IssueTracker

Thank you for your patience while our engineering team worked to resolve this issue. A fix for this issue is now available in:

• Android Studio Ladybug Feature Drop | 2024.2.2 Canary 3
• Android Gradle Plugin 8.8.0-alpha03

We encourage you to try the latest update.

If you notice further issues or have questions, please file a new bug report.

Thank you for taking the time to submit feedback — we really appreciate it!

I looked a bit at this today, and tried to add a few keep rules to narrow down where the issue happens. However, if I apply this diff:

diff --git a/libraries/apollo-gradle-plugin/rules.pro b/libraries/apollo-gradle-plugin/rules.pro
index 4b0867e580..af80c3ce44 100644
--- a/libraries/apollo-gradle-plugin/rules.pro
+++ b/libraries/apollo-gradle-plugin/rules.pro
@@ -45,3 +45,7 @@
 
 # The Gradle API jar and other compileOnly dependencies aren't added to the classpath, ignore the missing symbols
 -dontwarn **
+
+-keep class com.apollographql.apollo.ast.** {
+ *;
+}

the task :generateServiceApolloSources starts failing throwing

java.lang.NoSuchMethodError: 'kotlin.jvm.internal.ClassReference kotlin.jvm.internal.Reflection.getOrCreateKotlinClass(java.lang.Class)'

You mentioned a pull request - does that mean that you have found the issue locally?

the task :generateServiceApolloSources starts failing throwing

Right, I can also reproduce that. Not 100% sure what's happening there. I'll dig a bit.

You mentioned a pull request - does that mean that you have found the issue locally?

Kind of. I'm not 100% sure about the best way to fix it but I'm reliably confident this has to do invokedynamic callsites.

This was some time ago so I'll need to refresh my memories but IIRC, it had to do with AffectedMethodUseRegistry and argument propagation. I believe AffectedMethodUseRegistry should implement registerCallSite() and mark those as affected?

java.lang.NoSuchMethodError: 'kotlin.jvm.internal.ClassReference kotlin.jvm.internal.Reflection.getOrCreateKotlinClass(java.lang.Class)'

I did a bit of digging around. I think what's happening here is that, for some reason, keeping com.apollographql.apollo.ast.** disables relocation of kotlin.**. Since this whole test is about testing kotlin.** relocation, it fails bceause the Gradle embedded version of Kotlin takes over.

Interestingly enough, kotlin.** is kept but there is some rewriting. In this case, getOrCreateKotlinClass return type is changed from kotlin.reflect.KClass to kotlin.jvm.internal.ClassReference, which is what triggers the issue (but the real issue is that Gradle kotlin-stdlib takes precedence bceuase kotlin.** is not relocated).

I hope that makes sense? Let me know if you want me to dig into AffectedMethodUseRegistry and I can try to suggest something.

Thank you for the additional info and analysis of where this might go wrong.

Christoffer, can you take a look at the AffectedMethodUseRegistry to check up on the analysis in comment #4?

Dump attached.

Looking at the disassembly, it seems that most call sites have been rewritten correctly, but there is one call site that still seems to reference the old method name:

.class public final Lcom/apollographql/apollo/relocated/com/apollographql/apollo/ast/internal/Parser;

...

.method public final parseValue()Lcom/apollographql/apollo/relocated/com/apollographql/apollo/ast/GQLValue;
    .locals 1

