Generate KnownTags from a tag-conventions registry

[dougqh]

Code-generates the KnownTags registry from a language-agnostic tag-conventions.yaml (+ a Java
overlay), replacing the hand-maintained version. Based on dougqh/tag-id-api (the dense-store
branch). Draft — the generator works end to end (main compiles, tests pass); the committed-output
wiring is the remaining piece.

From Claude:

Pipeline (buildSrc task dd-trace-java.tag-registry-generator): parse → resolve per-type tag sets
(extends/include/applies) → assign ids → assign slots by deterministic graph coloring over
per-type layers + an explicit trace_level layer → emit KnownTags.java.

• Ids encode serial | slot | intercepted (no name hash — known ids are dense-store addressed;
  the name hash is the custom/bucket path). Emitted as literals (javac constant-folds them) with a
  // tagId(...) derivation comment for auditability.
• slot is a fast-path bitmask bit, not positional storage — co-occurrence is tolerated (the scan
  is authoritative). Coloring gives co-occurring tags distinct bits (fewer scans); colorability
  derives from required. The whole mask fits one long.
• Trace-level is its own tier (trace_level section → its own TagMap layer); the per-span base
  type holds parent_id/_dd.integration/_dd.svc_src (core-set but per-span).
• Virtual keys (service/resource.name/error/sampling.priority/…) get reserved ids; the
  generated id → handler dispatch table is the on-ramp to retiring the TagInterceptor.

Verified: internal-api main compiles; KnownTagsTest (property-based resolver round-trip over ~55
tags) passes; OptimizedTagMapDenseForkedTest passes; spotless clean. Note: literal ids fold, so a
constant reference no longer triggers <clinit> — registration is via KnownTags.init().

Remaining (why draft): the generated KnownTags.java is a manual copy in src/main right now;
wire the task to emit to a committed src/generated/java + srcDir + a verifyGenerated freshness
check + format-on-generate, and route the layout reports (folded-types/tag-assignment) there as
committed artifacts.

🤖 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	14.05 s	13.98 s	[-0.1%; +1.2%] (no difference)
startup:insecure-bank:tracing:Agent	12.94 s	13.09 s	[-2.2%; -0.0%] (maybe better)
startup:petclinic:appsec:Agent	16.93 s	16.74 s	[-0.2%; +2.4%] (no difference)
startup:petclinic:iast:Agent	16.86 s	16.96 s	[-1.6%; +0.4%] (no difference)
startup:petclinic:profiling:Agent	16.21 s	16.70 s	[-7.4%; +1.6%] (no difference)
startup:petclinic:sca:Agent	16.84 s	16.73 s	[-0.4%; +1.7%] (no difference)
startup:petclinic:tracing:Agent	16.07 s	15.73 s	[-2.2%; +6.6%] (no difference)

Commit: ad3c1f70 · 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.

[datadog-prod-us1-3]

🎯 Code Coverage (details)
• Patch Coverage: 97.40%
• Overall Coverage: 57.31% (-0.01%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: ad3c1f7 | Docs | Datadog PR Page | Give us feedback!}