Convert JMSDecorator.getDestination to ClassLatch

[dougqh]

Summary

Converts JMSDecorator.getDestination's catch (AbstractMethodError) fallback (JMS <=1.1 producers lack getDestination) to use ClassLatch, so the error is paid once per class rather than once per call, matching the pattern established in #12702 for JDBCDecorator.

One of the candidate sites scanned in APMLP-1895. Draft/trial PR to see how the conversion holds up for this call site.

Stacked on #12702 (ClassLatch) — not mergeable until that lands.

The <=1.1 getQueue/getTopic fallback lives in the latch's fallback override (added to ClassLatch in #12702), so getDestination is a single tryApplyOrNull call. This also settles a correctness nuance: MessageProducer.getDestination() legitimately returns null for a producer created with an unidentified destination (session.createProducer(null)), which must not be confused with a latched/unavailable method — otherwise a plain MessageProducer could hit the (TopicPublisher) ... .getTopic() cast and throw ClassCastException. With the fallback inside the latch, a real null is returned as-is and only a failed or latched call takes the fallback, with no isLatched re-check at the call site.

Behavior changes

Compared to the original catch (AbstractMethodError):

• UnsupportedOperationException from getDestination() now takes the getQueue/getTopic fallback (never latched, since it can't be attributed to a class) instead of propagating to the caller.
• A null producer now yields null instead of throwing NullPointerException.
• An AbstractMethodError whose message does not name the producer's class (e.g. thrown inside a wrapper's delegate) still takes the fallback, as before, but does not latch the class.

Test plan

• New JUnit 5 test JMSDecoratorGetDestinationTest (Proxy-based mocks, per the ParseDBInfoClientInfoTest precedent): real-destination passthrough, real-null passthrough for a plain producer, AbstractMethodError fallback to QueueSender.getQueue() with latching confirmed on a second call, fallback to TopicPublisher.getTopic(), fallback without latching for an error naming another class, and latch isolation across different proxy classes.
• ./gradlew :dd-java-agent:instrumentation:jms:javax-jms-1.1:test on JDK 8 and JDK 11 (before the fallback rework)
• Same task after moving the fallback into the latch, on the default local JDK only (149 tests pass)

🤖 Generated with Claude Code

[dd-octo-sts]

🟢 Java Benchmark SLOs — All performance SLOs passed

Suite	Status
Startup	🟢 pass

SLO thresholds are defined here based on automatically generated metrics. A warning is raised when results are within 5% of the threshold.

PR vs. master results

Scenario	Candidate	master	Δ (95% CI of mean)
startup:insecure-bank:iast:Agent	14.81 s	14.74 s	[-0.4%; +1.4%] (no difference)
startup:insecure-bank:tracing:Agent	13.77 s	13.75 s	[-0.6%; +0.9%] (no difference)
startup:petclinic:appsec:Agent	16.94 s	16.97 s	[-1.0%; +0.6%] (no difference)
startup:petclinic:iast:Agent	16.99 s	17.09 s	[-1.2%; +0.0%] (no difference)
startup:petclinic:profiling:Agent	16.65 s	16.90 s	[-2.6%; -0.3%] (maybe better)
startup:petclinic:sca:Agent	16.97 s	16.79 s	[+0.2%; +2.0%] (maybe worse)
startup:petclinic:tracing:Agent	16.09 s	15.31 s	[-0.8%; +11.1%] (unstable)

Commit: 39df7a5c · CI Pipeline · Benchmarking Platform UI

---

Load and DaCapo benchmarks can be triggered manually in the GitLab pipeline. Results will appear in the Benchmarking Platform UI after completion.