bluetooth: android: Add java build dependencies.

bluetooth: android: Add java build dependencies.

ChromePublic.apk when built from internal Chrome code failed to include
these dependencies, and would encounter runtime errors with
ClassNotFoundException.

BUG=490430
Committed: https://crrev.com/3a7891028f1357bc869a9495df6bff3da2944265
Cr-Commit-Position: refs/heads/master@{#338534}

[scheib, 2015-07-09 23:28:28]

scheib@chromium.org changed reviewers:
+ newt@chromium.org

[newt (away), 2015-07-09 23:50:35]

newt@chromium.org changed reviewers:
+ cjhopman@chromium.org

[newt (away), 2015-07-09 23:50:36]

+cjhopman: here's a weird build situation. C++ code in content, depends on C++
code in device/bluetooth, which needs access to Java code in device/bluetooth at
runtime. For some reason, even though content_browser depends on
device_bluetooth which depends on device_bluetooth_java, the bluetooth Java
files aren't included in the APK. This CL fixes that by adding a dependency from
content_java to device_bluetooth_java, which causes the bluetooth Java files to
be included in the APK. Voila.

However, Java code in content doesn't actually reference Java code in
device/bluetooth. So, is this the proper fix? Does our build system simply not
handle cases where an isolated island of Java code is depended upon by only a
C++ target?

[cjhopman, 2015-07-10 02:26:21]

On 2015/07/09 23:50:36, newt wrote:
> +cjhopman: here's a weird build situation. C++ code in content, depends on C++
> code in device/bluetooth, which needs access to Java code in device/bluetooth
at
> runtime. For some reason, even though content_browser depends on
> device_bluetooth which depends on device_bluetooth_java, the bluetooth Java
> files aren't included in the APK. This CL fixes that by adding a dependency
from
> content_java to device_bluetooth_java, which causes the bluetooth Java files
to
> be included in the APK. Voila.
> 
> However, Java code in content doesn't actually reference Java code in
> device/bluetooth. So, is this the proper fix? Does our build system simply not
> handle cases where an isolated island of Java code is depended upon by only a
> C++ target?

Hm, internal code does a "trick" to remove the dependency from the apk to the
native code. It ensures that the corresponding native code is included in the
apk, but doesn't handle any other dependencies inherited from that. Adding this
dependency seems correct.

[cjhopman, 2015-07-10 02:27:21]

On 2015/07/10 02:26:21, cjhopman wrote:
> On 2015/07/09 23:50:36, newt wrote:
> > +cjhopman: here's a weird build situation. C++ code in content, depends on
C++
> > code in device/bluetooth, which needs access to Java code in
device/bluetooth
> at
> > runtime. For some reason, even though content_browser depends on
> > device_bluetooth which depends on device_bluetooth_java, the bluetooth Java
> > files aren't included in the APK. This CL fixes that by adding a dependency
> from
> > content_java to device_bluetooth_java, which causes the bluetooth Java files
> to
> > be included in the APK. Voila.
> > 
> > However, Java code in content doesn't actually reference Java code in
> > device/bluetooth. So, is this the proper fix? Does our build system simply
not
> > handle cases where an isolated island of Java code is depended upon by only
a
> > C++ target?
> 
> Hm, internal code does a "trick" to remove the dependency from the apk to the
> native code. It ensures that the corresponding native code is included in the
> apk, but doesn't handle any other dependencies inherited from that. Adding
this
> dependency seems correct.

Also, having the dependency in the java part of the dependency graph (like this
change adds) is required in gn as java dependencies aren't transitively acquired
through native targets.

lgtm.

[scheib, 2015-07-10 15:46:30]

scheib@chromium.org changed reviewers:
+ tedchoc@chromium.org

[scheib, 2015-07-10 15:46:30]

tedchoc OWNERS review please.

[Ted C, 2015-07-10 16:14:04]

lgtm

[scheib, 2015-07-10 17:41:18]

The CQ bit was checked by scheib@chromium.org

[scheib, 2015-07-10 17:41:19]

The CQ bit was unchecked by scheib@chromium.org

[scheib, 2015-07-10 17:42:19]

scheib@chromium.org changed reviewers:
+ jam@chromium.org

[scheib, 2015-07-10 17:42:20]

jam content/content.gyp

[jam, 2015-07-13 15:55:43]

lgtm

[scheib, 2015-07-13 16:56:12]

The CQ bit was checked by scheib@chromium.org

[commit-bot: I haz the power, 2015-07-13 16:56:36]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1231883004/20001

[commit-bot: I haz the power, 2015-07-13 18:24:27]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-07-13 18:26:29]

Patchset 2 (id:??) landed as
https://crrev.com/3a7891028f1357bc869a9495df6bff3da2944265
Cr-Commit-Position: refs/heads/master@{#338534}