Use third_party/custom_tabs_client as the src for customtabs library

Use third_party/custom_tabs_client as the src for customtabs library

-Remove chrome/android/java/src/android/support/customtabs
-Start using third_party/custom_tabs_client/src/customtabs as the source
for customtabs library.
-Update gyp and gn build files to make the above work
-Update chrome/browser/customtabs to use the support library classes
-Remove explicit usages of aidl classes.

This need an update and roll on third_party/custom_tabs_client to land

BUG=522099

[Yusuf, 2015-08-20 21:34:00]

yusufo@chromium.org changed reviewers:
+ lizeb@chromium.org, newt@chromium.org, pasko@chromium.org

[Yusuf, 2015-08-20 21:34:01]

newt@ for build system changes
pasko@ and lizeb@ for customtabs/

[newt (away), 2015-08-20 21:55:23]

gyp/gn files lgtm

[Benoit L, 2015-08-21 11:20:47]

Thank you for doing that!
Looks good, but can we wait a bit to land this? In particular, it would be nice
to land the fix for crbug.com/523057 first, since we need to cherry-pick it to
M45.

Or we can cherry-pick this CL, but it seems a bit large.
What do you think?

https://codereview.chromium.org/1303023002/diff/1/chrome/android/java/src/org...
File
chrome/android/java/src/org/chromium/chrome/browser/customtabs/CustomTabsConnection.java
(right):

https://codereview.chromium.org/1303023002/diff/1/chrome/android/java/src/org...
chrome/android/java/src/org/chromium/chrome/browser/customtabs/CustomTabsConnection.java:311:
public Bundle extraCommand(String commandName, Bundle args) {
This no longer overrides a method?

https://codereview.chromium.org/1303023002/diff/1/chrome/android/java/src/org...
chrome/android/java/src/org/chromium/chrome/browser/customtabs/CustomTabsConnection.java:573:
prerenderUrl(session, null, null); // Cancels the pre-render.
Has we have discovered recently, this needs to be posted to the UI thread.

https://codereview.chromium.org/1303023002/diff/1/third_party/custom_tabs_cli...
File third_party/custom_tabs_client/custom_tabs_client.gyp (right):

https://codereview.chromium.org/1303023002/diff/1/third_party/custom_tabs_cli...
third_party/custom_tabs_client/custom_tabs_client.gyp:44:
'../android_tools/android_tools.gyp:android_support_v13_javalib',
Stupid question: Doesn't this bloat the final .dex file?
Hopefully proguard should help us in this case, but does it run for debug
builds?

[pasko, 2015-08-21 11:54:57]

thanks a lot!

lgtm, and I agree with lizeb@ that cherry-picking can be significantly more
complex if we land this before the branchpoint, we already have one conflicting
change ..

on the other hand, if we wait too long, a lot of things will have to be redone
here, so I would shoot in the sky and say 'one week after branchpoint maybe'? :)

https://codereview.chromium.org/1303023002/diff/1/third_party/custom_tabs_cli...
File third_party/custom_tabs_client/custom_tabs_client.gyp (right):

https://codereview.chromium.org/1303023002/diff/1/third_party/custom_tabs_cli...
third_party/custom_tabs_client/custom_tabs_client.gyp:44:
'../android_tools/android_tools.gyp:android_support_v13_javalib',
On 2015/08/21 11:20:47, Benoit L wrote:
> Stupid question: Doesn't this bloat the final .dex file?
> Hopefully proguard should help us in this case, but does it run for debug
> builds?

content_java already depends on it :(

Still a confusing dependency, please write a detailed comment here what
problems/conflicts pop up if we don't do this.

[Benoit L, 2015-08-21 12:07:32]

Oh, right, I said "looks good", but not the magic word.
LGTM.

[Benoit L, 2015-08-21 12:20:58]

On 2015/08/21 12:07:32, Benoit L wrote:
> Oh, right, I said "looks good", but not the magic word.
> LGTM.

Actually, I think that we should land this after the (hopefully last) M45
cherry-pick, but before M46 branch point, to make the (probably non-existent...
hum) M46 cherry-picks easier.

[pasko, 2015-08-21 12:55:12]

On 2015/08/21 12:20:58, Benoit L wrote:
> On 2015/08/21 12:07:32, Benoit L wrote:
> > Oh, right, I said "looks good", but not the magic word.
> > LGTM.
> 
> Actually, I think that we should land this after the (hopefully last) M45
> cherry-pick, but before M46 branch point, to make the (probably
non-existent...
> hum) M46 cherry-picks easier.

I don't have a strong feeling about this. I trust your intuition here.