Make gyp_chromium.py work with Python 2.7.

Make gyp_chromium.py work with Python 2.7.

gyp_chromium.py contained a Python 2.6 Windows bug workaround which turns out to be harmful with Python 2.7 since it will just include itself without doing anything.

BUG=
R=scottmg@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=202578

[Daniel Bratell, 2013-05-15 12:20:30]

This is really no big deal but I wasted some time trying to figure out why
things were broken so I imagines this could save some time for someone else.

[scottmg, 2013-05-15 15:22:39]

Will this work on 2.6? I don't think multiprocessing imports with __name__ ==
'__main__'?

[Daniel Bratell, 2013-05-15 15:34:49]

On 2013/05/15 15:22:39, scottmg wrote:
> Will this work on 2.6? I don't think multiprocessing imports with __name__ ==
> '__main__'?

It works when run from the console in with python 2.6, but I didn't test with
multiprocessing because I don't know what it is, but I think you're right.

If you point me to multiprocessing I can check what it does and see if I can put
together a better solution.

[scottmg, 2013-05-15 15:43:48]

On 2013/05/15 15:34:49, bratell wrote:
> On 2013/05/15 15:22:39, scottmg wrote:
> > Will this work on 2.6? I don't think multiprocessing imports with __name__
==
> > '__main__'?
> 
> It works when run from the console in with python 2.6, but I didn't test with
> multiprocessing because I don't know what it is, but I think you're right.
> 
> If you point me to multiprocessing I can check what it does and see if I can
put
> together a better solution.

If you set an environment variable GYP_PARALLEL=1 then gyp uses
http://docs.python.org/2/library/multiprocessing.html to parse the gyp input
files in parallel, and generate the output configs in parallel (for ninja). This
file was working around people running "python build\gyp_chromium" on windows,
but multiprocessing expecting to find a script named "gyp_chromium.py" when
relaunching subprocesses.

[Daniel Bratell, 2013-05-16 11:54:47]

I just removed the error handling code that tried to be nice to the user. If it
can't detect an error condition, well, then it can't.

[scottmg, 2013-05-16 16:44:30]

lgtm with comment

https://codereview.chromium.org/14670009/diff/6001/build/gyp_chromium.py
File build/gyp_chromium.py (right):

https://codereview.chromium.org/14670009/diff/6001/build/gyp_chromium.py#newc...
build/gyp_chromium.py:8: # their command line to add a .py when running
gyp_chromium.
please add "This is necessary when multiprocessing reinvokes gyp_chromium when
using GYP_PARALLEL=1" or similar.

[Nico, 2013-05-16 16:53:43]

(nit: cl descript says ".oy" instead of ".py". also might want to mention that
this mostly affects windows since gyp_chromium.py isn't used elsewhere afaik)

[Daniel Bratell, 2013-05-20 13:51:48]

On 2013/05/16 16:53:43, Nico wrote:
> (nit: cl descript says ".oy" instead of ".py". also might want to mention that
> this mostly affects windows since gyp_chromium.py isn't used elsewhere afaik)

Do I need a new lgtm for the updated patch or does the old one still apply?

[scottmg, 2013-05-20 15:24:03]

On 2013/05/20 13:51:48, bratell wrote:
> On 2013/05/16 16:53:43, Nico wrote:
> > (nit: cl descript says ".oy" instead of ".py". also might want to mention
that
> > this mostly affects windows since gyp_chromium.py isn't used elsewhere
afaik)
> 
> Do I need a new lgtm for the updated patch or does the old one still apply?

Just fix "gyp_chromuim.py" -> "gyp_chromium.py" in the description and lgtm.

[commit-bot: I haz the power, 2013-05-20 15:42:04]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bratell@opera.com/14670009/12001

[commit-bot: I haz the power, 2013-05-20 15:42:07]

Failed to apply patch for build/gyp_chromium.py:
While running patch -p1 --forward --force --no-backup-if-mismatch;
  patching file build/gyp_chromium.py
  Hunk #1 FAILED at 2.
  1 out of 1 hunk FAILED -- saving rejects to file build/gyp_chromium.py.rej

Patch:       build/gyp_chromium.py
Index: build/gyp_chromium.py
diff --git a/build/gyp_chromium.py b/build/gyp_chromium.py
index
29bac98c9253f29d0ad7af25d7f29bc9227c3623..c22398e2039e2ab110c59063eae59af365b3a246
100644
--- a/build/gyp_chromium.py
+++ b/build/gyp_chromium.py
@@ -2,8 +2,17 @@
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
-# This is a workaround for multiprocessing on Windows. Importing in Python on
-# Windows doesn't search for imports that don't end in .py (and aren't
-# directories with an __init__.py). So, add this wrapper to avoid having
-# people change their command line to add a .py when running gyp_chromium.
-__import__('gyp_chromium')
+# This file is (possibly, depending on python version) imported by
+# gyp_chromium when GYP_PARALLEL=1 and it creates sub-processes
+# through the multiprocessing library.
+
+# Importing in Python 2.6 (fixed in 2.7) on Windows doesn't search for
+# imports that don't end in .py (and aren't directories with an
+# __init__.py). This wrapper makes "import gyp_chromium" work with
+# those old versions and makes it possible to execute gyp_chromium.py
+# directly on Windows where the extension is useful.
+
+import os
+
+path = os.path.abspath(os.path.split(__file__)[0])
+execfile(os.path.join(path, 'gyp_chromium'))

[Daniel Bratell, 2013-05-20 16:59:04]

Any guess why the patch doesn't apply? I've uploaded it with "git cl upload" and
the file hasn't been changed for a year.

[scottmg, 2013-05-20 17:04:18]

On 2013/05/20 16:59:04, bratell wrote:
> Any guess why the patch doesn't apply? I've uploaded it with "git cl upload"
and
> the file hasn't been changed for a year.

Possibly line endings? I might have committed the original with dos line endings
mistakenly. I'm not sure how to fix it w/o dcommitting.

[Daniel Bratell, 2013-05-28 09:19:10]

scottmg, do you think you could help me land this? I have no other way of doing
it than checking the Commit checkbox and if the commitbot doesn't want to help
there is nothing I can do. (I did mail it about this but got no response.)

[scottmg, 2013-05-28 15:54:33]

On 2013/05/28 09:19:10, bratell wrote:
> scottmg, do you think you could help me land this? I have no other way of
doing
> it than checking the Commit checkbox and if the commitbot doesn't want to help
> there is nothing I can do. (I did mail it about this but got no response.)

OK, I'll land it. It seems to work OK on 2.6, but I haven't tried on 2.7.

[scottmg, 2013-05-28 16:02:19]

Committed patchset #3 manually as r202578 (presubmit successful).

[Daniel Bratell, 2013-05-28 18:11:05]

On 2013/05/28 16:02:19, scottmg wrote:
> Committed patchset #3 manually as r202578 (presubmit successful).

Thanks a lot for your help!