ia32 option for iOS does not produce i386 lib

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Compiling with the ia32 setting for iOS produces a x86_64 library 

What is the expected output? What do you see instead?
When compiling with ia32 I expect the i386 lib but receive the 64bit version.

What version of the product are you using? On what operating system?
yuv trunk on OS-X using xcode 8.3 tools.

Please provide any additional information below.
It is most likely just a minor detail in the build script but our iOS devs 
wants to maintain the possibility to use the older iOS simulators so we really 
need the i386 builds to build as well.

All other options, armv7, armv7s, arm64 and x86_64 works fine. 

Original issue reported on code.google.com by ilantzco...@gmail.com on 25 Jun 2015 at 11:12

• Blocked on: #456

[GoogleCodeExporter]

currently blocked on bug 456 - can't build on ios.

the ia32 behavior is likely defined by the chromium dependencies.  if you could 
track back to the last good version, we could try rolling chromium deps back to 
that

Original comment by fbarch...@google.com on 26 Jun 2015 at 9:43

• Now blocked on: #456

[GoogleCodeExporter]

I just posted a solution for that problem in issue 456.

Original comment by kjellan...@google.com on 28 Jun 2015 at 7:06

[GoogleCodeExporter]

When did this last work for ia32?
Try rolling back chromium deps to an older version?

Original comment by fbarch...@google.com on 29 Jun 2015 at 5:49

[GoogleCodeExporter]

The Getting started doesnt show the simulator settings.  Starting with arm 64 
bit:

arm64

 GYP_DEFINES="OS=ios target_arch=armv7 target_subarch=64" GYP_CROSSCOMPILE=1 GYP_GENERATOR_FLAGS="output_dir=out_ios" ./gyp_libyuv -f ninja --depth=. libyuv_test.gyp 
 ninja -j7 -C out_ios/Debug-iphoneos libyuv_unittest

we get 64 bit arm as expected

bash-3.2$ file ./out_ios/Release-iphoneos/libyuv_unittest.app/libyuv_unittest 
./out_ios/Release-iphoneos/libyuv_unittest.app/libyuv_unittest: Mach-O 64-bit 
executable

Do 32 similulator like this:
GYP_DEFINES="OS=ios target_arch=ia32 target_subarch=32" GYP_CROSSCOMPILE=1 
GYP_GENERATOR_FLAGS="output_dir=out_sim" ./gyp_libyuv -f ninja --depth=. 
libyuv_test.gyp 

ninja -j7 -C out_sim/Debug-iphonesimulator libyuv_unittest

bash-3.2$ file 
./out_sim/Debug-iphonesimulator/libyuv_unittest.app/libyuv_unittest 
./out_sim/Debug-iphonesimulator/libyuv_unittest.app/libyuv_unittest: Mach-O 
universal binary with 2 architectures
./out_sim/Debug-iphonesimulator/libyuv_unittest.app/libyuv_unittest (for 
architecture i386): Mach-O executable i386
./out_sim/Debug-iphonesimulator/libyuv_unittest.app/libyuv_unittest (for 
architecture x86_64):   Mach-O 64-bit executable x86_64

Original comment by fbarch...@google.com on 30 Jun 2015 at 11:16

• Changed state: Accepted

[GoogleCodeExporter]

webrtc documents parts of the build

without target_subarch=32 it produces only the x86_64 version.
with target_subarch=32 it produces both and uses lipo to join them.
I've updated Getting Started

Original comment by fbarch...@google.com on 1 Jul 2015 at 12:00

• Changed state: Fixed

[GoogleCodeExporter]

I have more information on this.  arm had the same issue:
https://code.google.com/p/libyuv/issues/detail?id=467

Looking in common.gypi there is a behavior change for arch/subarch to default 
to 64 bit
Its not well documented, but you can determine the valid setting from this 
snippet:

            # Limit the valid architectures depending on "target_subarch".
            # This need to include the "arm" architectures but also the "x86"
            # ones (they are used when building for the simulator).
            ['target_subarch=="arm32"', {
              'VALID_ARCHS': ['armv7', 'i386'],
            }],
            ['target_subarch=="arm64"', {
              'VALID_ARCHS': ['arm64', 'x86_64'],
            }],
            ['target_subarch=="both"', {
              'VALID_ARCHS': ['arm64', 'armv7', 'x86_64', 'i386'],
            }],

for 32 bit arm
target_arch=i386
target_subarch=arm32

I updated the Getting Started

Original comment by fbarch...@google.com on 15 Jul 2015 at 10:09

• Changed state: Started