chore(profiling-315): compile natives on 3.15 and add allow_failure job

[vlad-scherbich]

prev: #19269 | next: #19272

Description

Delta vs #19269.

setup.py lifts the four profiling / crashtracker / memalloc / _threading compile gates from < (3, 15) to < (3, 16), and setdefaults PYO3_USE_ABI3_FORWARD_COMPATIBILITY=1 on 3.15. profiling_native_py315 is a separate allow_failure job until the profiling_native image ships 3.15.

Does not add collector stubs, the asyncio current_task() swallow, the PY_UNWIND fallback, or riot 3.15 schedules.

Layer	Meaning	Which PR
Compiled	setup.py builds profiling natives on 3.15	this PR
Armed	profiling_native_py315 allow_failure	this PR
Observable	—	—

Testing

Risks

Additional Notes

CI chore, changelog/no-changelog. Leftovers extracted to sibling/child drafts.

Contributes to #17817

Pre-review checklist

• The changes have been tested in real services in staging
• The PR does not result in new crashes

[datadog-prod-us1-5]

🎉 All green!

🧪 All tests passed
❄️ No new flaky tests detected

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 33d1ad8 | Docs | View more details | Give us feedback!}

[cit-pr-commenter-54b7da]

Circular import analysis

⚠️ Existing circular imports

There are 3 circular imports that already exist on the base branch and have not been changed by this PR.

ddtrace.llmobs -> ddtrace.llmobs._evaluators -> ddtrace.llmobs._evaluators.format -> ddtrace.llmobs._experiment -> ddtrace.llmobs
ddtrace.errortracking._handled_exceptions.bytecode_injector -> ddtrace.errortracking._handled_exceptions.callbacks -> ddtrace.errortracking._handled_exceptions.collector -> ddtrace.errortracking._handled_exceptions.bytecode_reporting -> ddtrace.errortracking._handled_exceptions.bytecode_injector
ddtrace.appsec._asm_request_context -> ddtrace.appsec._iast._iast_request_context_base -> ddtrace.appsec._iast._iast_env -> ddtrace.appsec._iast.reporter -> ddtrace.appsec._exploit_prevention.stack_traces -> ddtrace.appsec._asm_request_context

[cit-pr-commenter-54b7da]

Codeowners resolved as

Resolved from the full PR diff against vlad/ddtracepy-315-profiling-native using the target branch CODEOWNERS file.
CODEOWNERS team requests not listed below are not required by the current file set.

.gitlab-ci.yml                                                          @DataDog/python-guild @DataDog/apm-core-python
setup.py                                                                @DataDog/python-guild

[cit-pr-commenter-54b7da]

Dependency direction analysis

⚠️ Existing dependency direction violations

There are 230 dependency direction violations that already exist on the base branch and have not been changed by this PR.

Show existing violations (showing 5 of 230 highest severity)

ddtrace.internal.tracemethods -×-> ddtrace.trace  (internal-core -> product:tracing, score=134)
ddtrace.internal.ci_visibility.api._base -×-> ddtrace.trace  (product:ci_visibility -> product:tracing, score=132)
ddtrace.internal.openfeature._span_enrichment -×-> ddtrace.trace  (product:openfeature -> product:tracing, score=132)
ddtrace.internal.ci_visibility.recorder -×-> ddtrace.trace  (product:ci_visibility -> product:tracing, score=132)
ddtrace.internal.ci_visibility.filters -×-> ddtrace.trace  (product:ci_visibility -> product:tracing, score=132)

To see all violations, download the layers-base.json and layers-pr.json artifacts from this CI job and run:

uv run --script scripts/import-analysis/layers.py compare layers-base.json layers-pr.json

[pr-commenter]

Benchmarks

Benchmark execution time: 2026-09-02 19:47:18

Comparing candidate commit 33d1ad8 in PR branch vlad/ddtracepy-315-profiling-collectors with baseline commit be2ca2c in branch vlad/ddtracepy-315-profiling-native.

📊 Benchmarking dashboard

Found 0 performance improvements and 5 performance regressions! Performance is the same for 579 metrics, 10 unstable metrics, 1 known flaky benchmarks, 17 flaky benchmarks without significant changes.

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:httppropagationinject-ids_only

• 🟥 execution_time [+1.916µs; +2.029µs] or [+11.812%; +12.505%]

scenario:iastaspects-rstrip_aspect

• 🟥 execution_time [+72.795µs; +78.596µs] or [+19.005%; +20.520%]

scenario:iastaspectsospath-ospathbasename_aspect

• 🟥 execution_time [+109.823µs; +117.760µs] or [+26.148%; +28.038%]

scenario:telemetryaddmetric-1-count-metric-1-times

• 🟥 execution_time [+396.509ns; +437.239ns] or [+14.926%; +16.459%]

scenario:tracer-small

• 🟥 execution_time [+37.856µs; +39.583µs] or [+11.284%; +11.798%]

Unstable benchmarks

These benchmarks have a confidence interval too wide to call a change; treat them as noise rather than signal.

