[dougqh]

What Does This Do

Extraction no longer formats hex strings for ids that are usually never read as hex. One commit each:

1. The extracted Datadog parent id is formatted only if read.
   • The extraction cache turned x-datadog-parent-id into a 16-character hex string on every extraction. Only a W3C span id override reads it, as _dd.parent_id.
   • The cache now keeps the parsed long and formats it on demand. It still parses as each header arrives, so invalid values are still skipped when the header repeats.
2. The traceparent trace id is built from its parsed parts.
   • DD128bTraceId.fromHex kept a 32-character substring of the traceparent as the id's cached hex form, once per extraction.
   • Both parse paths now read the high and low order bits the same way. The id is built with DD128bTraceId.from(high, low), which formats its hex lazily and caches it.

Motivation

Ids unique to a trace or span shouldn't be formatted or cached unless they're read. In the Kafka consumer macro benchmark (#13051, JFR allocation profile, 10M records), ExtractionCache.cacheDatadogSpanId alone accounts for about 0.2 GB of String allocation.

• Leaf services no longer allocate either string.
• Services that inject W3C downstream still format the trace id hex, once per trace, as before.

No JMH numbers yet. The propagation tests and the compound-extraction golden corpus (which covers _dd.parent_id) pass unchanged.

Additional Notes

Stacked on #13043.

🤖 Generated with Claude Code

[datadog-prod-us1-3]

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 59.99% (-0.02%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: bddcbf2 | Docs | 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	14.04 s	13.90 s	[+0.3%; +1.7%] (maybe worse)
startup:insecure-bank:tracing:Agent	12.86 s	13.02 s	[-1.8%; -0.5%] (maybe better)
startup:petclinic:appsec:Agent	17.65 s	17.42 s	[+0.4%; +2.2%] (maybe worse)
startup:petclinic:iast:Agent	17.50 s	17.12 s	[-2.0%; +6.4%] (no difference)
startup:petclinic:profiling:Agent	17.33 s	17.20 s	[-0.3%; +1.9%] (no difference)
startup:petclinic:sca:Agent	17.70 s	17.47 s	[+0.3%; +2.4%] (maybe worse)
startup:petclinic:tracing:Agent	16.55 s	16.63 s	[-1.6%; +0.6%] (no difference)

Commit: bddcbf27 · 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.