[dougqh]

What Does This Do

Moves four rarely read fields off DDSpanContext, which exists once per span:

field	now read from	read when
injectBaggageAsTags	the tracer, through the collector	serialization
injectLinksAsTags	the tracer, through the collector	serialization
disableSamplingMechanismValidation	the tracer, through the collector	a sampling decision fails validation
w3cBaggage	the trace collector, set by the local root	serialization, for the local root only (as before)

The three settings were always the tracer's own final fields, copied into every span. The DDSpanContext constructors lose those three parameters. The 21-argument variant became identical to the 19-argument one, so it is removed.

Motivation

TraceSegment work (APMLP-2049): per-trace and per-tracer state moves off the per-span object.

JOL shallow sizes (JDK 17, compressed oops):

class	before	after
DDSpanContext	128 B	120 B
PendingTrace	128 B	128 B
StreamingTraceCollector	72 B	72 B

DDSpanContext was full: 128 B with a 1-byte gap. Removing any single 4-byte reference would have saved nothing. These four fields free 7 bytes, which together with the gap reaches the next 8-byte boundary down. w3cBaggage fits in the collectors' existing padding, so this is −8 B per span with no per-trace cost.

Hops. No field read on span creation, tagging or activation moves:

• tagInterceptor, profilingContextIntegration and the trace id stay on the span context.
• The moved fields are read only on the serializer thread, or when a decision fails validation.

That was deliberate, because an extra dependent load on a hot path could cost more than the 8 bytes save. The trace id can move later, together with a primitive trace-id view, where its hop can be measured on its own.

Additional Notes

• Behavior is unchanged in production, where every span already carried its tracer's values.
• DDSpanSerializationTest's baggage cases set injectBaggageAsTags on the tracer instead of on each context. The other direct constructions passed the tracer defaults.
• CompoundExtractionGoldenTest passes.
• Tests: the full dd-trace-core suite passes locally, except for 2 Docker-dependent TracerConnectionReliabilityTest cases and the order-dependent PendingTraceBufferTest.testingTracerFlareDumpWithMultipleTraces. The OpenTracing and OpenTelemetry TypeConverterTests pass.

Stacked: master ← #13008 ← #13000 ← #13009 ← #13010 ← #13015 ← #13028 ← #13029 ← this PR.

🤖 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	13.95 s	13.84 s	[+0.0%; +1.5%] (maybe worse)
startup:insecure-bank:tracing:Agent	12.94 s	12.90 s	[-0.4%; +1.0%] (no difference)
startup:petclinic:appsec:Agent	17.16 s	17.10 s	[-0.7%; +1.3%] (no difference)
startup:petclinic:iast:Agent	17.01 s	17.02 s	[-0.9%; +0.8%] (no difference)
startup:petclinic:profiling:Agent	16.78 s	16.80 s	[-1.2%; +0.9%] (no difference)
startup:petclinic:sca:Agent	17.08 s	16.86 s	[+0.6%; +2.0%] (maybe worse)
startup:petclinic:tracing:Agent	16.27 s	16.12 s	[-0.0%; +1.9%] (no difference)

Commit: 55176e3f · 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.