    .line 1
    invoke-custom {p0}, call_site_283("invoke", (Lcom/apollographql/apollo/relocated/com/apollographql/apollo/ast/internal/Parser;)Lcom/apollographql/apollo/relocated/kotlin/jvm/functions/Function0;, ()Ljava/lang/Object;, invoke-static@Lcom/apollographql/apollo/relocated/com/apollographql/apollo/ast/internal/Parser;->parseValue$lambda$1(Lcom/apollographql/apollo/relocated/com/apollographql/apollo/ast/internal/Parser;)Lcom/apollographql/apollo/relocated/com/apollographql/apollo/ast/GQLValue;, ()Lcom/apollographql/apollo/relocated/com/apollographql/apollo/ast/GQLValue;)@Ljava/lang/invoke/LambdaMetafactory;->metafactory(Ljava/lang/invoke/MethodHandles$Lookup;Ljava/lang/String;Ljava/lang/invoke/MethodType;Ljava/lang/invoke/MethodType;Ljava/lang/invoke/MethodHandle;Ljava/lang/invoke/MethodType;)Ljava/lang/invoke/CallSite;

    move-result-object v0

    invoke-virtual {p0, v0}, Lcom/apollographql/apollo/relocated/com/apollographql/apollo/ast/internal/Parser;->parseTopLevel(Lcom/apollographql/apollo/relocated/kotlin/jvm/functions/Function0;)Ljava/lang/Object;

    move-result-object p0

    check-cast p0, Lcom/apollographql/apollo/relocated/com/apollographql/apollo/ast/GQLValue;

    return-object p0
.end method

From looking at this in more details, I agree that AffectedMethodUseRegistry needs to account for call sites. In the concrete build this would not fix the issue, however, since the affected method java.util.ArrayList com.apollographql.apollo.ast.internal.Parser.parseDirectives(boolean) is already marked for reprocessing. The (other) issue is that LensCodeRewriter#rewritePartialDefault does not rewrite the method names in call sites that correspond to lambdas.

Hi 👋 Thanks for looking into this 🙏.

I have a very limited understanding of the inner details here. I don't have much to add but thought I'd drop by to say I'm still around. Let me know if/how I can help.

This issue currently forces us on a very ancient version of R8 and I'd feel much better if we could update.

Fixing this should be rather straight-forward. I expect to be able to look into this tomorrow.

Excellent news, thanks! 💙

Project: r8
Branch: 8.6
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98381

Version 8.6.35

Version 8.6.35 
 
Bug: b/366932318 
Change-Id: I8e18fd6469749d4e3bb478415d370a157f3a6aeb

Project: r8
Branch: 8.6
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98380

Pin methods called from lambdas in argument propagation

Pin methods called from lambdas in argument propagation 
 
Bug: b/366932318 
Change-Id: Ie3888ec6788637b897ecdde1027bb1b9aef99305

Project: r8
Branch: 8.7
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98362

Pin methods called from lambdas in argument propagation

Pin methods called from lambdas in argument propagation 
 
Bug: b/366932318 
Change-Id: Ie3888ec6788637b897ecdde1027bb1b9aef99305

Project: r8
Branch: 8.7
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98382

Version 8.7.22

Version 8.7.22 
 
Bug: b/366932318 
Change-Id: I1236e2692269bc0159f4cbf1070de19eb3784fee

Project: r8
Branch: 8.8
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98363

Pin methods called from lambdas in argument propagation

Pin methods called from lambdas in argument propagation 
 
Bug: b/366932318 
Change-Id: Ie3888ec6788637b897ecdde1027bb1b9aef99305

Project: r8
Branch: 8.8
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98364

Version 8.8.19

Version 8.8.19 
 
Bug: b/366932318 
Change-Id: I11fdec29da7b98ed3b822661231cd729b9276c83

Project: r8
Branch: 8.6
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98380

Pin methods called from lambdas in argument propagation

Pin methods called from lambdas in argument propagation 
 
Bug: b/366932318 
Change-Id: Ie3888ec6788637b897ecdde1027bb1b9aef99305

Project: r8
Branch: 8.7
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98362

Pin methods called from lambdas in argument propagation

Pin methods called from lambdas in argument propagation 
 
Bug: b/366932318 
Change-Id: Ie3888ec6788637b897ecdde1027bb1b9aef99305

Project: r8
Branch: 8.5
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98383

