Update to Android Gradle Plugin 9.0.1 by inktomi · Pull Request #481 · runningcode/fladle

inktomi · 2026-03-02T21:55:02Z

Summary

Update to Android Gradle Plugin 9.0.1, resolving the "Could not find AppExtension error when newDsl=true" issue (the default in AGP 9.0).

This PR contains only AGP 9 functional migration changes. Code formatting/cleanup changes are in #482.

Changes

Migrate from legacy variant API (AppExtension, TestedExtension, BaseVariant) to new variant API (ApplicationExtension, ApplicationAndroidComponentsExtension, Variant)
Replace testVariants.configureEach with onVariants callbacks for APK path detection
Add VariantApkInfo data class to capture variant info during configuration phase
Update Gradle wrapper from 8.14.3 to 9.1.0 (minimum required by AGP 9.0)
Update Kotlin language/API version from 1.7 to 2.0
Remove kotlin-android plugin from sample projects (built-in in AGP 9.0)
Add <T : Any> type bounds for Kotlin 2.0 compatibility
Change task classes from open to abstract (AGP 9 requirement)
Update minimum Gradle version from 7.3 to 9.1
Applied Android Studio JVM migration (foojay toolchain resolution plugin)

Closes #478

Test plan

./gradlew :fladle-plugin:check — all tests pass, lint clean
./gradlew assembleDebug assembleDebugAndroidTest printYml — sample projects build with correct YAML output
./gradlew :sample-flavors-kotlin:printYml — flavor-specific APK paths resolve correctly
./gradlew printYml --configuration-cache — configuration cache works

inktomi · 2026-03-02T22:40:20Z

The workflow passes on my local fork: https://github.com/inktomi/fladle/actions/runs/22598741034/job/65475320428

inktomi · 2026-03-03T00:01:38Z

As part two of this, I'm working on reworking the fulladle plugin to support config caching. I will open that PR once this is reviewed, or ideally approved - I didn't want to work on it using old gradle APIs. I am introducing a new Settings plugin that people will have to apply so that we can get information about the modules without breaking caches - we can discuss more on that PR when the time comes.

emartynov

Super!

I would really ask you to split this into two PRs:

AGP 9 support
Code cleanup and improvements

Also not sure about hard cut of the lower supported version 9.1.0. I would probably at least support anything 9.x. And not sure about folks that are still on AGP 8.x.

inktomi · 2026-03-04T19:10:57Z

Happy to do so! Thank you!

…

On March 4, 2026 9:04:06 AM PST, Eugen Martynov ***@***.***> wrote: @emartynov commented on this pull request. Super! I would really ask you to split this into two PRs: 1. AGP 9 support 2. Code cleanup and improvements Also not sure about hard cut of the lower supported version 9.1.0. I would probably at least support anything 9.x. And not sure about folks that are still on AGP 8.x. -- Reply to this email directly or view it on GitHub: #481 (review) You are receiving this because you authored the thread. Message ID: ***@***.***>

- Migrate from legacy variant API (AppExtension, TestedExtension, BaseVariant) to new variant API (ApplicationExtension, ApplicationAndroidComponentsExtension, Variant) - Replace testVariants.configureEach with onVariants callbacks for APK path detection - Add VariantApkInfo data class to capture variant info during configuration - Update Gradle wrapper from 8.14.3 to 9.1.0 (minimum required by AGP 9.0) - Update Kotlin language/API version from 1.7 to 2.0 - Remove kotlin-android plugin from sample projects (built-in in AGP 9.0) - Add <T : Any> type bounds for Kotlin 2.0 compatibility - Change task classes from open to abstract (AGP 9 requirement) - Update minimum Gradle version from 7.3 to 9.1 Fixes runningcode#478

inktomi · 2026-03-04T21:02:34Z

The build fails on my fork - https://github.com/inktomi/fladle/actions/runs/22688439185

But it's because of issues that would be fixed by running the formatters - that's on the other branch now.

