[Bindings] Remove aggregation of generated binding code

Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.

BUG=634231, 388172
Committed: https://crrev.com/134a59bd48149d721ce06554570db1b2b5bdee55
Cr-Commit-Position: refs/heads/master@{#417216}

[peria, 2016-09-07 07:53:52]

Patchset #1 (id:1) has been deleted

[peria, 2016-09-07 08:02:35]

Description was changed from

==========
Remove aggregation of generated binding code

BUG=634231
==========

to

==========
Remove aggregation of generated binding code

With this change, it is expected to be able to update
IDL compiler to generate better source code
(e.g. http://crrev.com/332663006/) and each compile
can be run in parallel.


BUG=634231, 388172
==========

[peria, 2016-09-07 08:08:02]

Description was changed from

==========
Remove aggregation of generated binding code

With this change, it is expected to be able to update
IDL compiler to generate better source code
(e.g. http://crrev.com/332663006/) and each compile
can be run in parallel.


BUG=634231, 388172
==========

to

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.


BUG=634231, 388172
==========

[peria, 2016-09-07 08:08:54]

peria@chromium.org changed reviewers:
+ bashi@chromium.org, yukishiino@chromium.org

[peria, 2016-09-07 08:08:54]

PTL

[haraken, 2016-09-07 08:15:21]

Does this really improve the compile time? Historically we introduced the
aggregated files because it was too heavy to spawn one process per IDL
interface. Maybe you want to measure the time taken to compile all IDL
interfaces.

That said, I'm okay with making this change because the benefit of reducing # of
compiled files by getting rid of false-positive dependencies will be larger than
the cost of increasing the compile time per interface.

[peria, 2016-09-07 09:53:04]

On 2016/09/07 08:15:21, haraken wrote:
> Does this really improve the compile time? Historically we introduced the
> aggregated files because it was too heavy to spawn one process per IDL
> interface. Maybe you want to measure the time taken to compile all IDL
> interfaces.
> 
> That said, I'm okay with making this change because the benefit of reducing #
of
> compiled files by getting rid of false-positive dependencies will be larger
than
> the cost of increasing the compile time per interface.

Yes, this CL reduced about 30% of elapsed time in full-build on my machines,
and increased CPU time as you worried.
I'll update the description to show build time.

[peria, 2016-09-07 10:01:29]

Description was changed from

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.


BUG=634231, 388172
==========

to

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel
and to reduce actual build time.


On my local machine, full-build took
37m37s --> 23m38s  on Windows(Z420)
34m24s --> 14m30s  on MacOSX

BUG=634231, 388172
==========

[peria, 2016-09-07 10:02:03]

Description was changed from

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel
and to reduce actual build time.


On my local machine, full-build took
37m37s --> 23m38s  on Windows(Z420)
34m24s --> 14m30s  on MacOSX

BUG=634231, 388172
==========

to

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel
and to reduce actual build time.


On my local machine, full-build using Goma took
37m37s --> 23m38s  on Windows(Z420)
34m24s --> 14m30s  on MacOSX

BUG=634231, 388172
==========

[Yuki, 2016-09-07 10:11:02]

LGTM from my perspectives.

[peria, 2016-09-07 10:14:37]

Description was changed from

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel
and to reduce actual build time.


On my local machine, full-build using Goma took
37m37s --> 23m38s  on Windows(Z420)
34m24s --> 14m30s  on MacOSX

BUG=634231, 388172
==========

to

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel
and to reduce actual build time.



BUG=634231, 388172
==========

[peria, 2016-09-07 10:18:54]

Description was changed from

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel
and to reduce actual build time.



BUG=634231, 388172
==========

to

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.


BUG=634231, 388172
==========

[peria, 2016-09-07 10:21:42]

On 2016/09/07 09:53:04, peria wrote:
> On 2016/09/07 08:15:21, haraken wrote:
> > Does this really improve the compile time? Historically we introduced the
> > aggregated files because it was too heavy to spawn one process per IDL
> > interface. Maybe you want to measure the time taken to compile all IDL
> > interfaces.
> > 
> > That said, I'm okay with making this change because the benefit of reducing
#
> of
> > compiled files by getting rid of false-positive dependencies will be larger
> than
> > the cost of increasing the compile time per interface.
> 
> Yes, this CL reduced about 30% of elapsed time in full-build on my machines,
> and increased CPU time as you worried.

