Remove use_system_ffmpeg-related logic from media and other mainline gyp files.

build/linux/unbundle/ffmpeg.gyp

Index: build/linux/unbundle/ffmpeg.gyp
diff --git a/build/linux/unbundle/ffmpeg.gyp b/build/linux/unbundle/ffmpeg.gyp
index dfaaf0aee83668764b07c97846a035c933c4110d..e3c3723894ceae651920154dbb65cf83aa4d3d40 100644
--- a/build/linux/unbundle/ffmpeg.gyp
+++ b/build/linux/unbundle/ffmpeg.gyp
@@ -10,6 +10,31 @@
       'direct_dependent_settings': {
         'cflags': [
           '<!@(pkg-config --cflags libavcodec libavformat libavutil)',
+
+          '<!(python <(DEPTH)/tools/compile_test/compile_test.py '
+              '--code "#define __STDC_CONSTANT_MACROS\n'
+              '#include <libavcodec/avcodec.h>\n'
+              'int test() { return AV_CODEC_ID_OPUS; }" '
+              '--on-failure -DCHROMIUM_OMIT_AV_CODEC_ID_OPUS=1)',

[Ami GONE FROM CHROMIUM, 2013/05/05 03:41:53]

What will make use of this define once this CL lands?
(ditto for others below)

[Paweł Hajdan Jr., 2013/05/06 23:48:56]

This will need a custom distro patch. Still, IMHO what can be shared should be
shared, making the patch smaller.

[Ami GONE FROM CHROMIUM, 2013/05/06 23:55:04]

Isn't each distro going to need the same exact patch b/c they'll be patching the
same chromium source to do the same thing?
(I'm not complaining, just curious)

[Paweł Hajdan Jr., 2013/05/07 17:14:45]

Good question.

First, custom patching is really something to be avoided. You can see in various
distro guidelines, e.g.
https://fedoraproject.org/wiki/Staying_close_to_upstream_projects that all
changes that can be sent upstream should be - and at least this gyp part can
live upstream.

Then, most distro packagers are not familiar with gyp and our build system. C++
is something we can reasonably expect them to know however. When custom patching
it's valuable to understand the patch one's applying instead of doing sort of
cargo cult - it works for distro X so I'm taking it even though I have no idea
what's going on (applying patches without full understanding reminds me of
Debian PRNG debacle).

And then somewhat minor reason, but keeping the custom patch smaller is also of
value.

+
+          '<!(python <(DEPTH)/tools/compile_test/compile_test.py '
+              '--code "#define __STDC_CONSTANT_MACROS\n'
+              '#include <libavcodec/avcodec.h>\n'
+              'int test() { return AV_CODEC_ID_VP9; }" '
+              '--on-failure -DCHROMIUM_OMIT_AV_CODEC_ID_VP9=1)',
+
+          '<!(python <(DEPTH)/tools/compile_test/compile_test.py '
+              '--code "#define __STDC_CONSTANT_MACROS\n'
+              '#include <libavcodec/avcodec.h>\n'
+              'int test() { return AV_PKT_DATA_MATROSKA_BLOCKADDITIONAL; }" '
+              '--on-failure -DCHROMIUM_OMIT_AV_PKT_DATA_MATROSKA_BLOCKADDITIONAL=1)',
+
+          '<!(python <(DEPTH)/tools/compile_test/compile_test.py '
+              '--code "#define __STDC_CONSTANT_MACROS\n'
+              '#include <libavcodec/avcodec.h>\n'
+              'int test() { struct AVFrame frame;\n'
+              'return av_frame_get_channels(&frame); }" '
+              '--on-failure -DCHROMIUM_NO_AVFRAME_CHANNELS=1)',
         ],
         'defines': [
           '__STDC_CONSTANT_MACROS',