Switch depot_tools python to 2.7.4.

Switch depot_tools python to 2.7.4.

TBR=joi@chromium.org
BUG=

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=200799

[M-A Ruel, 2013-05-08 17:08:21]

+Joi

Sadly, this makes the "pops up a header file on gclient runhooks" happen on the
slave I tested with (a swarm canary slave). So either python 2.6. vs 2.7 has a
subtle difference or it's because of activestate python vs cpython.

[Jói, 2013-05-08 17:58:48]

D'oh!

So I guess this change is blocked on fixing that problem?  I have an old WIP
change that doesn't quite work but I think is in the right direction here:
https://codereview.chromium.org/12330155/

If you're blocked then I can prioritize figuring out the gyp issue.

Cheers,
Jói


On 2013/05/08 17:08:21, Marc-Antoine Ruel wrote:
> +Joi
> 
> Sadly, this makes the "pops up a header file on gclient runhooks" happen on
the
> slave I tested with (a swarm canary slave). So either python 2.6. vs 2.7 has a
> subtle difference or it's because of activestate python vs cpython.

[M-A Ruel, 2013-05-08 18:12:46]

On 2013/05/08 17:58:48, Jói wrote:
> D'oh!
> 
> So I guess this change is blocked on fixing that problem?  I have an old WIP
> change that doesn't quite work but I think is in the right direction here:
> https://codereview.chromium.org/12330155/
> 
> If you're blocked then I can prioritize figuring out the gyp issue.

It'd be appreciated if you could because the roll out is blocked on that
specific issue. I'm a bit lost about the actual details.

[Dirk Pranke, 2013-05-08 18:51:12]

On 2013/05/08 18:12:46, Marc-Antoine Ruel wrote:
> On 2013/05/08 17:58:48, Jói wrote:
> > D'oh!
> > 
> > So I guess this change is blocked on fixing that problem?  I have an old WIP
> > change that doesn't quite work but I think is in the right direction here:
> > https://codereview.chromium.org/12330155/
> > 
> > If you're blocked then I can prioritize figuring out the gyp issue.
> 
> It'd be appreciated if you could because the roll out is blocked on that
> specific issue. I'm a bit lost about the actual details.

Any particular reason we're upgrading to 2.7 on windows, apart from it being
newer and better? I just want to clarify that we're still only requiring 2.6
since at least Mac 10.6 doesn't have 2.7 by default (we could always install it
there, too).

[M-A Ruel, 2013-05-08 18:54:43]

On 2013/05/08 18:51:12, Dirk Pranke wrote:
> Any particular reason we're upgrading to 2.7 on windows, apart from it being
> newer and better? I just want to clarify that we're still only requiring 2.6
> since at least Mac 10.6 doesn't have 2.7 by default (we could always install
it
> there, too).

AFAIK, we already install python 2.7 on OSX 10.6 slaves

[Jói, 2013-05-08 22:29:33]

> It'd be appreciated if you could because the roll out is blocked on that
> specific issue. I'm a bit lost about the actual details.

I'll be in MTV next week (and the week after), can it wait until then?
 Will probably be easiest to fix if I can sit by one of the gyp guys'
desks while I fix it ;)  That and I don't have much left of this week,
tomorrow is a local holiday.

Cheers,
Jói


On Wed, May 8, 2013 at 7:12 PM,  <maruel@chromium.org> wrote:
> On 2013/05/08 17:58:48, Jói wrote:
>>
>> D'oh!
>
>
>> So I guess this change is blocked on fixing that problem?  I have an old
>> WIP
>> change that doesn't quite work but I think is in the right direction here:
>> https://codereview.chromium.org/12330155/
>
>
>> If you're blocked then I can prioritize figuring out the gyp issue.
>
>
> It'd be appreciated if you could because the roll out is blocked on that
> specific issue. I'm a bit lost about the actual details.
>
> https://codereview.chromium.org/14972010/

[M-A Ruel, 2013-05-16 15:43:20]

I tested manually on
http://build.chromium.org/p/chromium.swarm/builders/Windows%20Isolate%20Tests...
and no problem came up.

So this CL is ready to be committed (manually because of CRLF) as soon as it is
LGTM'ed. :)

Contrary to my comment in https://codereview.chromium.org/14962003, I'll do a
slightly different work plan than I initially stated:
1. Commit this CL.
2. Create another CL to bootstrap\win\win_tools.bat so it forcibly updates
python 2.6 users to 2.7, without deleting the 2.6 install so buildbot slaves are
not broken. This
will update all the slaves without manual intervention and will switch
automatically upon reboot.
3. See if anything blows up, revert and fix in that case.
4. Update depot_tools.zip.

[Jói, 2013-05-16 16:56:24]

LGTM

[M-A Ruel, 2013-05-17 12:44:13]

On 2013/05/16 16:56:24, Jói wrote:
> LGTM

I'll commit this morning as the tree happens to be closed and in a state I can
monitor for the next hours.

[M-A Ruel, 2013-05-17 12:46:20]

Committed patchset #1 manually as r200799.

[M-A Ruel, 2013-05-17 12:48:40]

On 2013/05/17 12:46:20, Marc-Antoine Ruel wrote:
> Committed patchset #1 manually as r200799.

Sadly, I had to bypass the presubmit checks because the tests are still broken
on Windows. This should be fixed, first step would be to include svnadmin in
svn_bin/.