Android: fix Ti.UI.overrideUserInterfaceStyle

[drauggres]

JIRA: https://jira.appcelerator.org/browse/TIMOB-28563
fix #13266

Known issues:

• TiGradientDrawable: doesn't have a proxy reference
• TiAnimationBuilder: proxy is assinged in start() methond, i.e. after parameters calculation.

In both cases classes could be refactored, but I'm not going to do it now.
As simple workaround it is possible to use tiApp.getRootOrCurrentActivity() instead of TiApplication.getInstance(). In most cases this will give the desired result, but is not really correct, since it's possible that a proxy will have different Activity from current.

If someone will be thinking about merging this, please consider refactoring classes mentioned above or at least creating issues, so this don't get lost.

[drauggres]

We need to use Activity (its Configuration) stored in a Proxy when we resolve colors from a semantic-name string.
Now in the SDK we use "global" Application context instead, which is wrong.

I changed behaviour for some basic cases as a PoC.
The same must be done for each Proxy and each color-related property (will try to finish next week).

@hansemannn

[build]

	Fails
🚫	🔬 There are library changes, but no changes to the unit tests. That's OK as long as you're refactoring existing code, but will require an admin to merge this PR. Please see README.md#unit-tests for docs on unit testing.

	Messages
📖	✊ The commits in this PR match our conventions! Feel free to Rebase and Merge this PR when ready.
📖	✅ All tests are passing Nice one! All 19694 tests are passing. (There are 1157 skipped tests not included in that total)
📖	🎉 Another contribution from our awesome community member, drauggres! Thanks again for helping us make Titanium SDK better. 👍
📖	💾 Here's the generated SDK zipfile.

Generated by 🚫 dangerJS against 315d018

[drauggres]

Apparently, failed tests are not related to changes in this PR.
Same for example here: #13260

[hansemannn]

Thats pretty cool, thank you @drauggres! I did not expect to see any progress regarding this, wow! Will try to test in over the next weeks!

[hansemannn]

@drauggres Can you please pull from master again? The build will then succeed :)

[hansemannn]

There is one new merge conflict because of the recent separator-color changes.

[drauggres]

There is one new merge conflict because of the recent separator-color changes.

rebased and fixed

[hansemannn]

Thank you! But the PR looks much larger now. And kind of mixed with #13273, is that possible? Due to the force-push, this is not clearly visible anymore.

[drauggres]

But the PR looks much larger now.

It is not. This is what it has before update : 315d018. And this is what it has now: 8608c6a

And kind of mixed with #13273, is that possible?

It is other way around: #13273 includes this PR (and #13265). I mentioned it in the description.

[drauggres]

In this PR I had to update every single line where color is assigned and add an Activity reference.
In #13273 I had to update the same lines, and add support for Ti.UI.Color instance in arguments.

[hansemannn]

Understood, thank you. Didn't mean to be rude here 🧐.

[drauggres]

Understood, thank you. Didn't mean to be rude here monocle_face.

What? When? You wasn't rude. Was I?
I just tried to state the facts straight.

[hansemannn]

Looks fine from my end. @m1ga ?

[m1ga]

It adds another function to the helper so parseColor stays the same in case modules use it, so we are safe there 👍
And most of the other changes are just using the new method. All tests passed so that awesome too 👍

I will checkout the PR later today and give it a test run

[m1ga]

@hansemannn

Using the test app from the ticket I've tested:

• starting in dark or light mode -> shows correct mode ✔️
• overriding the mode works fine ✔️
• removing the override shows the correct system mode again ✔️

Tested my existing app ✔️ :

• no issues in showing the correct colors
• changing the system dark/light mode updated the app without any issue
• no issues using different android modules (like FCM, ti.animation, map)

Side note:
not having a semantic.json file BUT using label colors will result in some hiccups (#13266 (comment)) but it is not related to this PR and is the same in the current GA too. It was of course my fault since I didn't include the file. Maybe we can investigate in a different issue if there is a way to prevent that. Adding it as a default file (#13241) should be one step closer 😄