When running git presubmit hooks, use the changelist's description,...

When running git presubmit hooks, use the changelist's description,
rather than the local commit message's.

BUG=31695
TEST=git cl upload where commit message is missing BUG= but CL description has one. No presubmit error.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=35682

[M-A Ruel, 2010-01-06 21:41:54]

Eh, was much easier than I thought. Thanks for looking into it.

lgtm

[chase, 2010-01-08 19:25:00]

I think this patch breaks in my Cygwin setup that might be exposing some
underlying issue.  When I run git-cl upload now, I see this error:

Traceback (most recent call last):
  File "/d/dev/chromium/src/.git/hooks/pre-cl-upload", line 40, in <module>
    exec git_cl_hooks.RunHooks(hook_name=program_name,
upstream_branch=sys.argv[1])
  File "/d/dev/depot_tools/git_cl_hooks.py", line 58, in RunHooks
    options = ChangeOptions(commit=commit, upstream_branch=upstream_branch)
  File "/d/dev/depot_tools/git_cl_hooks.py", line 48, in __init__
    description = gcl.GetIssueDescription(int(issue))
ValueError: invalid literal for int() with base 10: '\x1b[?1034hNone'
Command "/d/dev/chromium/src/.git/hooks/pre-cl-upload refs/remotes/origin/trunk"
failed.

I don't recognize the escape sequence that's getting prepended to the git-cl
output.  Reverting r35682 lets me upload my patch.

[M-A Ruel, 2010-01-08 19:27:44]

On 2010/01/08 19:25:00, chase wrote:
> I think this patch breaks in my Cygwin setup that might be exposing some
> underlying issue.  When I run git-cl upload now, I see this error:
> 
> Traceback (most recent call last):
>   File "/d/dev/chromium/src/.git/hooks/pre-cl-upload", line 40, in <module>
>     exec git_cl_hooks.RunHooks(hook_name=program_name,
> upstream_branch=sys.argv[1])
>   File "/d/dev/depot_tools/git_cl_hooks.py", line 58, in RunHooks
>     options = ChangeOptions(commit=commit, upstream_branch=upstream_branch)
>   File "/d/dev/depot_tools/git_cl_hooks.py", line 48, in __init__
>     description = gcl.GetIssueDescription(int(issue))
> ValueError: invalid literal for int() with base 10: '\x1b[?1034hNone'
> Command "/d/dev/chromium/src/.git/hooks/pre-cl-upload
refs/remotes/origin/trunk"
> failed.
> 
> I don't recognize the escape sequence that's getting prepended to the git-cl
> output.  Reverting r35682 lets me upload my patch.

IIRC, it's a color escape sequence. Is your "None" colored?

[chase, 2010-01-08 19:30:42]

On 2010/01/08 19:27:44, Marc-Antoine Ruel wrote:
> IIRC, it's a color escape sequence. Is your "None" colored?

No, according to the output of 'git-cl status --field=id'.

cmp@cphillips4-corp:/work/chromium/src$ git-cl status --field=id|md5sum
390d3398a07a90b813bbf4fe52f1e8f1 *-
cmp@cphillips4-corp:/work/chromium/src$ echo "None"|md5sum
08e9de24ab150d083240557cec091d53 *-

[M-A Ruel, 2010-01-08 19:34:15]

On 2010/01/08 19:30:42, chase wrote:
> On 2010/01/08 19:27:44, Marc-Antoine Ruel wrote:
> > IIRC, it's a color escape sequence. Is your "None" colored?
> 
> No, according to the output of 'git-cl status --field=id'.
> 
> cmp@cphillips4-corp:/work/chromium/src$ git-cl status --field=id|md5sum
> 390d3398a07a90b813bbf4fe52f1e8f1 *-
> cmp@cphillips4-corp:/work/chromium/src$ echo "None"|md5sum
> 08e9de24ab150d083240557cec091d53 *-

It'd be simpler to just catch exception for int() and not look for hardcoded
strings.

[Robert Sesek, 2010-01-08 19:40:35]

On 2010/01/08 19:34:15, Marc-Antoine Ruel wrote:
> On 2010/01/08 19:30:42, chase wrote:
> > On 2010/01/08 19:27:44, Marc-Antoine Ruel wrote:
> > > IIRC, it's a color escape sequence. Is your "None" colored?
> > 
> > No, according to the output of 'git-cl status --field=id'.
> > 
> > cmp@cphillips4-corp:/work/chromium/src$ git-cl status --field=id|md5sum
> > 390d3398a07a90b813bbf4fe52f1e8f1 *-
> > cmp@cphillips4-corp:/work/chromium/src$ echo "None"|md5sum
> > 08e9de24ab150d083240557cec091d53 *-
> 
> It'd be simpler to just catch exception for int() and not look for hardcoded
> strings.

Easy enough. CL coming your way in a few minutes.