Grpc client interceptors are not registered when injected via constructor

[RoMiRoSSaN]

Describe the bug

Since version 3.9 client interceptors are not registered when injected via constructor, if there are more than 1 of interceptor.
If there is one, then it registers successfully.
When inject by field - all works.

How I understand, there were changes in InjectionPointInfo or somewhere else that affected this. If more than 1 interceptor is listed, they are not included in the qualifiers. As a result, not a single interceptor is registered.
https://github.com/quarkusio/quarkus/blob/main/extensions/grpc/deployment/src/main/java/io/quarkus/grpc/deployment/GrpcClientProcessor.java#L433

Before version 3.9 it works normal. I checked from 3.9 to 3.17, and this bug containing.

Expected behavior

1. If 1 interceptor - interceptor registered
2. If 2 or more interceptors - interceptor registered

Actual behavior

1. If 1 interceptor - interceptor registered
2. If 2 or more interceptors - none are registered

How to Reproduce?

Reproducer: grpc-demo.zip

Output of uname -a or ver

No response

Output of java -version

No response

Quarkus version or git rev

No response

Build tool (ie. output of mvnw --version or gradlew --version)

No response

Additional information

No response

[quarkus-bot]

/cc @alesj (grpc), @cescoffier (grpc)

[geoand]

cc @mkouba

[mkouba]

It's very likely a regression caused by d34eb5a#diff-c176df2cb8a1cd56ae30ff243a7c6f817604c94f4b755ff424a79bdf0fd6b798. The code filters the qualifiers by annotation target but for nested annotations (which is the case for repeatable annotations used in a container) it's null.

CC @manovotn

[manovotn]

@RoMiRoSSaN in your reproducer, I don't see the FirstClientInterceptor and SecondClientInterceptor being invoked regardless of whether you use field injection or constructor. Am I missing something?
I do see the GlobalClientInterceptor being invoked but none other.

It's very likely a regression caused by d34eb5a#diff-c176df2cb8a1cd56ae30ff243a7c6f817604c94f4b755ff424a79bdf0fd6b798. The code filters the qualifiers by annotation target but for nested annotations (which is the case for repeatable annotations used in a container) it's null.

Yes, after some digging it seems that the code here is a leftover of the commit you posted that should have been removed with #40841. We should now be able to just return transformationContext.getQualifiers()
That doesn't seem to break any ArC tests but doesn't fix the reproducer for me either 🤔
I'll get back to it tomorrow...

[RoMiRoSSaN]

@manovotn Hi. I have redesigned the reproducer for better understanding.
Start it and call GET/client method. In out will be

Start call from injected by constructor one - first interceptor will be
FIRST CLIENT
End call from injected by constructor one. If you see FIRST CLIENT - all works

Start call from injected by constructor two - must be two interceptors messages
End call from injected by constructor two. Nothing. Interceptors not working     <------ problem

Start call from injected by field one - first interceptor will be
FIRST CLIENT
End call from injected by field one. If you see FIRST CLIENT - all works

Start call from injected by field two - will be first and second interceptors
FIRST CLIENT
SECOND CLIENT
End call from injected by field two. If you see FIRST and SECOND CLIENT - all works

Reproducer: grpc-demo.zip

[mkouba]

That doesn't seem to break any ArC tests but doesn't fix the reproducer for me either

In any case, we should add an alternative to ClientInterceptorRegistrationTest with constructor injection.

[alesj]

with constructor injection

Yeah, just copy/paste the test and should be fine.
Or perhaps some base class for both cases - field and ctor injection test.

    public CtorClientInterceptorRegistrationTest(
        @RegisterClientInterceptor(MyLastClientInterceptor.class)
        @RegisterClientInterceptor(MyThirdClientInterceptor.class)
        @GrpcClient("hello-service") GreeterGrpc.GreeterBlockingStub client)
    {
        this.client = client;
    }

[manovotn]

Ah, I see, I misunderstood it originally, thanks for clarification @RoMiRoSSaN :)

The fix I mentioned above works for your reproducer, so I'll push that along with a test momentarily.