FIX: Preserve Python-specific release dependencies

[gargsaumya]

Work Item / Issue Reference

AB#47861

---

Summary

Preserves Python-version environment markers when release dependency locks are refreshed, preventing the Python 3.10-only asyncio backport from being selected on Python 3.11 and newer.

Changes:

• Compiles Linux, macOS, and Windows release locks with universal marker preservation.
• Leaves the ODBC lock generation path unchanged.
• Regenerates affected locks without upgrading existing package versions and retains SHA256 hashes only.
• Adds focused regression tests for universal generation and the asyncio backport marker.
• Preserves Gaurav's fix to skip profiler contracts when repository-only profiler tools are absent from isolated wheel tests.
• Supplies the narrow set of repository-side mssql-odbc runner contract files to isolated Linux wheel tests and reports runner stdout/stderr if the status artifact is absent.
• Excludes source-only release-lock/workflow contracts from both manylinux and musllinux installed-wheel suites, with a source-checkout contract guarding both selections.

Validation

• Release dependency contract module: 3 passed, including the exact asyncio backport regression.
• Linux pipeline contract tests: 5 passed; the complete runner/pipeline module was 5 passed, 5 skipped on Windows because runner execution requires Linux.
• Reproduced the isolated fixture failure before the fix; after the fix, pipeline-equivalent collection excluded the repository contract module while an installed-wheel sentinel passed.
• Targeted dependency-lock and runner contract tests: 12 passed on Linux in the prior follow-up.
• Isolated Linux wheel-test fixture simulation: 10 passed, 1 profiler module skipped as intended in the prior follow-up.
• Budget-exhaustion runner test repeated 50 times on Linux in the prior follow-up: 50 passed.
• Black (line length 100), whitespace, and YAML parse checks: passed.
• Lock regeneration with repository-pinned uv 0.12.6 repeated byte-for-byte: passed.
• Production-like pip dry runs with require-hashes on Windows Python 3.10 through 3.14: passed.
• Confirmed Python 3.10 selects the backport and Python 3.11 through 3.14 do not.
• Simulated macOS and manylinux x64 resolution on Python 3.10 through 3.14: passed; the ADO native build matrix remains the final validation.

[github-actions]

PR Performance Report

✅ No regression detected

No consistent slowdowns detected across all 2 environments.

0 IMPROVEMENTS 0 SLOWDOWNS 2/2 ENVIRONMENTS

Coverage: 2 of 2 environments completed. Advisory result; does not block merging.

Performance diagnostics

Phase times are inclusive diagnostics and must not be added together. They identify where measured time changed, not why it changed.

No affected phases or call-count changes were recorded.

All database tasks and timings

Unix / SQL Server 2022

Database task	Before	After	Paired change	Result
Connection opening	10.571 ms	10.460 ms	-1.0%	no signal
SELECT queries	1.102 ms	1.091 ms	-1.0%	no signal
Row insertion	34.871 ms	34.786 ms	-0.8%	no signal
Executemany inserts	161.224 ms	160.796 ms	-1.0%	no signal
Fetch-all queries	120.984 ms	122.055 ms	+0.4%	no signal
Row-by-row fetching	14.321 ms	14.287 ms	+1.1%	no signal
Batched row fetching	117.170 ms	117.983 ms	-0.1%	no signal
Transaction commit and rollback	115.997 ms	116.010 ms	+0.1%	no signal
Arrow row fetching	94.447 ms	96.402 ms	+1.8%	no signal
100,000-row insertion	445.381 ms	447.466 ms	-0.2%	no signal
Row fetching in batches of 100	121.036 ms	122.602 ms	+1.2%	no signal
Row fetching in batches of 10,000	137.735 ms	127.672 ms	-4.2%	no signal
Repeated positional queries	34.548 ms	34.414 ms	-0.4%	no signal
Repeated named-parameter queries	36.912 ms	36.343 ms	-3.6%	no signal
Legacy 100,000-row insertion	353.173 ms	354.023 ms	-1.3%	no signal
Insertion with explicit input sizes	490.917 ms	496.298 ms	+1.1%	no signal
Joined aggregation queries	178.082 ms	183.336 ms	+3.8%	no signal
Large joined-result fetching	184.617 ms	191.787 ms	+3.6%	no signal
1.2-million-row fetching	3525.263 ms	3480.431 ms	-2.8%	no signal
Common table expression queries	5.319 ms	5.343 ms	+0.7%	no signal
256 KiB VARCHAR(MAX) / fetchall()	1.349 ms	1.376 ms	+2.5%	no signal
10,000 scalar values / fetchval() (debug disabled)	106.309 ms	106.602 ms	-0.0%	no signal

