Use the icu data file on Linux (but not on CrOS)

Use the icu data file on Linux (but not on CrOS)

Set icu_use_data_file_flag to 1 for POSIX (other than Mac/iOS/Android).

Mac and Android will be handled in a separate CL.

This has to wait until https://codereview.chromium.org/111723007/ is rolled in.
(done by https://codereview.chromium.org/118313004/ )

BUG=72633
TEST=All linux builds go through and all tests pass and there's no perf (start-up etc) regression.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=246678

R=mark@chromium.org, mmoss@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=247045

[Mark Mentovai, 2013-12-16 18:51:57]

LGTM

[commit-bot: I haz the power, 2013-12-16 21:32:11]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jshin@chromium.org/102413007/1

[commit-bot: I haz the power, 2013-12-16 22:43:19]

Retried try job too often on linux_aura for step(s) app_list_unittests,
aura_unittests, browser_tests, compositor_unittests, content_browsertests,
content_unittests, events_unittests, interactive_ui_tests, unit_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_aura...

[jungshik at Google, 2013-12-17 00:17:31]

Non-shared builds do not work with this. I have to get 'copies' out of
'link_settings'. With that, both shared and non-shared builds work.  I had to do
that for Windows and it seems that I have to do the same for Linux.

[jungshik at Google, 2013-12-19 09:26:44]

On 2013/12/17 00:17:31, Jungshik Shin wrote:
> Non-shared builds do not work with this. I have to get 'copies' out of
> 'link_settings'. With that, both shared and non-shared builds work.  I had to
do
> that for Windows and it seems that I have to do the same for Linux.

linux_chromeos failure is apparently due to that 'icudtl.dat' is not listed in
the corresponding *isolate file.  content_unittests appears to be run as an
isolated test according to the log (copied below) : 
rtl.cc failure is due to a missing icudtl.dat. 

/usr/bin/python /mnt/scratch0/b_used/build/scripts/slave/runisolatedtest.py
--test_name content_unittests --builder_name linux_chromeos --checkout_dir
/mnt/scratch0/b_used/build/slave/linux_chromeos/build
/mnt/scratch0/b_used/build/slave/linux_chromeos/build/src/out/Release/content_unittests
--
/mnt/scratch0/b_used/build/slave/linux_chromeos/build/src/out/Release/content_unittests
--brave-new-test-launcher --test-launcher-bot-mode --gtest_print_time
--test-launcher-summary-output=/tmp/tmpjHn4yq

/mnt/scratch0/b_used/build/slave/linux_chromeos/build/src/out/Release/content_unittests
--brave-new-test-launcher --test-launcher-bot-mode --gtest_print_time
--test-launcher-summary-output=/tmp/tmpjHn4yq
[1219/010333:FATAL:rtl.cc(145)] Check failed: U_SUCCESS(status).

[jungshik at Google, 2014-01-22 00:21:28]

On 2013/12/19 09:26:44, Jungshik Shin wrote:
> On 2013/12/17 00:17:31, Jungshik Shin wrote:
> > Non-shared builds do not work with this. I have to get 'copies' out of
> > 'link_settings'. With that, both shared and non-shared builds work.  I had
to
> do
> > that for Windows and it seems that I have to do the same for Linux.
> 
> linux_chromeos failure is apparently due to that 'icudtl.dat' is not listed in
> the corresponding *isolate file.  content_unittests appears to be run as an
> isolated test according to the log (copied below) : 
> rtl.cc failure is due to a missing icudtl.dat. 
> 
> /usr/bin/python /mnt/scratch0/b_used/build/scripts/slave/runisolatedtest.py
> --test_name content_unittests --builder_name linux_chromeos --checkout_dir
> /mnt/scratch0/b_used/build/slave/linux_chromeos/build
>
/mnt/scratch0/b_used/build/slave/linux_chromeos/build/src/out/Release/content_unittests
> --
>
/mnt/scratch0/b_used/build/slave/linux_chromeos/build/src/out/Release/content_unittests
> --brave-new-test-launcher --test-launcher-bot-mode --gtest_print_time
> --test-launcher-summary-output=/tmp/tmpjHn4yq
> 
>
/mnt/scratch0/b_used/build/slave/linux_chromeos/build/src/out/Release/content_unittests
> --brave-new-test-launcher --test-launcher-bot-mode --gtest_print_time
> --test-launcher-summary-output=/tmp/tmpjHn4yq
> [1219/010333:FATAL:rtl.cc(145)] Check failed: U_SUCCESS(status).

https://codereview.chromium.org/124143002/ took care of isolate-related issues.
I'll land after the corresponding windows and mac CLs got it (to avoid a
potential conflict).