Pin methods called from lambdas in argument propagation

Pin methods called from lambdas in argument propagation 
 
Bug: b/366932318 
Change-Id: Ie3888ec6788637b897ecdde1027bb1b9aef99305

Project: r8
Branch: 8.5
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98365

Version 8.5.47

Version 8.5.47 
 
Bug: b/366932318 
Change-Id: Ib3884dfe3fcf4796a4c0c549e2fd2eaca21e8d8d

Hi 👋

I finally had the time to test the fix using r8:8.8.19 but sadly it doesn't seem to fix my issue. I just rebased the use-r8 branch:

git clone https://github.com/apollographql/apollo-kotlin
cd apollo-kotlin
git checkout 6980cf48a4600b06042054d9e6bbf8f013e5e68d
./gradlew :apollo-gradle-plugin:testJava11 --rerun --tests '*operationIdGenerator is working*' --no-build-cache --no-configuration-cache

Still outputs the same:

    Caused by: java.lang.AbstractMethodError: Receiver class com.apollographql.apollo.relocated.com.apollographql.apollo.ast.internal.Parser$$Lambda$1679/0x0000000800d5fc40 does not define or inherit an implementation of the resolved method 'abstract java.lang.Object invoke$3()' of interface com.apollographql.apollo.relocated.kotlin.jvm.functions.Function0.
        at com.apollographql.apollo.relocated.com.apollographql.apollo.ast.internal.Parser.parseTopLevel(SourceFile:4)
        at com.apollographql.apollo.relocated.com.apollographql.apollo.ast.internal.Parser.parseValue(SourceFile:1)
        at com.apollographql.apollo.relocated.com.apollographql.apollo.ast.ApolloParser__ApiKt.parseAsGQLValue$lambda$7$ApolloParser__ApiKt(SourceFile:1)

Is there something I'm missing?

Woops, bad commit, 8881b5481db7a66abcdabba7571fc7b72b3cf2dd is the commit to use:

git clone https://github.com/apollographql/apollo-kotlin
cd apollo-kotlin
git checkout 8881b5481db7a66abcdabba7571fc7b72b3cf2dd
./gradlew :apollo-gradle-plugin:testJava11 --rerun --tests '*operationIdGenerator is working*' --no-build-cache --no-configuration-cache

Project: r8
Branch: main
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98360

Account for lambda call sites in argument propagation

Account for lambda call sites in argument propagation 
 
Bug: b/366932318 
Change-Id: I7d6378034d4bf7e607badbe266653e273fdcd14a

Sorry about that. This should be fixed by https://r8.googlesource.com/r8/+/887704078a06fc0090e7772c921a30602bf1a49f.

Can you check if the main version fixes your issue (jar is at https://storage.googleapis.com/r8-releases/raw/main/887704078a06fc0090e7772c921a30602bf1a49f/r8lib.jar)?

Running tests seems to work for me:

> Task :apollo-gradle-plugin:testJava11

OperationIdGeneratorTests > operationIdGenerator is working PASSED

Works like a charm, thanks again!

Project: r8
Branch: 8.5
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98605

Version 8.5.48

Version 8.5.48 
 
Bug: b/366932318 
Change-Id: Iddbe33a40bee481c66e67030690c704d69ce35b3

Project: r8
Branch: 8.5
Author: Christoffer Adamsen <christofferqa@google.com>
Link:      https://r8-review.googlesource.com/98604

Account for lambda call sites in argument propagation

Account for lambda call sites in argument propagation 
 
Bug: b/366932318 
Change-Id: I7d6378034d4bf7e607badbe266653e273fdcd14a

Thank you for your patience while our engineering team worked to resolve this issue. A fix for this issue is now available in:

• Android Studio Ladybug Feature Drop | 2024.2.2 Patch 1
• Android Gradle Plugin 8.8.1

We encourage you to try the latest update.

If you notice further issues or have questions, please file a new bug report.

Thank you for taking the time to submit feedback — we really appreciate it!