Use Clang's update.py --print-revision everywhere.

Use Clang's update.py --print-revision everywhere.

Since yesterday, it works on Windows. This allows us to not
completely clobber the build from update.py each time, since the
build system tracks the dependency instead.

It should also be possible to use this from gn, so we can delete the
posix-print-revision.py script. The only problem was that the
hack we use for fixing the output in gclient runhooks didn't work.
I've adjusted that to not fail when the hack doesn't work.

BUG=483065
R=scottmg@chromium.org, thakis@chromium.org

Committed: https://chromium.googlesource.com/chromium/src/+/cd20b07ab7edf2a7bab8e34362cebe5f92b41a27

[hans, 2015-05-06 18:54:50]

hans@chromium.org changed reviewers:
+ scottmg@chromium.org, thakis@chromium.org

[hans, 2015-05-06 18:54:51]

thakis: please take a look

scottmg: owners for gn stuff

[Nico, 2015-05-06 18:59:05]

https://codereview.chromium.org/1127233002/diff/20001/build/common.gypi
File build/common.gypi (right):

https://codereview.chromium.org/1127233002/diff/20001/build/common.gypi#newco...
build/common.gypi:2634: 'defines':
['CR_CLANG_REVISION=<!(<(DEPTH)/tools/clang/scripts/update.py
--print-revision)'],
Does this work? update.py always goes to svn, no? Or does it forward to
update.sh on non-win still?

[hans, 2015-05-06 19:32:15]

https://codereview.chromium.org/1127233002/diff/20001/build/common.gypi
File build/common.gypi (right):

https://codereview.chromium.org/1127233002/diff/20001/build/common.gypi#newco...
build/common.gypi:2634: 'defines':
['CR_CLANG_REVISION=<!(<(DEPTH)/tools/clang/scripts/update.py
--print-revision)'],
On 2015/05/06 18:59:05, Nico wrote:
> Does this work? update.py always goes to svn, no? Or does it forward to
> update.sh on non-win still?

It forwards on non-win.

[Nico, 2015-05-06 19:48:52]

lgtm then

[Nico, 2015-05-06 19:48:53]

lgtm then

[hans, 2015-05-06 19:55:21]

Scott, does the gn part look good to you?

[scottmg, 2015-05-06 19:59:03]

lgtm

[hans, 2015-05-06 19:59:50]

The CQ bit was checked by hans@chromium.org

[commit-bot: I haz the power, 2015-05-06 20:00:33]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1127233002/20001

[commit-bot: I haz the power, 2015-05-06 20:15:19]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-05-06 20:15:20]

Try jobs failed on following builders:
  win8_chromium_rel on tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win8_chromium_rel...)

[hans, 2015-05-06 20:21:27]

On 2015/05/06 20:15:20, I haz the power (commit-bot) wrote:
> Try jobs failed on following builders:
>   win8_chromium_rel on tryserver.chromium.win (JOB_FAILED,
>
http://build.chromium.org/p/tryserver.chromium.win/builders/win8_chromium_rel...)

Interesting...

       ['clang==1 or host_clang==1', {
         # This is here so that all files get recompiled after a clang roll and
         # when turning clang on or off.
         # (defines are passed via the command line, and build systems rebuild
         # things when their commandline changes). Nothing should ever read this
         # define.
         'defines':
['CR_CLANG_REVISION=<!(<(DEPTH)/tools/clang/scripts/update.py
--print-revision)'],

seems gyp runs the <!(command> even though it's guarded by clang==1 and it's not
going to use it?

I guess we can make update.py --print-revision return something benign when
clang=1 isn't set.

[hans, 2015-05-06 20:32:49]

On 2015/05/06 20:21:27, hans wrote:
> On 2015/05/06 20:15:20, I haz the power (commit-bot) wrote:
> > Try jobs failed on following builders:
> >   win8_chromium_rel on tryserver.chromium.win (JOB_FAILED,
> >
>
http://build.chromium.org/p/tryserver.chromium.win/builders/win8_chromium_rel...)
> 
> Interesting...
> 
>        ['clang==1 or host_clang==1', {
>          # This is here so that all files get recompiled after a clang roll
and
>          # when turning clang on or off.
>          # (defines are passed via the command line, and build systems rebuild
>          # things when their commandline changes). Nothing should ever read
this
>          # define.
>          'defines':
> ['CR_CLANG_REVISION=<!(<(DEPTH)/tools/clang/scripts/update.py
> --print-revision)'],
> 
> seems gyp runs the <!(command> even though it's guarded by clang==1 and it's
not
> going to use it?
> 
> I guess we can make update.py --print-revision return something benign when
> clang=1 isn't set.

Trying that.

