Fix javac command never caching input .md5s

Fix javac command never caching input .md5s

This was broken over a year ago by http://crrev.com/a78eb479

This change also removes --classes-dir (which is currently unused), and makes --jar-path (always used) required.

It also forces the command to run if the output is missing (rather than relying completely on the .stamp file)

On my ubiquity machine:
Add a log statement to a file in base_java then build chrome_apk:

Before:
real    2m25.188s
user    4m52.950s
sys     1m10.064s

After:
real    1m16.989s
user    1m13.795s
sys     0m44.040s

BUG=523420, 379190
Committed: https://crrev.com/1c366880cb68fbf25050f1c6d78684eb6ae2e644
Cr-Commit-Position: refs/heads/master@{#346975}

[agrieve, 2015-08-21 20:36:48]

agrieve@chromium.org changed reviewers:
+ jbudorick@chromium.org

[agrieve, 2015-08-21 20:37:42]

On 2015/08/21 20:36:48, agrieve wrote:
> mailto:agrieve@chromium.org changed reviewers:
> + mailto:jbudorick@chromium.org

Note: I think there's still an improvement to be had, as I'm pretty sure some
.jars are still being rebuilt unnecessarily.  Suspect those with generated
srcjars. Will investigate that next.

[jbudorick, 2015-08-23 02:50:51]

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py
File build/android/gyp/javac.py (left):

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py#...
build/android/gyp/javac.py:122: record_path = os.path.join(classes_dir,
'javac.md5.stamp')
This is the fix, right?

It's kinda hard to tell with all of the shuffling you've done in here :/

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py
File build/android/gyp/javac.py (right):

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py#...
build/android/gyp/javac.py:71: ret = []
return [f for f in paths if any(fnmatch.fnmatch(f, i) for i in javac_includes)]

?

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py#...
build/android/gyp/javac.py:147: entries = map(lambda e: e.split(":"),
manifest_entry)
entries = [e.split(':') for e in manifest_entry]

[agrieve, 2015-08-26 05:52:58]

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py
File build/android/gyp/javac.py (left):

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py#...
build/android/gyp/javac.py:122: record_path = os.path.join(classes_dir,
'javac.md5.stamp')
On 2015/08/23 02:50:50, jbudorick wrote:
> This is the fix, right?
> 
> It's kinda hard to tell with all of the shuffling you've done in here :/

Yes, it's the crux of it. The script shouldn't be creating a temporary directory
and extracting srcjars if nothing has changed though, so that's the reason for
the re-jigger. Arguably could have split that out...

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py
File build/android/gyp/javac.py (right):

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py#...
build/android/gyp/javac.py:71: ret = []
On 2015/08/23 02:50:51, jbudorick wrote:
> return [f for f in paths if any(fnmatch.fnmatch(f, i) for i in
javac_includes)]
> 
> ?

Done (and moved to build_utils.py since it was copy&pasted there as well)

https://codereview.chromium.org/1308083002/diff/1/build/android/gyp/javac.py#...
build/android/gyp/javac.py:147: entries = map(lambda e: e.split(":"),
manifest_entry)
On 2015/08/23 02:50:51, jbudorick wrote:
> entries = [e.split(':') for e in manifest_entry]

Done.

[agrieve, 2015-09-01 14:59:57]

agrieve@chromium.org changed reviewers:
+ dpranke@google.com

[jbudorick, 2015-09-01 16:28:08]

jbudorick@chromium.org changed reviewers:
- dpranke@google.com

[jbudorick, 2015-09-01 16:28:09]

https://codereview.chromium.org/1308083002/diff/60001/build/android/gyp/util/...
File build/android/gyp/util/build_utils.py (right):

https://codereview.chromium.org/1308083002/diff/60001/build/android/gyp/util/...
build/android/gyp/util/build_utils.py:240: return filters and
any(fnmatch.fnmatch(path, f) for f in filters)
I don't think this is the same as Allow, which would return True if
exclude_patterns was None. Is that intentional? (It doesn't look like it, but
I'm not sure.) If not, this should be

  return not filters or any(...)

[agrieve, 2015-09-01 16:58:18]

https://codereview.chromium.org/1308083002/diff/60001/build/android/gyp/util/...
File build/android/gyp/util/build_utils.py (right):

https://codereview.chromium.org/1308083002/diff/60001/build/android/gyp/util/...
build/android/gyp/util/build_utils.py:240: return filters and
any(fnmatch.fnmatch(path, f) for f in filters)
On 2015/09/01 16:28:09, jbudorick wrote:
> I don't think this is the same as Allow, which would return True if
> exclude_patterns was None. Is that intentional? (It doesn't look like it, but
> I'm not sure.) If not, this should be
> 
>   return not filters or any(...)

It is intentionally different, in that it simplifies the usages a bit. It's
still the case in MergeZips that no filters results in the action happening.

[jbudorick, 2015-09-02 01:18:16]

lgtm

[agrieve, 2015-09-02 15:02:49]

The CQ bit was checked by agrieve@chromium.org

[agrieve, 2015-09-02 15:02:50]

The patchset sent to the CQ was uploaded after l-g-t-m from
jbudorick@chromium.org
Link to the patchset: https://codereview.chromium.org/1308083002/#ps80001
(title: "rebase")

[commit-bot: I haz the power, 2015-09-02 15:02:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1308083002/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1308083002/80001

[commit-bot: I haz the power, 2015-09-02 15:19:54]

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

[commit-bot: I haz the power, 2015-09-02 15:19:55]

Try jobs failed on following builders:
  android_chromium_gn_compile_dbg on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/android_chromiu...)

[agrieve, 2015-09-02 16:33:56]

The CQ bit was checked by agrieve@chromium.org

[agrieve, 2015-09-02 16:33:57]

The patchset sent to the CQ was uploaded after l-g-t-m from
jbudorick@chromium.org
Link to the patchset: https://codereview.chromium.org/1308083002/#ps100001
(title: "empty libs")

[commit-bot: I haz the power, 2015-09-02 16:34:24]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1308083002/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1308083002/100001

[commit-bot: I haz the power, 2015-09-02 16:51:43]

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

[commit-bot: I haz the power, 2015-09-02 16:51:44]

Try jobs failed on following builders:
  linux_android_rel_ng on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_android_r...)

[agrieve, 2015-09-02 17:11:53]

The CQ bit was checked by agrieve@chromium.org

[agrieve, 2015-09-02 17:11:53]

The patchset sent to the CQ was uploaded after l-g-t-m from
jbudorick@chromium.org
Link to the patchset: https://codereview.chromium.org/1308083002/#ps140001
(title: "remove flag it actually is unused like i thought before")

[commit-bot: I haz the power, 2015-09-02 17:12:07]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1308083002/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1308083002/140001

[commit-bot: I haz the power, 2015-09-02 18:33:25]

Committed patchset #8 (id:140001)

[commit-bot: I haz the power, 2015-09-02 18:34:08]

Patchset 8 (id:??) landed as
https://crrev.com/1c366880cb68fbf25050f1c6d78684eb6ae2e644
Cr-Commit-Position: refs/heads/master@{#346975}

[agrieve, 2015-09-02 20:30:54]

A revert of this CL (patchset #8 id:140001) has been created in
https://codereview.chromium.org/1309663005/ by agrieve@chromium.org.

The reason for reverting is: Causing try-bot failures:

e.g.:
http://build.chromium.org/p/tryserver.chromium.linux/builders/android_chromiu....