inktomi · 2026-03-04T21:03:45Z

gradle/wrapper/gradle-wrapper.properties

 distributionBase=GRADLE_USER_HOME
 distributionPath=wrapper/dists
-distributionUrl=https\://services.gradle.org/distributions/gradle-8.14.3-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-9.1.0-bin.zip


9.1 is the min compatible version for AGP 9.0.1

https://developer.android.com/build/releases/about-agp#updating-gradle

My bad - I was mixing AGP version with Gradle version

The min gradle version is also changed; it was good to call out for visibility! AGP doesn't give much flexibility here.

Thank you. Did you update the wrapper task as well to go along with this change? run ./gradlew wrapper twice to make sure.

See the "tip" here about running it "again": https://docs.gradle.org/current/userguide/gradle_wrapper.html#sec:upgrading_wrapper

Running it the second time now. Good catch.

inktomi · 2026-03-04T21:25:14Z

build.gradle


 tasks.wrapper.configure {
-    gradleVersion = '8.14.3'
+    gradleVersion = '9.1.0'


As with the settings for the overall project, this is the minimum required version for AGP 9.0.1

emartynov · 2026-03-05T11:31:49Z

Looks like some merge conflicts.

But the rest (Even some changes looks like not related) the change is LGTM.

@runningcode how does it look for you?

runningcode · 2026-03-05T11:38:28Z

Thank you so much for this work! It looks great.

Looks like there's a small lint error on CI.
Could you please also add an entry to the docs/changelog.md ?

runningcode · 2026-03-05T15:56:52Z

docs/changelog.md

 * Minimum required JVM version is now 17.
-* Minimum supported Gradle version is now 7.3.
+* Minimum supported Gradle version is now 9.1 (the minimum supported version for AGP 9.0.1).
+* Fixed support for Android Gradle Plugin version 9.0.1


This will go in the next release, could you put it in the unreleased section?

Hah wooop! Sorry! Will fix this right now.

…n lint issue

runningcode

Thanks, I gave it a thorough look and added some comments!

runningcode · 2026-03-05T16:31:56Z

fladle-plugin/src/main/java/com/osacky/flank/gradle/FladlePluginDelegate.kt

+    base.flankCoordinates.finalizeValueOnRead()
+    base.serviceAccountCredentials.finalizeValueOnRead()
+
+    // Register onVariants callbacks before afterEvaluate for APK path detection


Just curious is there a reason for the order change here?

This is part of the 9.0.1 upgrade - we used testVariants.configureEach before which was a lazy API that worked fine inside afterEvaluate. The new version must be registered during the configuration phase, before variants are finalized and so if we kept it using afterEvaluate it could be executed too late in the lifecycle which could make the callback flaky. The change in the execution order prevents that.

got it. 👍

runningcode · 2026-03-05T16:32:34Z

fladle-plugin/src/main/java/com/osacky/flank/gradle/FladlePluginDelegate.kt

      val useOrchestrator =
-        androidExtension.testOptions.getExecutionEnum() == TestOptions.Execution.ANDROIDX_TEST_ORCHESTRATOR ||
-          androidExtension.testOptions.getExecutionEnum() == TestOptions.Execution.ANDROID_TEST_ORCHESTRATOR
+        execution == "ANDROIDX_TEST_ORCHESTRATOR" ||


Any reason not to use the constants here?

TestOptions actually goes away with the upgrade, so we can't use it anymore. I replaced them with string comparison since the enum was removed from the public API.

ah makes sense!

runningcode · 2026-03-05T16:32:51Z

fladle-plugin/src/main/java/com/osacky/flank/gradle/FladlePluginDelegate.kt

+          execution == "ANDROID_TEST_ORCHESTRATOR"
      if (useOrchestrator) {
        log("Automatically detected the use of Android Test Orchestrator")
+        config.useOrchestrator.set(true)


:O was this a bug earlier?