[hans, 2015-05-06 20:32:54]

The CQ bit was checked by hans@chromium.org

[hans, 2015-05-06 20:32:55]

The patchset sent to the CQ was uploaded after l-g-t-m from thakis@chromium.org,
scottmg@chromium.org
Link to the patchset: https://codereview.chromium.org/1127233002/#ps40001
(title: "Make --print-version not fall over and die when there's no clang
checkout")

[commit-bot: I haz the power, 2015-05-06 20:34:14]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1127233002/40001

[commit-bot: I haz the power, 2015-05-06 20:49:49]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-05-06 20:49:50]

Try jobs failed on following builders:
  win8_chromium_rel on tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win8_chromium_rel...)

[hans, 2015-05-06 21:03:34]

The CQ bit was checked by hans@chromium.org

[hans, 2015-05-06 21:03:35]

The patchset sent to the CQ was uploaded after l-g-t-m from thakis@chromium.org,
scottmg@chromium.org
Link to the patchset: https://codereview.chromium.org/1127233002/#ps60001
(title: "Fix Windows for realz")

[commit-bot: I haz the power, 2015-05-06 21:04:30]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1127233002/60001

[commit-bot: I haz the power, 2015-05-06 22:39:03]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-05-06 22:40:33]

Patchset 4 (id:??) landed as
https://crrev.com/828bba82c5ed315cf073ad9599c0d5334a9a403d
Cr-Commit-Position: refs/heads/master@{#328634}

[Nico, 2015-05-07 01:14:25]

A revert of this CL (patchset #4 id:60001) has been created in
https://codereview.chromium.org/1131813002/ by thakis@chromium.org.

The reason for reverting is: Looks like this broke the world:
http://build.chromium.org/p/chromium.fyi/console?category=win%20clang.

[hans, 2015-05-07 01:43:50]

On 2015/05/07 01:14:25, Nico wrote:
> A revert of this CL (patchset #4 id:60001) has been created in
> https://codereview.chromium.org/1131813002/ by mailto:thakis@chromium.org.
> 
> The reason for reverting is: Looks like this broke the world:
> http://build.chromium.org/p/chromium.fyi/console?category=win%20clang.

Sorry, I'm not sure why it failed on the bots but not on my machine.

The script makes svn checkouts successfully on these machines, but it's always
run with 'shell=True', so it probably should in this instance as well. I've
uploaded a new patch that does this. OK to commit? I'll watch the bots like a
hawk.

[scottmg, 2015-05-07 02:06:51]

On 2015/05/07 01:43:50, hans wrote:
> On 2015/05/07 01:14:25, Nico wrote:
> > A revert of this CL (patchset #4 id:60001) has been created in
> > https://codereview.chromium.org/1131813002/ by mailto:thakis@chromium.org.
> > 
> > The reason for reverting is: Looks like this broke the world:
> > http://build.chromium.org/p/chromium.fyi/console?category=win%20clang.
> 
> Sorry, I'm not sure why it failed on the bots but not on my machine.
> 
> The script makes svn checkouts successfully on these machines, but it's always
> run with 'shell=True', so it probably should in this instance as well. I've
> uploaded a new patch that does this. OK to commit? I'll watch the bots like a
> hawk.

I'd recommend taking svn.exe out of your local PATH and testing with that. But
shell=True lgtm.

[hans, 2015-05-07 02:12:54]

On 2015/05/07 02:06:51, scottmg wrote:
> I'd recommend taking svn.exe out of your local PATH and testing with that. But
> shell=True lgtm.

Yeah, that's it. I had d:\src\depot_tools\svn_bin (which contains svn.exe) on my
PATH. If I remove that, the script fails as on the bot. When I add shell=True,
it works.

[commit-bot: I haz the power, 2015-05-07 03:50:22]

Patchset 5 (id:??) landed as
https://crrev.com/cd20b07ab7edf2a7bab8e34362cebe5f92b41a27
Cr-Commit-Position: refs/heads/master@{#328699}

[hans, 2015-05-07 03:50:43]

Committed patchset #5 (id:80001) manually as
cd20b07ab7edf2a7bab8e34362cebe5f92b41a27 (presubmit successful).

[hans, 2015-05-07 04:21:29]

The first bot made it through:
http://build.chromium.org/p/chromium.fyi/builders/CrWinClang64%28dll%29/build...

[Nico, 2015-05-10 05:43:48]

One side-effect that wasn't clear to me: The clang/win fyi bots no longer do
clobber builds if clang didn't change. That's not a problem, but it just got
very scared for a few minutes when I saw two bots rebuilding only 3k TUs between
builds. It looks like this was correct and working as intended; I just didn't
think of this being intentional immediately.

[hans, 2015-05-10 18:34:49]

On 2015/05/10 05:43:48, Nico wrote:
> One side-effect that wasn't clear to me: The clang/win fyi bots no longer do
> clobber builds if clang didn't change. That's not a problem, but it just got
> very scared for a few minutes when I saw two bots rebuilding only 3k TUs
between
> builds. It looks like this was correct and working as intended; I just didn't
> think of this being intentional immediately.

Right, and this matches the non-Windows Clang ToT builders. I think it's good
they're all mostly using the same scheme.