Execution failed for task ':app:mergeReleaseClasses' after Updating AGP from 8.0.2 to 8.1.0 [295205663]

Fixed

Bug

Status Update

No update yet.

Description

mk...@opera.com

created issue #1

Aug 9, 2023 08:55PM

After Updating AGP from 8.0.2 to 8.1.0 I have the following build error

> Task :app:mergeReleaseClasses FAILED

FAILURE: Build failed with an exception.

* What went wrong:
Execution failed for task ':app:mergeReleaseClasses'.
> A failure occurred while executing com.android.build.gradle.internal.tasks.MergeClassesTask$MergeClassesWorkAction
> Zip file '/Users/.../build/app/intermediates/module_and_runtime_deps_classes/release/base.jar' already contains entry 'com/google/android/youtube/player/internal/u.class', cannot overwrite

The u.class mentioned in the error comes from a .jar file that I have in the libs dir.

I was looking in which AGP it started to break. When I get back to AGP 8.1.0-alpha10 it started to break with different error, this time related to R8 but mentioning the same base.jar file. I found the following issue

https://issuetracker.google.com/issues/277166577, then I changed some eager task creation to their lazy equivalents (basically replaced tasks.all with tasks.forEach and task with tasks.register) and the issue disappeared. But that didn't solve the original issue with the ZIP file that I mentioned at the top. This starts to appear in AGP 8.1.0-beta01. I still have many places using eager task creation across the build files, not sure if this is related though.

------------------

Version of Gradle Plugin: 8.1.0
Version of Gradle: 8.1.1
Version of Java: 17
OS: MacOS Ventura 13.4.1

Comments

ha...@google.com <ha...@google.com> Aug 9, 2023 10:57PM

Assigned to an...@google.com.

ga...@google.com <ga...@google.com> Aug 11, 2023 04:51PM

Reassigned to sp...@google.com.

sp...@google.com <sp...@google.com> #2Aug 11, 2023 05:22PM

Is the problem reproducible for you in Studio Arctic Fox? I wasn't able to reproduce it.

mk...@opera.com <mk...@opera.com> #3Aug 12, 2023 07:15PM

@2 Yes, it still occurs, even today. Tested on:

Android Studio Chipmunk | 2021.2.1 Canary 6
Build #AI-212.5457.46.2112.7968471, built on December 4, 2021
Runtime version: 11.0.12+7-b1504.28-7817840 amd64
VM: OpenJDK 64-Bit Server VM by Oracle Corporation
Windows 10 10.0
GC: G1 Young Generation, G1 Old Generation
Memory: 4000M
Cores: 12
Registry: external.system.auto.import.disabled=true, debugger.watches.in.variables=false
Non-Bundled Plugins: com.intellij.marketplace (212.5457.65), Show As ... (1.0.3), com.dubreuia (2.3.0), String Manipulation (8.24.203.5981.1), com.google.mad-scorecard (1.2), net.aquadc.mike.plugin (0.19)

mk...@opera.com <mk...@opera.com> #4Aug 13, 2023 04:22PM

I wasn't able to reproduce the problem with My Application.zip attached to comment #1 using Chipmunk Beta 4. Is the problem still reproducible for you?

Message last modified on Aug 14, 2023 11:33AM

sp...@google.com <sp...@google.com> #5Aug 15, 2023 05:27PM

If the problem is reproducible, what is the value of Gradle JDK in Settings?

ma...@gmail.com <ma...@gmail.com> #6Aug 15, 2023 05:34PM

@4 @5 It still occurs.
Show me what you've tried. Do exactly as on the video.
Also, I've noticed a different issue, that the mouse-back-key doesn't always go backward correctly on this project. Reported here:

https://issuetracker.google.com/issues/224564027

Android Studio Bumblebee | 2021.1.1 Patch 2
Build #AI-211.7628.21.2111.8193401, built on February 17, 2022
Runtime version: 11.0.11+9-b60-7590822 amd64
VM: OpenJDK 64-Bit Server VM by Oracle Corporation
Windows 10 10.0
GC: G1 Young Generation, G1 Old Generation
Memory: 8048M
Cores: 12
Registry: external.system.auto.import.disabled=true, ide.settings.move.mouse.on.default.button=true, debugger.watches.in.variables=false
Non-Bundled Plugins: GenerateSerialVersionUID (3.0.3), GenerateSerializationHelpers (1.0.6), Show As ... (1.0.3), com.dubreuia (2.2.0), com.github.beansoft.jadx.gui.idea (2021.2.1), com.intellij.marketplace (211.7628.36), String Manipulation (9.4.0), org.jetbrains.kotlin (211-1.6.10-release-923-AS7442.40), com.google.mad-scorecard (1.2), com.ppismerov.ksvu (0.0.1), net.aquadc.mike.plugin (0.21), izhangzhihao.rainbow.brackets (6.21), org.intellij.plugins.markdown (211.7142.37)

sp...@google.com <sp...@google.com> #7Aug 15, 2023 06:17PM

Reassigned to am...@google.com.

