 2014/04/24 22:04:44, agable wrote:
> Ping. Alternatively, this can not land in favor of

Ugh, sorry for the delay.

lgtm as long as you keep the pieces. :)

[agable, 2014-04-25 21:39:50]

The CQ bit was checked by agable@chromium.org

[commit-bot: I haz the power, 2014-04-25 21:41:04]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/agable@chromium.org/244253007/20001

[commit-bot: I haz the power, 2014-04-25 21:43:42]

Change committed as 266251