Closed gmazzo closed 7 months ago
Can you update the Gradle version(s) in checks matrix so the tests can pass?
Can you update the Gradle version(s) in the checks matrix so the tests can pass?
Can you provide better actionable feedback? Which versions of Gradle should we be targeting?
I've checked locally, and main
seems to be currently broken. Running ./gradlew check -s
from a fresh checkout prints:
java.lang.StackOverflowError
at java.base/java.util.TreeSet.iterator(TreeSet.java:181)
at org.gradle.execution.plan.FinalizerGroup.isCanCancel(FinalizerGroup.java:153)
at org.gradle.execution.plan.CompositeNodeGroup.isCanCancel(CompositeNodeGroup.java:101)
at org.gradle.execution.plan.Node.isCanCancel(Node.java:232)
at org.gradle.execution.plan.FinalizerGroup.isCanCancel(FinalizerGroup.java:155)
at org.gradle.execution.plan.CompositeNodeGroup.isCanCancel(CompositeNodeGroup.java:101)
at org.gradle.execution.plan.Node.isCanCancel(Node.java:232)
It seems my changes didn't introduce this issue, but rather this project needs more active contributions. Is it even maintained?
Fixing the whole build should be addressed in a separate PR, not in this one.
Can you update the Gradle version(s) in the checks matrix so the tests can pass?
Can you provide better actionable feedback? Which versions of Gradle should we be targeting?
I've checked locally, and
main
seems to be currently broken. Running./gradlew check -s
from a fresh checkout prints:java.lang.StackOverflowError at java.base/java.util.TreeSet.iterator(TreeSet.java:181) at org.gradle.execution.plan.FinalizerGroup.isCanCancel(FinalizerGroup.java:153) at org.gradle.execution.plan.CompositeNodeGroup.isCanCancel(CompositeNodeGroup.java:101) at org.gradle.execution.plan.Node.isCanCancel(Node.java:232) at org.gradle.execution.plan.FinalizerGroup.isCanCancel(FinalizerGroup.java:155) at org.gradle.execution.plan.CompositeNodeGroup.isCanCancel(CompositeNodeGroup.java:101) at org.gradle.execution.plan.Node.isCanCancel(Node.java:232)
It seems my changes didn't introduce this issue, but rather this project needs more active contributions. Is it even maintained?
Fixing the whole build should be addressed in a separate PR, not in this one.
I thought it was broken with your update, but seems to just be broken in general. I'll see if I can get it fixed.
I thought it was broken with your update, but seems to just be broken in general. I'll see if I can get it fixed.
Actually, I have it fixed locally already.
The main offender is the com.guardsquare.proguard
plugin, which it seems discontinued (up to AGP 7.x
) and redundant, as AGP always had it embedded and later replaced with R8.
After removing it, and therefore the sample/proguard
project I was able to fix the build, and even bump both Gradle and AGP versions to the latest.
I just submitted #158 fixing the build
Closing in favor of #159
What has changed
Addresses #153 with a minimal set of changes to unblock the consumers of this plugin.
The breaking change is produced because the plugin is accessing an internal AGP API for getting the Android SDK location.
As suggested by @SimonMarquis, this can be done with a public API in
BaseExtension.sdkDirectory
. I made minimal changes to read from there, falling back to anANDROID_HOME
environment variable if not provided to keep the coupling at the bare minimum.Additionally, I've moved the whole
ANDROID_GRADLE_PLUGIN
dependency as acompileOnly
to avoid leaking it to the consumersWhy was it changed
Because since
AGP 8.3.0
, the internal API signature has changed, and it breaks: #153Related issues
AGP 8.3.0 fails with NoSuchMethodError #153