This measurement was wrong. the reduction was caused by cache on goma.
Elapsed compile time does not change.

[haraken, 2016-09-07 15:12:21]

On 2016/09/07 10:21:42, peria wrote:
> On 2016/09/07 09:53:04, peria wrote:
> > On 2016/09/07 08:15:21, haraken wrote:
> > > Does this really improve the compile time? Historically we introduced the
> > > aggregated files because it was too heavy to spawn one process per IDL
> > > interface. Maybe you want to measure the time taken to compile all IDL
> > > interfaces.
> > > 
> > > That said, I'm okay with making this change because the benefit of
reducing
> #
> > of
> > > compiled files by getting rid of false-positive dependencies will be
larger
> > than
> > > the cost of increasing the compile time per interface.
> > 
> > Yes, this CL reduced about 30% of elapsed time in full-build on my machines,
> > and increased CPU time as you worried.
> 
> This measurement was wrong. the reduction was caused by cache on goma.
> Elapsed compile time does not change.

LGTM

[peria, 2016-09-07 16:11:07]

Description was changed from

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.


BUG=634231, 388172
==========

to

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.

Note: this change strongly depends on GN, and it breaks
builds with GYP.


BUG=634231, 388172
==========

[bashi, 2016-09-07 22:39:16]

lgtm

[peria, 2016-09-08 01:26:18]

The CQ bit was checked by peria@chromium.org

[commit-bot: I haz the power, 2016-09-08 01:26:40]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-08 01:29:03]

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

[commit-bot: I haz the power, 2016-09-08 01:29:04]

Try jobs failed on following builders:
  mac_chromium_rel_ng on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_...)

[peria, 2016-09-08 01:30:02]

The CQ bit was unchecked by peria@chromium.org

[peria, 2016-09-08 02:03:57]

Description was changed from

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.

Note: this change strongly depends on GN, and it breaks
builds with GYP.


BUG=634231, 388172
==========

to

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.



BUG=634231, 388172
==========

[peria, 2016-09-08 02:06:55]

The CQ bit was checked by peria@chromium.org

[peria, 2016-09-08 02:06:55]

The patchset sent to the CQ was uploaded after l-g-t-m from
haraken@chromium.org, bashi@chromium.org, yukishiino@chromium.org
Link to the patchset: https://codereview.chromium.org/2318933002/#ps60001
(title: "Fix import order")

[commit-bot: I haz the power, 2016-09-08 02:07:24]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-08 03:09:35]

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

[commit-bot: I haz the power, 2016-09-08 03:09:36]

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

[peria, 2016-09-08 03:34:37]

The CQ bit was checked by peria@chromium.org

[commit-bot: I haz the power, 2016-09-08 03:34:57]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-09-08 07:15:01]

Description was changed from

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.



BUG=634231, 388172
==========

to

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.



BUG=634231, 388172
==========

[commit-bot: I haz the power, 2016-09-08 07:15:02]

Committed patchset #3 (id:60001)

[commit-bot: I haz the power, 2016-09-08 07:18:29]

Description was changed from

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.



BUG=634231, 388172
==========

to

==========
Remove aggregation of generated binding code.

After this change, we maybe able to skip needless compiles
based on redundant dependencies. (issue 388172)
Plus, it is expected to compile those files in parallel.

BUG=634231, 388172

Committed: https://crrev.com/134a59bd48149d721ce06554570db1b2b5bdee55
Cr-Commit-Position: refs/heads/master@{#417216}
==========

[commit-bot: I haz the power, 2016-09-08 07:18:30]

Patchset 3 (id:??) landed as
https://crrev.com/134a59bd48149d721ce06554570db1b2b5bdee55
Cr-Commit-Position: refs/heads/master@{#417216}

[peria, 2016-09-08 08:30:28]

A revert of this CL (patchset #3 id:60001) has been created in
https://codereview.chromium.org/2325583002/ by peria@chromium.org.

The reason for reverting is: Build break
http://build.chromium.org/p/chromium.chrome/builders/Google%20Chrome%20Win/bu....