[jungshik at Google, 2014-01-22 21:29:33]

mmoss@chromium.org: Please review changes in chrome/{installer,build}/linux.  

Thank you.

[Michael Moss, 2014-01-22 22:06:13]

https://codereview.chromium.org/102413007/diff/40001/chrome/installer/linux/c...
File chrome/installer/linux/common/installer.include (right):

https://codereview.chromium.org/102413007/diff/40001/chrome/installer/linux/c...
chrome/installer/linux/common/installer.include:136: if [ -r
"${BUILDDIR}/icudtl.dat" ]; then
If this file is not optional (if it should always be built, or if Chrome will
fail if it's missing), then don't wrap it in an "if -r". Otherwise, LGTM.

[jungshik at Google, 2014-01-22 22:17:39]

On 2014/01/22 22:06:13, Michael Moss wrote:
>
https://codereview.chromium.org/102413007/diff/40001/chrome/installer/linux/c...
> File chrome/installer/linux/common/installer.include (right):
> 
>
https://codereview.chromium.org/102413007/diff/40001/chrome/installer/linux/c...
> chrome/installer/linux/common/installer.include:136: if [ -r
> "${BUILDDIR}/icudtl.dat" ]; then
> If this file is not optional (if it should always be built, or if Chrome will
> fail if it's missing), then don't wrap it in an "if -r". Otherwise, LGTM.

If Chrome is built with icu_use_data_file_flag == 1, it's not optional. This CL
is setting  icu_use_data_file_flag to 1 so that I guess I should not wrap it
with 'if -r'.

[Michael Moss, 2014-01-22 22:38:37]

lgtm

[jungshik at Google, 2014-01-23 00:06:18]

On 2014/01/22 22:38:37, Michael Moss wrote:
> lgtm

Thanks. 

linux_chromeos failed content_unittests due to the abence of icudtl.dat, which
is a mystery because other unittests (base, net,) succeeded and icu.isolate is
included in content_unittests.isolate exactly the same way as included in
{base,net}_unittests.isolate. Moreover, linux_rel does not have that problem
even though both bots use runisolate*py.  

I'll talk to maruel about it.

[jungshik at Google, 2014-01-23 01:20:46]

linux_arm_crosscompile_cros also failed. ninja complained about multiple rules
for icudtl.dat. 

It looks like I'd better turn this on only on desktop_linux (but not on cros). 

------------------Error copied below ------------

Environment variables modified in compile.py:
  GOMA_API_KEY_FILE=/mnt/data/b/build/goma/goma.key
  NO_NACL_GOMA=true
 
PATH=/mnt/data/b/build/goma:/home/chrome-bot/slavebin:/mnt/data/b/depot_tools:/usr/bin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin


ninja -C /mnt/data/b/build/slave/linux_arm_cross_compile/build/src/out/Release
chromium_builder_tests -j50
ninja: Entering directory
`/mnt/data/b/build/slave/linux_arm_cross_compile/build/src/out/Release'
ninja: warning: multiple rules generate icudtl.dat. builds involving this target
will not be correct; continuing anyway
[1/21313] ACTION devtools_html: devtools_html_ba7b562d358c47af873b93ea32409e7f
[2/21313] ACTION Generating resources from ui_unscaled_resources.grd

[jungshik at Google, 2014-01-23 11:15:30]

I disabled icu_use_data_file_flag on CrOS for now. I'll get the latest patchset
landed.

[commit-bot: I haz the power, 2014-01-23 17:33:19]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jshin@chromium.org/102413007/180001

[commit-bot: I haz the power, 2014-01-23 20:38:54]

Change committed as 246678

[jungshik at Google, 2014-01-25 00:58:59]

Committed patchset #4 manually as r247045 (presubmit successful).

[jungshik at Google, 2014-01-25 01:02:04]

On 2014/01/25 00:58:59, Jungshik Shin wrote:
> Committed patchset #4 manually as r247045 (presubmit successful).

This will cut down the chrome-text size by 10MB. Perf_IMPROVEMENT from that size
reduction should NOT be regarded as a reason to revert. Its a good thing (TM)
:-)