Modify how gpu.gyp includes angle.gyp.

Modify how gpu.gyp includes angle.gyp.

Instead of including dependencies on angle.gyp and then removing
them if skia_angle is not true, only include angle.gyp to begin
with skia_angle is true.

This allows gyp_to_android.py to work when run in the Android tree,
where we don't have third_party/externals (since we do only a git
merge, and no gclient sync.

Committed: http://code.google.com/p/skia/source/detail?r=13618

[scroggo, 2014-02-26 22:35:09]

Alternatively, I could just have the bot run gyp_to_android BEFORE removing
third_party/

Unfortunately, it's gyp (and not my script) that tries to read the included file
even though it's inside a condition that evaluated to false.

[djsollen, 2014-02-27 13:44:13]

lgtm

[robertphillips, 2014-02-27 13:44:30]

lgtm

[djsollen, 2014-02-27 13:48:19]

on second thought does it not work to update angle.gyp to move the includes from
the top level conditions block into the conditions within the angle target.

[bsalomon, 2014-02-27 14:37:18]

I'm not going to stand in the way of this but I do like the include-and-remove
pattern because the files still show up in the IDE when the gyp flag is not
present. It makes it easier to navigate through the code in VS. In gpu-land
we've been pretty consistent about this.

[scroggo, 2014-02-27 14:38:37]

On 2014/02/27 13:48:19, djsollen wrote:
> on second thought does it not work to update angle.gyp to move the includes
from
> the top level conditions block into the conditions within the angle target.

No, that does not work.

[scroggo, 2014-02-27 15:09:14]

On 2014/02/26 22:35:09, scroggo wrote:
> Alternatively, I could just have the bot run gyp_to_android BEFORE removing
> third_party/

Derek pointed out that my alternate approach would not have worked, since the
file included by angle.gyp is in third_party/external, which we don't have when
merging to Android (since we do not gclient sync).

PTAL; the latest patch adds the Skia angle sources (but not the dependencies)
always, and then removes them if skia_angle is not set. I think this should
satisfy Brian's needs and mine.

[djsollen, 2014-02-27 15:15:21]

lgtm part 2

[bsalomon, 2014-02-27 15:23:29]

On 2014/02/27 15:15:21, djsollen wrote:
> lgtm part 2

lgtm

[scroggo, 2014-02-27 20:17:36]

The CQ bit was checked by scroggo@google.com

[commit-bot: I haz the power, 2014-02-27 20:17:46]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/scroggo@google.com/178223015/20001

[commit-bot: I haz the power, 2014-02-27 20:40:36]

Change committed as 13618