[vjfridge]

Description

Motivation

PHP now emits flagevaluation track EVP events, but its current libdatadog
dependency does not attach the application's identity. This adopts
libdatadog #2716 so those
events identify the right application and are not combined across different
applications. Implements FFL-3439
and follows PHP #4288.

Dependent draft: #2716 is not merged. This currently pins its exact commit,
58b738075f440d4a1d94ea49a545a0fac2e52e2c. Before merging, adopt a merged
libdatadog revision containing the fix and revalidate any changes. Both PRs
are prerequisites for releasing this functionality.

Changes and key decisions

• Keep PHP's per-evaluation application context empty. The sidecar uses the
  application already registered on the connection, before aggregation.
• Require batch context.service, context.env, and context.version, plus
  context.dd.service on ordinary events. Do not add per-event environment or
  version tags, or restore customer context on protected/degraded events.
• Test default identity and changes to service, environment, and version
  before a single flush. Earlier evaluations must keep their original
  identity, even when every evaluation uses the same flag.
• Update the native pin, required Cargo dependencies, and generated bindings
  using the repository's normal build workflow.

Blast radius

• No new PHP setting, worker, queue, FFI call, or per-evaluation string copy.
  Privacy rules, evaluation results, exposures, and SDK-origin headers retain
  their existing behavior.
• The native update also includes intervening upstream changes, not only
  Enable sidecar on 8.3 #2716: CDN support, TLS roots, trace internals, and crashtracking. Linux
  ARM64 validation passed; supported-platform CI is still required.
• Header regeneration removes three Windows crash-callback declarations that
  upstream changed to the Windows calling convention. PHP does not call those
  callbacks directly; Windows coverage remains a CI requirement.

Testing

• PHP 8.2 NTS / Linux ARM64: optimized build and 23 PHPT tests pass;
  three require development services or ZTS. The three attribution regression
  tests fail against the old native dependency and pass with this update.
• Frozen system-tests, all five FFE files forced on: 51/53 pass, versus
  48/53 on master. All three consent modes pass, including application
  attribution and protected identity/context checks.
• Two system-test failures remain: a context fixture that leaves consent off
  but rejects the newly correct service tag (tracked in
  FFL-3407), and the 24,000-event
  stress test. Master delivered 23,734 and this branch 23,038; reported
  queue_overflow drops account for both shortfalls. This does not establish
  equivalent stress performance or lossless delivery.
• Frozen ffe-dogfooding: native PHP and OpenFeature both ready; 36/36
  successful evaluations and 36 OTel metrics on each build, with identical
  evaluation responses. All 36 new events have the correct application
  service. Live configuration enabled full consent; deterministic system
  tests cover absent/false consent separately.

Remaining before merge

• Enable sidecar on 8.3 #2716 merged; replace provisional native pin if necessary and validate.
• Supported-platform CI and reviewer approval, including native-update
  compatibility beyond the Linux ARM64 run.

Reviewer checklist

• Test coverage seems ok.
• Appropriate labels assigned.

[datadog-prod-us1-5]

 

✨ Unblock PR with BitsAI

❌ Errors

Your PR has failed checks. Please review the issues below and take necessary action before merging.

🚦 5 Pipeline jobs failed

DataDog/apm-reliability/dd-trace-php | merge-gate — 🔧 Needs a code fix, caused by this PR

View more details · View in GitLab

DataDog/apm-reliability/dd-trace-php | System Tests: [parametric]

View more details · View in GitLab

DataDog/apm-reliability/dd-trace-php | framework test: [flow]

View more details · View in GitLab

View all 5 failed jobs.

ℹ️ Info

No other issues found (see more)

🧪 All tests passed
❄️ No new flaky tests detected

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 74.40% (+0.04%)

Useful? React with 👍 / 👎

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: e9e88ca | Docs | View more details | Give us feedback!}

[pr-commenter]

Benchmarks [ tracer ]

Benchmark execution time: 2026-10-23 04:04:19

Comparing candidate commit e9e88ca in PR branch vickie/FFL-3439/php-flagevaluation-application-identity with baseline commit 04b8525 in branch master.

📊 Benchmarking dashboard

Found 1 performance improvements and 7 performance regressions! Performance is the same for 186 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:ContextPropagationBench/benchInject64Bit-opcache

• 🟥 execution_time [+95.202ns; +240.798ns] or [+4.121%; +10.424%]

scenario:EmptyFileBench/benchEmptyFileDdprof-opcache

• 🟩 execution_time [-295.584µs; -103.496µs] or [-6.809%; -2.384%]

scenario:MessagePackSerializationBench/benchMessagePackSerialization

• 🟥 execution_time [+8.430µs; +9.890µs] or [+7.979%; +9.361%]

scenario:MessagePackSerializationBench/benchMessagePackSerialization-opcache

• 🟥 execution_time [+7.200µs; +11.340µs] or [+6.519%; +10.268%]

scenario:SamplingRuleMatchingBench/benchRegexMatching1

• 🟥 execution_time [+75.248ns; +180.352ns] or [+5.187%; +12.432%]

scenario:SamplingRuleMatchingBench/benchRegexMatching2

• 🟥 execution_time [+97.109ns; +184.891ns] or [+6.715%; +12.786%]

scenario:SamplingRuleMatchingBench/benchRegexMatching3

• 🟥 execution_time [+83.122ns; +183.078ns] or [+5.695%; +12.543%]

scenario:SamplingRuleMatchingBench/benchRegexMatching4-opcache

• 🟥 execution_time [+114.147ns; +218.653ns] or [+7.153%; +13.702%]