Fix stale mojom-generated Java problems with gyp builds

Fix stale mojom-generated Java problems with gyp builds

This causes any gyp targets which use the mojom bindings generator
to regenerate all outputs when any of the input mojom files have
changed. In such cases, the Java output directory is first wiped
out.

This avoids problems with stale mojom-generated Java files hanging
around and being compiled by build/android/javac.py, which tries
to build everything in its generated sources location.

BUG=461622
Committed: https://crrev.com/99b81f3496f7e5f9badf9504904da20545823d30
Cr-Commit-Position: refs/heads/master@{#318190}

[Ken Rockot(use gerrit already), 2015-02-26 01:34:54]

rockot@chromium.org changed reviewers:
+ cjhopman@chromium.org, jamesr@chromium.org

[Ken Rockot(use gerrit already), 2015-02-26 01:34:54]

Guys, here's my somewhat hacky solution (I mean... rmdir_and_stamp... really?)
for the gyp problem of stale generated Java. This CL only fixes mojom-generated
Java so I can unblock the mojo->chrome roll. Other targets would have to be
updated separately.

My gyp-fu is weak, so I'm open to suggestions if you see a cleaner low-effort
way to accomplish this.

Thanks

[cjhopman, 2015-02-26 02:15:37]

Yeah, this is probably about as good as it gets in gyp.

https://codereview.chromium.org/958773002/diff/40001/third_party/mojo/mojom_b...
File third_party/mojo/mojom_bindings_generator.gypi (right):

https://codereview.chromium.org/958773002/diff/40001/third_party/mojo/mojom_b...
third_party/mojo/mojom_bindings_generator.gypi:15: 'action_name':
'<(_target_name)_mojom_bindings_stamp',
this could probably use a really short comment about why:

# The java output directory is deleted to ensure that the java library doesn't
try to compile stale files.

https://codereview.chromium.org/958773002/diff/40001/third_party/mojo/mojom_b...
third_party/mojo/mojom_bindings_generator.gypi:76: 'variables': {
It looks like this gypi doesn't satisfy (1) from
https://code.google.com/p/chromium/issues/detail?id=461622#c8

it should have something like:

'additional_input_paths': [
  '<@(_sources)',  # or '<(stamp_filename)'
  '<@(mojom_bindings_generator_sources)',
]

I think that, without that, changing the mojom won't actually trigger rebuilding
the java library that it's a part of (it should be pretty easy to confirm that).

Actually, I think that gyp now supports depfiles for actions and so we should
use those for the gyp actions (they are already used when using the scripts from
gn) and remove the requirement that inputs need to be added here. I've filed
http://crbug.com/462075 for that.

[Ken Rockot(use gerrit already), 2015-02-26 02:36:16]

https://codereview.chromium.org/958773002/diff/40001/third_party/mojo/mojom_b...
File third_party/mojo/mojom_bindings_generator.gypi (right):

https://codereview.chromium.org/958773002/diff/40001/third_party/mojo/mojom_b...
third_party/mojo/mojom_bindings_generator.gypi:15: 'action_name':
'<(_target_name)_mojom_bindings_stamp',
On 2015/02/26 02:15:36, cjhopman wrote:
> this could probably use a really short comment about why:
> 
> # The java output directory is deleted to ensure that the java library doesn't
> try to compile stale files.

Done.

https://codereview.chromium.org/958773002/diff/40001/third_party/mojo/mojom_b...
third_party/mojo/mojom_bindings_generator.gypi:76: 'variables': {
On 2015/02/26 02:15:37, cjhopman wrote:
> It looks like this gypi doesn't satisfy (1) from
> https://code.google.com/p/chromium/issues/detail?id=461622#c8
> 
> it should have something like:
> 
> 'additional_input_paths': [
>   '<@(_sources)',  # or '<(stamp_filename)'
>   '<@(mojom_bindings_generator_sources)',
> ]
> 
> I think that, without that, changing the mojom won't actually trigger
rebuilding
> the java library that it's a part of (it should be pretty easy to confirm
that).
> 
> Actually, I think that gyp now supports depfiles for actions and so we should
> use those for the gyp actions (they are already used when using the scripts
from
> gn) and remove the requirement that inputs need to be added here. I've filed
> http://crbug.com/462075 for that.

Ah yeah, you're right. I've added additional_input_sources for now, assuming
depfiles use will be addressed in some future CL(s).

[cjhopman, 2015-02-26 02:38:39]

lgtm

https://codereview.chromium.org/958773002/diff/60001/third_party/mojo/mojom_b...
File third_party/mojo/mojom_bindings_generator.gypi (right):

https://codereview.chromium.org/958773002/diff/60001/third_party/mojo/mojom_b...
third_party/mojo/mojom_bindings_generator.gypi:83: 'additional_input_paths': [
this needs to go in the variables block with generated_src_dirs

[Ken Rockot(use gerrit already), 2015-02-26 02:46:10]

fixed. thanks!

[jamesr, 2015-02-26 03:05:04]

rslgtm

[jamesr, 2015-02-26 03:05:09]

rslgtm

[jamesr, 2015-02-26 03:06:04]

On 2015/02/26 03:05:09, jamesr wrote:
> rslgtm

rs lgtm

(Is that the magic?)

[Ken Rockot(use gerrit already), 2015-02-26 03:28:19]

The CQ bit was checked by rockot@chromium.org

[Ken Rockot(use gerrit already), 2015-02-26 03:28:20]

The patchset sent to the CQ was uploaded after l-g-t-m from
cjhopman@chromium.org
Link to the patchset: https://codereview.chromium.org/958773002/#ps100001
(title: "actually set additional_input_paths correctly -_-")

[Ken Rockot(use gerrit already), 2015-02-26 03:29:23]

The CQ bit was unchecked by rockot@chromium.org

[Ken Rockot(use gerrit already), 2015-02-26 03:30:18]

The CQ bit was checked by rockot@chromium.org

[commit-bot: I haz the power, 2015-02-26 03:31:44]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/958773002/100001

[commit-bot: I haz the power, 2015-02-26 05:20:42]

Committed patchset #6 (id:100001)

[commit-bot: I haz the power, 2015-02-26 05:21:09]

Patchset 6 (id:??) landed as
https://crrev.com/99b81f3496f7e5f9badf9504904da20545823d30
Cr-Commit-Position: refs/heads/master@{#318190}