I've finally noticed the difference between my and your actions. I was using Navigate > Declaration or Usages and you were invoking Navigate > Implementation(s). Navigate > Implementation(s) indeed doesn't navigate anywhere. I don't really know what "implementation" is supposed to mean for an Android resource, but the current behavior is clearly wrong.

mk...@opera.com <mk...@opera.com> #8Aug 15, 2023 07:31PM

@7 Why shouldn't it allow to reach the only other place/s it can reach (meaning everything except the current place)?

As for "usages", I prefer not to use it, because it often shows results that I don't want to see, and it puts them all at the top.
See for example the case of translated strings. I want to see the usages of a string, and instead I get to see all the translations of all languages, and only below it all I can see the real usages of it...

Message last modified on Aug 15, 2023 07:48PM

am...@google.com <am...@google.com> #9Aug 22, 2023 11:04AM

Why mark as fixed? What did you do?

mk...@opera.com <mk...@opera.com> #10Aug 22, 2023 07:45PM

I improved behavior of Navigate > Implementation(s) for resource references in R classes, but unfortunately the list of suggestions now contains two elements, one is the element in the R class under the cursor and another is the resource file. The first suggestion is pretty useless since it doesn't navigate anywhere, but I was not able to get rid of it because it is controlled by the Java plugin without a mechanism to inject any custom logic. I'll continue experimenting with it in an attempt to find a cleaner solution.

am...@google.com <am...@google.com> #11Aug 23, 2023 08:48AM

@10 Can you please just make it reach the resource itself, and only if it's in multiple places, show a list to reach ?

mk...@opera.com <mk...@opera.com> #12Aug 23, 2023 06:50PM

Re #11 This is what I'm trying to do, but it looks infeasible at this point without upstream changes.

am...@google.com <am...@google.com> #13Aug 25, 2023 12:22PM

@12 Maybe tell something to IntelliJ ?

:(

mk...@opera.com <mk...@opera.com> #14Aug 25, 2023 03:39PM

Figured out a way to make it navigate directly. Rejoyce.

Message last modified on Aug 25, 2023 05:15PM

mk...@opera.com <mk...@opera.com> #15Aug 29, 2023 09:55AM

I wonder if it makes sense to flip the roles of Navigate > Declaration or Usages and Navigate > Implementation(s) for Android resources. This way Ctrl+B would navigate directly and Ctrl+Alt+B would display a list of locale-specific options. What is your opinion?

am...@google.com <am...@google.com> #16Aug 29, 2023 03:36PM

@15 Please don't publish my name here.
As for shortcuts, I don't use a lot of the official shortcuts (too many keys, sometimes very far from one another, so I customized them) so please write their names instead.

What exactly did you do?

The declaration&implementation of a resource are the same, because they are both in the same place. Maybe when they exist in multiple places, you could offer to reach each of them (please prefer the main one, like English for translations, as the first item).

As you look currently at "R.layout.some_layout", there is no point in staying there when performing any kind of navigation.
Same for strings. Same for drawables....

am...@google.com <am...@google.com> #17Aug 29, 2023 03:38PM

With the fix Navigate > Implementation(s) goes directly to the resource for the default configuration. Navigate > Declaration or Usages shows a list of resources with the same name if there are more than one. I got a feeling that it would be more logical if they behaved the other way around. What do you think?

mk...@opera.com <mk...@opera.com> #18Aug 29, 2023 04:31PM

@17 It's fine.
For "find usages", I just wish it would first show the usages (or scroll to them) in the list. Sadly it first shows the resources instead.
For example, when I want to see all usages of some string, it first shows me all the translations to the string, and only at the bottom of the list the actual usages of it.
It's very annoying because I don't at translations as often as I look at actual usages of something...

output.txt

31 KB

View

Download

mk...@opera.com <mk...@opera.com> #19Aug 31, 2023 07:15PM

@17 BTW what you've done it will work whether I'm in Java, Kotlin or even XML , right?

Message last modified on Aug 31, 2023 07:15PM

am...@google.com <am...@google.com> #20Aug 31, 2023 07:20PM

@18 The ordering of find usages is a separate issue. It is tracked by b/124294806.

mk...@opera.com <mk...@opera.com> #21Sep 1, 2023 06:04AM

@19 Thank you for reminding me. My fix was for Java and Kotlin only. Doesn't work properly in XML yet.

Message last modified on Sep 1, 2023 06:16AM

am...@google.com <am...@google.com> #22Sep 1, 2023 10:43AM

@20 I'm not talking only about strings. It's true for all resources.

@21 I'm also not familiar with how Compose works, so maybe check there too.

am...@google.com <am...@google.com> Sep 4, 2023 11:58AM

Marked as fixed.

an...@google.com <an...@google.com> #23Sep 7, 2023 11:02PM

Dolphin Canary 8 contains the fix a for navigation from XML. Please let me know if you find any scenarios that are not working properly yet.

an...@google.com <an...@google.com> #24Sep 13, 2023 06:59PM

@23 Is it available right now?

st...@gmail.com <st...@gmail.com> #25Mar 12, 2024 04:20AM

Yes.

zt...@gmail.com <zt...@gmail.com> #26Jun 25, 2024 08:37AM

Comment has been deleted.

Message last modified on Jun 25, 2024 09:03AM

zt...@gmail.com <zt...@gmail.com> #27Jun 25, 2024 09:05AM

Sorry, I made a mistake. The fix for the navigation from XML will be included in Dolphin Canary 9, not 8.