Extract peer-connection tagging into a TagExtractor<InetSocketAddress>

[dougqh]

What

Second TagExtractor canary (stacked on #11817). Lifts the pure peer-connection field→tag logic — peer.hostname / peer.ipv4 / peer.ipv6 / peer.port — out of BaseDecorator.onPeerConnection(...) into PeerConnectionExtractor, a named singleton TagExtractor<InetSocketAddress>.

Why this one

• It's the extrinsic TagExtractor case — extraction from a JDK type we don't own (InetSocketAddress) — complementing the JDBC canary's intrinsic case (our own DBInfo). Together they exercise both shapes of the API.
• Bucket (a) pure field-extraction: no service/operation/peer-service derivation, no AppSec/IAST/DSM side-effects — the cleanest, safest slice to start dissolving the hierarchy.
• It exercises the cache-in-extractor pattern cleanly: the hostName(...) resolver cache becomes a class static the extractor consults (not lambda-captured state) — which is exactly why it's promoted to a named class.

Named singleton class

Promoted from an inline lambda to PeerConnectionExtractor implements TagExtractor<InetSocketAddress> with a private-ctor INSTANCE. Buys a name at call sites (span.setTags(addr, PeerConnectionExtractor.INSTANCE)), a home for the pure statics + the resolver cache, and composability — the axis the Decorator inheritance chain lacked. This is the convention the rest of the dissolution follows.

Scope / honesty

onPeerConnection is a concrete shared base method (52 call sites), not a megamorphic template method — so this is a structural dissolution step and a demonstrator of the extrinsic pattern, largely perf-neutral (it's not one of the per-leaf overridden dispatch sites where the megamorphic cost lives). The dispatch win comes later, from the overridden template methods (see #11823).

Behavior preservation

Identical tag output. Internal plumbing invokes extract() directly (so it runs on Spock test doubles too — a default-method span.setTags wouldn't); the span-first span.setTags(source, extractor) sugar is reserved for hand-written integration call sites. BaseDecoratorTest / ClientDecoratorTest / ServerDecoratorTest pass.

🤖 Generated with Claude Code

[datadog-official]

🎯 Code Coverage (details)
• Patch Coverage: 72.22%
• Overall Coverage: 54.93% (-0.01%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 07dc157 | Docs | Datadog PR Page | Give us feedback!}

[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	13.99 s	14.00 s	[-0.8%; +0.6%] (no difference)
startup:insecure-bank:tracing:Agent	12.92 s	12.98 s	[-1.3%; +0.3%] (no difference)
startup:petclinic:appsec:Agent	16.89 s	16.74 s	[-0.1%; +2.0%] (no difference)
startup:petclinic:iast:Agent	16.82 s	16.93 s	[-1.7%; +0.4%] (no difference)
startup:petclinic:profiling:Agent	16.24 s	16.89 s	[-8.0%; +0.4%] (no difference)
startup:petclinic:sca:Agent	16.88 s	16.63 s	[+0.6%; +2.4%] (maybe worse)
startup:petclinic:tracing:Agent	15.59 s	16.16 s	[-7.5%; +0.4%] (no difference)

Commit: 07dc1576 · 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.