Make git-cl work with OWNERS files in a non .git/hooks/pre-cl-* world

Make git-cl work with OWNERS file hooks properly.

This version calls into presubmit_support directly to support the OWNERS hooks. We do not need both this patch and http://codereview.chromium.org/6646009/

This patch depends on http://codereview.chromium.org/6665018/ .

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=77898

[Dirk Pranke, 2011-03-11 05:39:03]

Okay, here's the version w/o hooks. Substantially smaller.

[M-A Ruel, 2011-03-11 18:00:06]

Looks sane to me. Maybe it could implemented differently eventually but right
now it's fine.

lgtm

http://codereview.chromium.org/6674014/diff/3008/git_cl/git_cl.py
File git_cl/git_cl.py (right):

http://codereview.chromium.org/6674014/diff/3008/git_cl/git_cl.py#newcode497
git_cl/git_cl.py:497: reviewer_regexp = re.compile('ADD: R=(.+)')
I wonder if instead it could be a more generic implementation:
The presubmit check could return the new description.
This way BUG=, TEST= and others could be added by the presubmit check instead of
git-cl/gcl.

[Dirk Pranke, 2011-03-11 21:06:31]

http://codereview.chromium.org/6674014/diff/3008/git_cl/git_cl.py
File git_cl/git_cl.py (right):

http://codereview.chromium.org/6674014/diff/3008/git_cl/git_cl.py#newcode497
git_cl/git_cl.py:497: reviewer_regexp = re.compile('ADD: R=(.+)')
On 2011/03/11 18:00:06, Marc-Antoine Ruel wrote:
> I wonder if instead it could be a more generic implementation:
> The presubmit check could return the new description.
> This way BUG=, TEST= and others could be added by the presubmit check instead
of
> git-cl/gcl.

Yeah, that would probably be a nice enhancement. However, I think it would be
better to find a way that wasn't parsing text to begin with. Depends on how we
cross process boundaries, I suppose.

[M-A Ruel, 2011-03-11 21:21:54]

http://codereview.chromium.org/6674014/diff/3008/git_cl/git_cl.py
File git_cl/git_cl.py (right):

http://codereview.chromium.org/6674014/diff/3008/git_cl/git_cl.py#newcode789
git_cl/git_cl.py:789: if not res:
What about PresubmitNotifyResult?

[Dirk Pranke, 2011-03-11 21:39:18]

http://codereview.chromium.org/6674014/diff/3008/git_cl/git_cl.py
File git_cl/git_cl.py (right):

http://codereview.chromium.org/6674014/diff/3008/git_cl/git_cl.py#newcode789
git_cl/git_cl.py:789: if not res:
On 2011/03/11 21:21:55, Marc-Antoine Ruel wrote:
> What about PresubmitNotifyResult?

Good catch. It would get printed in some but not all other cases,  and having
the logic scattered around is a bad thing. Will fix.

[Dirk Pranke, 2011-03-11 23:07:27]

On 2011/03/11 21:53:33, dpranke wrote:

Per IM w/ anush@ and msb@ we should be safe for chromiumos as well.

[M-A Ruel, 2011-03-11 23:11:14]

http://codereview.chromium.org/6674014/diff/9002/git_cl/git_cl.py
File git_cl/git_cl.py (right):

http://codereview.chromium.org/6674014/diff/9002/git_cl/git_cl.py#newcode788
git_cl/git_cl.py:788: default_presubmit=None, may_prompt=None, tbr=tbr,
You disable may_prompt so that converts all warnings into errors. Is it what you
want? Since stdout is redirected, I'm not sure about the best thing to do there.

[Dirk Pranke, 2011-03-12 00:31:57]

On 2011/03/11 23:11:14, Marc-Antoine Ruel wrote:
> http://codereview.chromium.org/6674014/diff/9002/git_cl/git_cl.py
> File git_cl/git_cl.py (right):
> 
> http://codereview.chromium.org/6674014/diff/9002/git_cl/git_cl.py#newcode788
> git_cl/git_cl.py:788: default_presubmit=None, may_prompt=None, tbr=tbr,
> You disable may_prompt so that converts all warnings into errors. Is it what
you
> want? Since stdout is redirected, I'm not sure about the best thing to do
there.

AFAICT, git-cl was consistently swallowing the prompts and/or bypassing the
hooks when you did -f.

I have attempted to make things a little more rational and consistent. 

I've also deleted a bit more unnecessary code and done some more linting. And I
fixed the post-commit hook :)

[M-A Ruel, 2011-03-12 03:02:21]

lgtm with verification about may_prompt.

http://codereview.chromium.org/6674014/diff/6003/git_cl/git_cl.py
File git_cl/git_cl.py (right):

http://codereview.chromium.org/6674014/diff/6003/git_cl/git_cl.py#newcode862
git_cl/git_cl.py:862: may_prompt=(not options.force))
Can you test it to see what happens when may_prompt=True and a PromptWarning is
generated?

[Dirk Pranke, 2011-03-12 04:38:44]

On 2011/03/12 03:02:21, Marc-Antoine Ruel wrote:
> lgtm with verification about may_prompt.
> 
> http://codereview.chromium.org/6674014/diff/6003/git_cl/git_cl.py
> File git_cl/git_cl.py (right):
> 
> http://codereview.chromium.org/6674014/diff/6003/git_cl/git_cl.py#newcode862
> git_cl/git_cl.py:862: may_prompt=(not options.force))
> Can you test it to see what happens when may_prompt=True and a PromptWarning
is
> generated?

Of course, it didn't work :( But I had already committed the change.

I've committed a follow-on that actually does work (tested by hand) and TBR'ed
that to you. 

http://codereview.chromium.org/6682018/

Sorry!