1879132 - Upgrade AGP to latest version (8.3.0)

Status: RESOLVED FIXED

(Firefox for Android :: General, task, P2)

Description

[Mihai Adrian Carare [:mcarare]]

This is needed for upgrading Compose to 1.6.0.
At the moment AGP 8.3.0 is not released yet.

Comment 1

[BMO Github Automation]

Attachment: [mozilla-mobile/firefox-android] Bug 1879132 - Upgrade AGP to 8.3.0 version. (#5495)

Comment 2

[Ryan VanderMeulen [:RyanVM]]

It's released now.

Comment 3

[BMO Github Automation]

Authored by https://github.com/mcarare
https://github.com/mozilla-mobile/firefox-android/commit/9db0adfe4ad016c9aa09cb4c0d11581e4465c3e8
[main] Bug 1879132 - Upgrade AGP to 8.3.0 version.

Authored by https://github.com/mcarare
https://github.com/mozilla-mobile/firefox-android/commit/dfcb84dafee347edd1c351d6651e5fe50e7bf571
[main] Bug 1879132 - Use SdkSuppress instead of RequiresApi in tests.

Comment 4

[BMO Github Automation]

Authored by https://github.com/rvandermeulen
https://github.com/mozilla-mobile/firefox-android/commit/6e68351872d08075272be21f3aa397a5b12b989b
[main] Revert "Bug 1879132 - Upgrade AGP to 8.3.0 version." for likely causing bug 1884359

Comment 5

[Ryan VanderMeulen [:RyanVM]]

I can reproduce the android-startup-test-fenix-arm-nightly-robo-opt crashes on Try with the AGP bump and can confirm that setting android.enableNewResourceShrinker.preciseShrinking to false fixes them. I guess we still may want to figure out why that particular resource is being optimized away in the first place, but we can at least land the upgrade with that change as a workaround for the time-being and unblock the Compose update.
https://treeherder.mozilla.org/jobs?repo=try&revision=7fec6e76a2993549c2126ecb8e640f6f0e727631&group_state=expanded

Comment 6

[Ryan VanderMeulen [:RyanVM]]

Attachment: Bug 1879132- Upgrade AGP to version 8.3.0. r=mcarare

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/b9c4f8219055
Upgrade AGP to version 8.3.0. r=mcarare

Comment 8

[Ryan VanderMeulen [:RyanVM]]

Should we file a follow-up for investigating why that new optimization broke Fenix builds? I guess the longer-term concern is if they eventually remove the ability to disable it in some future release.

Comment 9

[Narcis Beleuzu [:NarcisB]]

https://hg.mozilla.org/mozilla-central/rev/b9c4f8219055

Comment 10

[Mihai Adrian Carare [:mcarare]]

I think we should. I am also considering making a minimal reproducible example to open an issue for Google to investigate.