Yes, it was sort of a bug! Consider this:

testOptions.execution = AGP's setting (in the android {} block)

config.useOrchestrator = Fladle's setting (in the fladle {} block)

The auto-detection reads from AGP's testOptions and writes to Fladle's config.useOrchestrator. Both happen in afterEvaluate, but the auto-detection runs last, so it wins.

An example: a user sets useOrchestrator = true in their Fladle config, but does not configure testOptions.execution for orchestrator in AGP. The old code reads AGP's testOptions → sees no orchestrator → calls config.useOrchestrator.set(false) → stomps on the user's explicit Fladle setting.

Setting it to true explicitly like this preserves their config when it's not also configured in fladle.

Thank you for fixing this!

runningcode · 2026-03-05T16:34:22Z

fladle-plugin/src/main/java/com/osacky/flank/gradle/FulladlePlugin.kt

      extensions.create("fulladleModuleConfig", FulladleModuleExtension::class.java)
+
+      // Register onVariants callbacks to capture APK info during configuration
+      pluginManager.withPlugin("com.android.application") {


Can this code be moved or shared with the FladlePluginDelegate ?

Not really, we're doing different things here:

FladlePluginDelegate reads variant info to set debugApk/instrumentationApk on the Fladle extension directly. FulladlePlugin collects variant info into VariantApkInfo objects on FulladleModuleExtension.variantApks for later processing in configureModule(). The outputs and data flows are different between the two, so we can't directly share the code.

I'm also working on migrating FulladlePlugin into a settings plugin to allow compatibility with the configuration cache. This code will have to move then, because the root-level settings plugin won't have access to the variants anymore - there will be a secondary settings-applied convention plugin that gets applied to each module which gathers that data internally.

I see. Oh that migration to a settings plugin really makes sense!

runningcode

Thank you!
Looks like github actions are having issues right now https://www.githubstatus.com/

I'll merge this when the action runs and passes

emartynov reviewed Mar 4, 2026

View reviewed changes

inktomi mentioned this pull request Mar 4, 2026

Code formatting & cleanup - broken out from AGP 9 update PR #482

Merged

inktomi force-pushed the agp-9.0.1 branch from 4a086dc to dc91e2c Compare March 4, 2026 20:34

inktomi added 2 commits March 4, 2026 12:35

Applied Android Studio JVM migration

8bf89f3

inktomi force-pushed the agp-9.0.1 branch from dc91e2c to 4954305 Compare March 4, 2026 20:43

inktomi commented Mar 4, 2026

View reviewed changes

inktomi requested a review from emartynov March 4, 2026 21:27

Merge branch 'master' into agp-9.0.1

02b0acd

inktomi added 2 commits March 5, 2026 07:50

Merge branch 'master' into agp-9.0.1

523e54c

Updated changelog with blurbs about changes for AGP 9.0.1 support

9d715e7

runningcode reviewed Mar 5, 2026

View reviewed changes

inktomi added 2 commits March 5, 2026 07:59

Added changelog info to unreleased section

0eb233f

Updated gradle wrapper binary, and fixes for chain-method-continuatio…

e9fdbb5

…n lint issue

inktomi requested a review from runningcode March 5, 2026 16:28

runningcode reviewed Mar 5, 2026

View reviewed changes

runningcode approved these changes Mar 5, 2026

View reviewed changes

Uh oh!

Conversation

inktomi commented Mar 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

Test plan

Uh oh!

inktomi commented Mar 2, 2026

Uh oh!

inktomi commented Mar 3, 2026

Uh oh!

emartynov left a comment

Choose a reason for hiding this comment

Uh oh!

inktomi commented Mar 4, 2026 via email

Uh oh!

inktomi commented Mar 4, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

emartynov commented Mar 5, 2026

Uh oh!

runningcode commented Mar 5, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

runningcode left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

runningcode left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

inktomi commented Mar 2, 2026 •

edited

Loading