Get rid of build_angle.gypi

[GoogleCodeExporter]

It is weird to have all targets in .gypi and a wrapper .gyp that does not do 
anything.

I see that .gypi file is only being used by skia. I do not understand why they 
need it. There is some discussion here https://codereview.appspot.com/6495055, 
but I still do not understand the real issue. If chrome can use .gyp files in 
other directories, why can't skia?

Original issue reported on code.google.com by alokp@chromium.org on 13 Aug 2013 at 9:28

• Blocking: #744
• Blocked on: #537

[GoogleCodeExporter]

Eric/Elliot: Can you help me understand the issue. I would like to get rid of 
this .gypi file. We use the .gyp file in chrome and webkit without any problem.

Original comment by alokp@chromium.org on 13 Aug 2013 at 9:30

[GoogleCodeExporter]

Elliot probably remembers more of this than I do, but we have been unable to 
use gyp files in other directories successfully.  For other external projects, 
we've just duplicated their gyp file in our repo, which we'd hoped to avoid 
with ANGLE.  This bug chronicles Elliot's attempt with libjpeg: 
https://code.google.com/p/skia/issues/detail?id=543

Original comment by bore...@google.com on 14 Aug 2013 at 2:04

[GoogleCodeExporter]

alokp: if you can change 
https://code.google.com/p/skia/source/browse/trunk/gyp/angle.gyp so that it 
depends on build_angle.gyp instead of build_angle.gypi, and "make all" still 
works on Windows, we're all for it.

As Eric mentioned, our previous attempts at something similar (with libjpeg) 
are documented at https://code.google.com/p/skia/issues/detail?id=543 .  I have 
come to believe that gyp's behavior is quite fragile when directory paths are 
changed.

Given our previously fruitless attempts at this, I'm not raring to spend a lot 
of time on it (given that I suspect we will ultimately fail yet again).  Is 
keeping the .gypi file separate causing substantial pain, or just a mild 
annoyance?

Original comment by epoger@google.com on 14 Aug 2013 at 7:10

[GoogleCodeExporter]

Original comment by jmad...@chromium.org on 9 Dec 2013 at 9:04

• Now blocked on: #537

[GoogleCodeExporter]

Original comment by geofflang@chromium.org on 10 Dec 2013 at 3:49

[GoogleCodeExporter]

Was considering tackling this once we finish the MANGLE stuff, but it's up for 
grabs as of now.

Original comment by jmad...@chromium.org on 2 Jul 2014 at 8:38

[GoogleCodeExporter]

Original comment by jmad...@chromium.org on 19 Sep 2014 at 3:14

• Now blocking: #744

[GoogleCodeExporter]

Going to tackle this clean-up now.

Original comment by jmad...@chromium.org on 19 Sep 2014 at 3:14

• Changed state: Started

[GoogleCodeExporter]

see issue chromium:415983 for most of the changes.

Original comment by jmad...@chromium.org on 30 Sep 2014 at 9:56

[GoogleCodeExporter]

As far as Chromium/Blink/ANGLE are concerned, we should be able to delete 
build_angle.gypi. There's a question about WebKit/Safari... or Firefox, or 
other projects. Maybe we can alert people this file will be going away soon?

Original comment by jmad...@chromium.org on 2 Oct 2014 at 6:36

[GoogleCodeExporter]

I think Safari/general WebKit probably maintains separate build files, but am 
unsure. Firefox, I think, recently converted to use gyp, at least to build 
ANGLE. Do they just need to use angle.gyp the same as they used to use 
build_angle.gyp/gypi?

Original comment by shannonw...@chromium.org on 2 Oct 2014 at 6:44

[GoogleCodeExporter]

Yeah, build_angle.gyp (and angle.gyp as well) is just a passthrough to 
angle.gypi, where the logic of our gyp stuff gets included. The fix to anyone 
using build_angle.gyp is just to change that to angle.gyp.

Original comment by jmad...@chromium.org on 2 Oct 2014 at 7:03

[GoogleCodeExporter]

Closing this out, we no longer have build_angle.gypi.

Original comment by jmad...@chromium.org on 10 Nov 2014 at 9:06

• Changed state: Fixed