Ignore GraalVM features in HibernateValidatorProcessor

[gsmet]

Fixes #47033

[marko-bekhta]

I was thinking whether we could say something like:

if (hasAnyConstraints(indexView, implementor)) {

If there are no constraints, there's nothing to cascade into, right? But I don't remember if we need to keep the classes without constraints.. will try to take a look later

[gsmet]

I think:

• if the class is a leaf class
• it doesn't have any constraints

we can skip it.

So I think we could go with your approach... but... I would still try to exclude the Feature classes as you never know if someone adds a @NotNull by mistake.

[marko-bekhta]

yeah +1 just to be safe

[gsmet]

I let you check that excluding the leaf classes without constraints would be fine?

[marko-bekhta]

yes 👍🏻 🙂

[marko-bekhta]

ok checked aand that won't fly 🙈:

Validator validator = validator( Set.of( A.class ) );

assertThat( validator.validate( new A() ) ).containsOnlyViolations(
		violationOf( NotNull.class ).withProperty( "name" )
);

// fails as we get the `UninitializedBeanMetaData` and hence no constraints ... 
assertThat( validator.validate( new B() ) ).containsOnlyViolations(
		violationOf( NotNull.class ).withProperty( "name" )
);

If we skip the class, even if it's a leaf with no constraints (or if it's anywhere in the hierarchy for that matter), we will end up having the UninitializedBeanMetaData for it once someone wants to validate an instance of it and it simply has 0 constraints (the ones from the super classes are there).

[gsmet]

I think we should have started with: WTH is this class in the index? I wouldn't expect it to be.

[gsmet]

OK so the problem comes from CXF, which aggressively indexes several jars including: https://github.com/quarkiverse/quarkus-cxf/blob/main/extensions/core/deployment/src/main/java/io/quarkiverse/cxf/deployment/JakartaActivationProcessor.java

I pinged Peter here: https://quarkusio.zulipchat.com/#narrow/channel/187038-dev/topic/CXF.20and.20indexing/with/510702132 .

I will pursue this anyway.

[gsmet]

@marko-bekhta I just pushed an updated version on top of @Ladicek 's work in #47213 .

I will rebase and get it out of draft once the Jandex update PR is merged.

[quarkus-bot]

Status for workflow Quarkus CI

This is the status report for running Quarkus CI on commit 27b475b.

✅ The latest workflow run for the pull request has completed successfully.

It should be safe to merge provided you have a look at the other checks in the summary.

You can consult the Develocity build scans.

[gsmet]

@marko-bekhta I'm interested in your review!

[marko-bekhta]

looks nice! 🎉