Use relative path to reference the libraries in the NDK.

Use relative path to reference the libraries in the NDK.

BUG=462153
Committed: https://crrev.com/a7219b31e06f25b2fb94d79253e9f436ea4b9fb0
Cr-Commit-Position: refs/heads/master@{#318905}

[Fabrice (no longer in Chrome), 2015-03-03 16:42:27]

fdegans@chromium.org changed reviewers:
+ thakis@chromium.org, yang.gu@intel.com

[Fabrice (no longer in Chrome), 2015-03-03 16:42:27]

PTAL
This changes android_ndk_root to be a relative path. We'll only use the absolute
path where it is strictly necessary.

[Nico, 2015-03-03 16:46:27]

lgtm

Awesome :-)

https://codereview.chromium.org/969383004/diff/1/build/common.gypi
File build/common.gypi (left):

https://codereview.chromium.org/969383004/diff/1/build/common.gypi#oldcode1747
build/common.gypi:1747: 'gcc_version%': 49,
Is this no longer needed?

[Fabrice (no longer in Chrome), 2015-03-03 16:54:21]

https://codereview.chromium.org/969383004/diff/1/build/common.gypi
File build/common.gypi (left):

https://codereview.chromium.org/969383004/diff/1/build/common.gypi#oldcode1747
build/common.gypi:1747: 'gcc_version%': 49,
On 2015/03/03 16:46:27, Nico wrote:
> Is this no longer needed?

This is from a time when we were using GCC 4.8 for Android and the mips64
toolchain needed  GCC 4.9. We're now using GCC 4.9 everywhere on Android.

[Fabrice (no longer in Chrome), 2015-03-03 16:54:25]

The CQ bit was checked by fdegans@chromium.org

[commit-bot: I haz the power, 2015-03-03 16:55:41]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/969383004/1

[commit-bot: I haz the power, 2015-03-03 17:58:37]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-03-03 17:59:57]

Patchset 1 (id:??) landed as
https://crrev.com/a7219b31e06f25b2fb94d79253e9f436ea4b9fb0
Cr-Commit-Position: refs/heads/master@{#318905}

[Yang Gu, 2015-03-05 14:22:34]

@Fabrice, today I found your patch would break build with the situation output
is at a specific directory other than 'output'. To support multi-archs build, we
put outputs for each arch in a specific directory. For example, here is the way
we build x86:
1. Generate ninja files
build/gyp_chromium -Dwerror= -Duse_goma=0 -Dtarget_arch=ia32 --generator-output
out-x86
The ninja files would be in out-x86/out/Release/xxx.
2. Build some module, such as content shell
ninja -j16 -C src/out-x86/out/Release content_shell_apk

Then I will get lots of errors. 

If the output directory is just out/Release, everything is OK (I think trybot
just tested this situation). And I think the reason is in my case, it has one
more directory level than just "out". 
If I changed every android_ndk_root to android_ndk_absolute_root in common.gypi,
everything would be ok again. But I'm afraid this might not be the best fix. 

Your idea?

[Nico, 2015-03-05 14:55:17]

Don't use --generator-output, use -Goutput_dir?
On Mar 5, 2015 6:22 AM, <yang.gu@intel.com> wrote:

> @Fabrice, today I found your patch would break build with the situation
> output
> is at a specific directory other than 'output'. To support multi-archs
> build, we
> put outputs for each arch in a specific directory. For example, here is
> the way
> we build x86:
> 1. Generate ninja files
> build/gyp_chromium -Dwerror= -Duse_goma=0 -Dtarget_arch=ia32
> --generator-output
> out-x86
> The ninja files would be in out-x86/out/Release/xxx.
> 2. Build some module, such as content shell
> ninja -j16 -C src/out-x86/out/Release content_shell_apk
>
> Then I will get lots of errors.
>
> If the output directory is just out/Release, everything is OK (I think
> trybot
> just tested this situation). And I think the reason is in my case, it has
> one
> more directory level than just "out".
> If I changed every android_ndk_root to android_ndk_absolute_root in
> common.gypi,
> everything would be ok again. But I'm afraid this might not be the best
> fix.
>
> Your idea?
>
> https://codereview.chromium.org/969383004/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Fabrice (no longer in Chrome), 2015-03-05 15:07:10]

On 2015/03/05 14:22:34, Yang Gu wrote:
> @Fabrice, today I found your patch would break build with the situation output
> is at a specific directory other than 'output'. To support multi-archs build,
we
> put outputs for each arch in a specific directory. For example, here is the
way
> we build x86:
> 1. Generate ninja files
> build/gyp_chromium -Dwerror= -Duse_goma=0 -Dtarget_arch=ia32
--generator-output
> out-x86
> The ninja files would be in out-x86/out/Release/xxx.
> 2. Build some module, such as content shell
> ninja -j16 -C src/out-x86/out/Release content_shell_apk
> 
> Then I will get lots of errors. 
> 
> If the output directory is just out/Release, everything is OK (I think trybot
> just tested this situation). And I think the reason is in my case, it has one
> more directory level than just "out". 
> If I changed every android_ndk_root to android_ndk_absolute_root in
common.gypi,
> everything would be ok again. But I'm afraid this might not be the best fix. 
> 
> Your idea?

I think you are supposed to have 2 levels anyway. I use cr to take care of all
that for me, and strongly recommend it.
https://code.google.com/p/chromium/wiki/CRUserManual
I think what is happening is that you are not setting the output directory
correctly, and the proper way to do it is to add -Goutput_dir=<dirname> to the
gyp_chromium command.

[Yang Gu, 2015-03-06 01:32:17]

Thank you both, -Goutput_dir works for me like a charm!
I read the cr description. It's something like what I have done in my script
(https://github.com/gyagp/share/blob/master/python/chromium.py). We use this
script to do everything about Chromium, from code checkout, build to all kinds
of tests. Actually I think it might be a good idea for Chromium project to have
such a script, so that we don't need to read code and document here and there,
and use various commands/scripts to interact with Chromium.