scenario:coreapiscenario-context_with_data_listeners

• unstable execution_time [-790.379ns; +705.217ns] or [-7.090%; +6.326%]

scenario:coreapiscenario-core_dispatch_1_listener

• unstable execution_time [-34.470ns; +34.534ns] or [-5.534%; +5.544%]

scenario:coreapiscenario-core_dispatch_50_listeners

• unstable execution_time [-1694.281ns; +1640.477ns] or [-9.823%; +9.511%]

scenario:coreapiscenario-core_dispatch_exception_listeners

• unstable execution_time [-1229.885ns; +1224.251ns] or [-9.551%; +9.507%]

scenario:coreapiscenario-core_dispatch_listeners

• unstable execution_time [-338.974ns; +323.037ns] or [-9.090%; +8.663%]

scenario:coreapiscenario-core_dispatch_no_args_listeners

• unstable execution_time [-259.873ns; +257.265ns] or [-8.800%; +8.712%]

scenario:coreapiscenario-core_dispatch_with_results_1_listener

• unstable execution_time [-72.727ns; +75.804ns] or [-6.209%; +6.471%]

scenario:coreapiscenario-core_dispatch_with_results_50_listeners

• unstable execution_time [-4119.011ns; +3970.253ns] or [-10.030%; +9.667%]

scenario:coreapiscenario-core_dispatch_with_results_listeners

• unstable execution_time [-750.429ns; +794.160ns] or [-9.203%; +9.740%]

scenario:packagesupdateimporteddependencies-import_many_stdlib_cached

• unstable execution_time [-47.095µs; +29.878µs] or [-8.162%; +5.178%]

Known flaky benchmarks

These benchmarks are marked as flaky and will not trigger a failure. Modify FLAKY_BENCHMARKS_REGEX to control which benchmarks are marked as flaky.

scenario:span-start

• 🟥 execution_time [+1.529ms; +1.660ms] or [+10.886%; +11.821%]

Known flaky benchmarks without significant changes:

• scenario:errortrackingflasksqli-baseline
• scenario:flasksimple-iast-get
• scenario:iastaspects-casefold_aspect
• scenario:iastaspects-casefold_noaspect
• scenario:iastaspects-index_aspect
• scenario:iastaspects-ljust_noaspect
• scenario:iastaspects-lower_aspect
• scenario:iastaspects-replace_aspect
• scenario:iastaspects-swapcase_aspect
• scenario:iastaspects-title_noaspect
• scenario:iastaspects-translate_aspect
• scenario:iastaspects-translate_noaspect
• scenario:iastaspects-upper_noaspect
• scenario:packagespackageforrootmodulemapping-cache_off
• scenario:packagespackageforrootmodulemapping-cache_on
• scenario:sethttpmeta-all-enabled
• scenario:telemetryaddmetric-record-100-metrics

[Copilot]

Pull request overview

Updates profiling collectors and CI/build wiring to better support Python 3.15, including making collector imports resilient when profiling Cython extensions are not present and extending the riot/CI matrices and lockfiles to exercise 3.15.

Changes:

• Add ImportError fallbacks/stubs for profiling collectors that depend on Cython extensions (_lock, _task, _exception) so profiling can be imported and collectors can be gracefully skipped when extensions aren’t available.
• Update setup.py gating and environment to allow building profiling/crashtracker natives on Python 3.15 (incl. PyO3 ABI3 forward-compat safety net).
• Extend riot/lockfiles and CI configuration to include Python 3.15 coverage for relevant suites/jobs.

Reviewed changes

Copilot reviewed 17 out of 17 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
tests/profiling/test_scheduler.py	Stabilizes scheduler logging test by patching upload and adds type hints.
setup.py	Enables profiling-related native builds through 3.15 and sets PyO3 ABI3 forward-compat env var on 3.15+.
scripts/requirements_to_csv.py	Improves typing (untyped toml) and normalizes CSV line endings.
riotfile.py	Adds/selects 3.15 in specific venv definitions and updates a protobuf incompat comment.
ddtrace/profiling/collector/threading.py	Adds ImportError fallback stubs when _lock extension isn’t available.
ddtrace/profiling/collector/asyncio.py	Adds ImportError fallback stubs when _lock extension isn’t available.
ddtrace/profiling/collector/exception.py	Adds ImportError fallback stub for _exception extension.
ddtrace/profiling/collector/stack.py	Adds ImportError fallback stub for _task extension to keep StackCollector importable.
ddtrace/profiling/_asyncio.py	Adjusts task-linking wrappers to avoid failures when called outside a running loop.
ddtrace/internal/monitoring.py	Adds a guarded fallback when PY_UNWIND is rejected as a local monitoring event.
.riot/requirements/e26245b.txt	Adds a newly generated Python 3.15 lockfile.
.riot/requirements/95077af.txt	Adds a newly generated Python 3.15 lockfile.
.riot/requirements/222bcd0.txt	Adds a newly generated lockfile (currently contains merge-conflict markers).
.riot/requirements/1c6cb02.txt	Adds a newly generated Python 3.15 lockfile.
.gitlab/templates/build-base-venvs.yml	Extends artifact paths for profiling-native test outputs.
.gitlab-ci.yml	Adds an allow-failure profiling native job for Python 3.15 pending base-image support.
.github/workflows/generate-package-versions.yml	Adds Python 3.15 to the workflow’s installed Python set.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[emmettbutler]

Does this have a regression test?

[emmettbutler]

Why's this necessary?

[emmettbutler]

Suggested change

	pys=select_pys(min_version="3.12", max_version="3.14") + ["3.15"],
	pys=select_pys(min_version="3.12", max_version="3.15"),

[emmettbutler]

Why do these need to be passed as artifacts? Are they generated dynamically?

[vlad-scherbich]

Folded into #19269. GitHub marked this MERGED because #19269 fast-forwarded to the same SHA — not merged to main.