feat: implement stable configuration

[bm1549]

Description

Implement Phase 1 Stable Configuration support for dd-trace-cpp, as defined in the RFC.

Reads application_monitoring.yaml files from disk at tracer initialization, parses the apm_configuration_default map, and applies values with correct configuration precedence across all supported tracer configs.

Key changes

• YAML parsing via yaml-cpp (yaml_parser.h/cpp): Clean interface — parse_yaml(string) → {config_id, values map}. Uses yaml-cpp 0.8.0 (FetchContent, static build). Handles all YAML features correctly.
• Stable config loading (stable_config.h/cpp): Reads two files (user-managed local + fleet-managed), 256KB size limit. Platform paths for Linux and Windows (registry-based agent directory with C:\ProgramData\Datadog\ fallback).
• Config precedence: fleet_stable > env > user/code > local_stable > default — matches all other tracer implementations. All tracer configs participate in stable config precedence, including tags, propagation styles, sampling rules, baggage, and resource renaming.
• Telemetry: Two new ConfigMetadata::Origin values (LOCAL_STABLE_CONFIG, FLEET_STABLE_CONFIG). Fleet config_id attached at serialization time.
• Extended resolve_and_record_config(): 5-parameter overload with fleet/local stable sources. The 3-parameter version delegates to it.
• Parametric test server: /config endpoint dynamically exposes all stable config values.

Files

File	Change
cmake/deps/yaml.cmake	New: yaml-cpp 0.8.0 FetchContent dependency
src/datadog/yaml_parser.h/cpp	New: YAML parser interface using yaml-cpp
src/datadog/stable_config.h/cpp	New: StableConfig structs, file loading, path resolution, parse_stable_config_rules template
include/datadog/config.h	New origin enum values, 5-param resolve_and_record_config
include/datadog/tracer_config.h	Stable config value maps on FinalizedTracerConfig (for parametric test server)
include/datadog/telemetry/configuration.h	fleet_stable_config_id on FinalizedConfiguration
include/datadog/trace_sampler_config.h	StableConfigs* parameter on finalize_config
include/datadog/span_sampler_config.h	StableConfigs* parameter on finalize_config
src/datadog/tracer_config.cpp	StablePair helpers, all configs wired to 5-param resolve
src/datadog/trace_sampler_config.cpp	Stable config for sample rate, rate limit, sampling rules
src/datadog/span_sampler_config.cpp	Stable config for span sampling rules
src/datadog/telemetry/telemetry_impl.cpp	Origin serialization, fleet config_id at serialization time
test/test_yaml_parser.cpp	26 YAML parser tests
test/test_stable_config.cpp	16 stable config integration tests
test/system-tests/request_handler.cpp/h	Parametric server /config endpoint

Testing

• 148 total tests (42 new), all passing
• 26 YAML parser tests: valid/invalid input, quoting, document markers, anchors, JSON-in-YAML
• 16 stable config tests: lookup, precedence, path resolution, finalize_config integration
• System tests pass locally: Test_Stable_Config_Default (20/20) and Test_Stable_Configuration_Origin (all passing)
• Related system-tests PR: feat(cpp): enable stable config telemetry tests system-tests#6554

Motivation

Enable fleet-managed and user-managed APM product configuration for C++ tracers (dd-trace-cpp, nginx-datadog, httpd-datadog). This is the last major language to get stable config support. nginx-datadog and httpd-datadog inherit the capability transparently through finalize_config().

Additional Notes

Jira ticket: [PROJ-IDENT]

🤖 Generated with Claude Code

[pr-commenter]

Benchmarks

Benchmark execution time: 2026-03-21 14:54:31

Comparing candidate commit 20488e9 in PR branch bm1549/stable-configuration with baseline commit 14280af in branch main.

Found 0 performance improvements and 1 performance regressions! Performance is the same for 0 metrics, 0 unstable metrics.

Explanation

This is an A/B test comparing a candidate commit's performance against that of a baseline commit. Performance changes are noted in the tables below as:

• 🟩 = significantly better candidate vs. baseline
• 🟥 = significantly worse candidate vs. baseline

We compute a confidence interval (CI) over the relative difference of means between metrics from the candidate and baseline commits, considering the baseline as the reference.

If the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD), the change is considered significant.

Feel free to reach out to #apm-benchmarking-platform on Slack if you have any questions.

More details about the CI and significant changes

You can imagine this CI as a range of values that is likely to contain the true difference of means between the candidate and baseline commits.

CIs of the difference of means are often centered around 0%, because often changes are not that big:

---------------------------------(------|---^--------)-------------------------------->
                              -0.6%    0%  0.3%     +1.2%
                                 |          |        |
         lower bound of the CI --'          |        |
sample mean (center of the CI) -------------'        |
         upper bound of the CI ----------------------'

As described above, a change is considered significant if the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD).

For instance, for an execution time metric, this confidence interval indicates a significantly worse performance:

----------------------------------------|---------|---(---------^---------)---------->
                                       0%        1%  1.3%      2.2%      3.1%
                                                  |   |         |         |
       significant impact threshold --------------'   |         |         |
                      lower bound of CI --------------'         |         |
       sample mean (center of the CI) --------------------------'         |
                      upper bound of CI ----------------------------------'

scenario:BM_TraceTinyCCSource

• 🟥 execution_time [+3.170ms; +3.334ms] or [+4.224%; +4.442%]

[datadog-datadog-prod-us1-2]

🎯 Code Coverage (details)
• Patch Coverage: 61.70%
• Overall Coverage: 88.22% (-2.61%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 20488e9 | Docs | Datadog PR Page | Was this helpful? React with 👍/👎 or give us feedback!}

[dmehala]

I took a quick look, and this PR is a hard no. There are many problems, with the most obvious being the custom YAML parser. I sincerely appreciate your effort to add feature to the tracer. Coding agents can be useful, but they need an experienced person to guide them properly. Without that, the result is just a pile of code that is hard to maintain and fix.

[bm1549]

I took a quick look, and this PR is a hard no. There are many problems, with the most obvious being the custom YAML parser. I sincerely appreciate your effort to add feature to the tracer. Coding agents can be useful, but they need an experienced person to guide them properly. Without that, the result is just a pile of code that is hard to maintain and fix.

@dmehala I was the one that nudged it towards hand-rolled yaml 😅 justification was because it was a 600KB dependency (8% of library size). After reviewing further you're right

Made a bunch of tweaks that improve it, but still happy to close if you think this is far from solving the problem correctly