ArC fixes for spec compatibility, round 2

[Ladicek]

Related to #28558

• fix obtaining priority of a producer-based alternative from class-level @Priority
  • producers take priority from the declaring class
• fix generating Bean.create() to wrap checked exceptions in CreationException
  • the path of least resistance: what was up to now generated as create() is now called doCreate() and create() just wraps that into try/catch
• add support for static disposer methods
  • also fixes a NPE in case of static producer method declared on a @Dependent bean
• improve runtime representation of array bean types
  • we always used GenericArrayType to represent array bean types at runtime, which is unexpected for people, and also for at least one TCK test that expects a java.lang.Class representation of String[]
• add validation for array-typed producers

[gsmet]

@Ladicek I don't think I have seen this test fail before ^

[Ladicek]

Interesting. The failure is:

2023-01-30T14:38:49.7646913Z [INFO] Running io.quarkus.context.test.SimpleContextPropagationTest
2023-01-30T14:38:51.9399577Z 2023-01-30 14:38:51,927 INFO  [io.quarkus] (main) quarkus-integration-test-smallrye-context-propagation 999-SNAPSHOT on JVM (powered by Quarkus 999-SNAPSHOT) started in 1.155s. Listening on: http://localhost:8081
2023-01-30T14:38:51.9400595Z 2023-01-30 14:38:51,927 INFO  [io.quarkus] (main) Profile test activated. 
2023-01-30T14:38:51.9401730Z 2023-01-30 14:38:51,927 INFO  [io.quarkus] (main) Installed features: [agroal, cdi, hibernate-orm, hibernate-orm-panache, jdbc-h2, narayana-jta, resteasy-reactive, smallrye-context-propagation, vertx]
2023-01-30T14:38:56.0898796Z 2023-01-30 14:38:56,081 INFO  [io.quarkus] (main) quarkus-integration-test-smallrye-context-propagation stopped in 0.037s
2023-01-30T14:38:56.1016170Z [ERROR] Tests run: 11, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 6.325 s <<< FAILURE! - in io.quarkus.context.test.SimpleContextPropagationTest
2023-01-30T14:38:56.1017059Z [ERROR] io.quarkus.context.test.SimpleContextPropagationTest.testArcTCContextPropagationDisabled  Time elapsed: 0.012 s  <<< FAILURE!
2023-01-30T14:38:56.1018082Z org.opentest4j.AssertionFailedError: expected: <2> but was: <3>
2023-01-30T14:38:56.1018574Z 	at org.junit.jupiter.api.AssertionFailureBuilder.build(AssertionFailureBuilder.java:151)
2023-01-30T14:38:56.1019163Z 	at org.junit.jupiter.api.AssertionFailureBuilder.buildAndThrow(AssertionFailureBuilder.java:132)
2023-01-30T14:38:56.1019713Z 	at org.junit.jupiter.api.AssertEquals.failNotEqual(AssertEquals.java:197)
2023-01-30T14:38:56.1020208Z 	at org.junit.jupiter.api.AssertEquals.assertEquals(AssertEquals.java:150)
2023-01-30T14:38:56.1020674Z 	at org.junit.jupiter.api.AssertEquals.assertEquals(AssertEquals.java:145)
2023-01-30T14:38:56.1021146Z 	at org.junit.jupiter.api.Assertions.assertEquals(Assertions.java:528)
2023-01-30T14:38:56.1021813Z 	at io.quarkus.context.test.SimpleContextPropagationTest.testArcTCContextPropagationDisabled(SimpleContextPropagationTest.java:92)
2023-01-30T14:38:56.1022503Z 	at ...

[manovotn]

Note that this test used to be flaky, but I am not aware of it being problematic since we swapped tests to use RR instead of classic RE (Oct last year).
See the history - https://github.com/quarkusio/quarkus/commits/main/integration-tests/smallrye-context-propagation/src/test/java/io/quarkus/context/test/SimpleContextPropagationTest.java

[manovotn]

Frankly speaking, I doubt the CP test failure has anything to do with changes in this PR and it's not reproducible locally.
I'd say if the next CI run passes it, we should be fine merging it.
In case the test fails elsewhere, we should start treating it as flaky, probably temporarily excluding it, but I haven't seen it fail anywhere else yet.

[quarkus-bot]

✔️ 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.