Unix / SQL Server 2025

Database task	Before	After	Paired change	Result
Connection opening	97.560 ms	96.503 ms	-1.1%	no signal
SELECT queries	1.047 ms	1.063 ms	+1.1%	no signal
Row insertion	34.532 ms	34.561 ms	-0.4%	no signal
Executemany inserts	150.357 ms	151.198 ms	+0.8%	no signal
Fetch-all queries	120.464 ms	123.125 ms	+1.8%	no signal
Row-by-row fetching	14.506 ms	14.458 ms	+0.3%	no signal
Batched row fetching	120.764 ms	117.779 ms	-2.5%	no signal
Transaction commit and rollback	117.255 ms	115.248 ms	-1.4%	no signal
Arrow row fetching	93.493 ms	95.277 ms	+2.8%	no signal
100,000-row insertion	427.952 ms	448.047 ms	+5.9%	no signal
Row fetching in batches of 100	119.779 ms	126.649 ms	+4.4%	no signal
Row fetching in batches of 10,000	138.649 ms	129.772 ms	-8.4%	no signal
Repeated positional queries	33.522 ms	33.950 ms	+1.0%	no signal
Repeated named-parameter queries	36.392 ms	36.624 ms	+0.8%	no signal
Legacy 100,000-row insertion	350.879 ms	357.941 ms	+1.0%	no signal
Insertion with explicit input sizes	485.640 ms	482.053 ms	-1.1%	no signal
Joined aggregation queries	166.547 ms	160.066 ms	-0.4%	no signal
Large joined-result fetching	180.534 ms	183.133 ms	+1.3%	no signal
1.2-million-row fetching	3528.892 ms	3539.988 ms	-1.1%	no signal
Common table expression queries	5.139 ms	5.126 ms	+0.2%	no signal
256 KiB VARCHAR(MAX) / fetchall()	1.455 ms	1.494 ms	-1.9%	no signal
10,000 scalar values / fetchval() (debug disabled)	105.756 ms	107.244 ms	+0.9%	no signal

Build and measurement details

ADO build 180846

PR head: b5c11999461079f986a746471de5a5951f44c572
Base: 666f3cb6d23981bb23cd182ec273df10a7b2c805
Measured merge: 85e4eed54f937849036e63e3625102f6b4c824be

• Unix / SQL Server 2022: Python 3.12.3, x86_64, SQL 16.0.4295.3; 5 paired comparisons and 1 warmup.
• Unix / SQL Server 2025: Python 3.12.3, x86_64, SQL 17.0.5005.3; 5 paired comparisons and 1 warmup.

A consistent change requires more than 20% median paired movement, at least 1 ms between the median runtimes, and at least 80% of pairs exceeding the relative threshold in the same direction. A slowdown without enough pair agreement is reported as inconsistent.

The displayed change is the median of paired before-and-after ratios. It is not recalculated from the two displayed median runtimes.

Both revisions use profiling-enabled builds on the same agent and database, with alternating order and discarded warmups. Results are diagnostic and do not represent production-wheel latency.

Raw samples and logs are attached to the ADO run as profiler-* artifacts.

[Copilot]

Copilot review overview

🟡 Changes recommended

The new regression tests contain a cross-entry false positive and hard-code a dependency version that future lock refreshes may legitimately upgrade.

Review effort: Balanced
Findings: 2

Open (2)

• Bound assertion to the macOS matrix item · New
• Avoid hard-coding the version in the marker assertion · New

What changed in this PR

Preserves conditional dependencies across supported Python versions during release-lock generation.

Changes:

• Enables universal marker preservation for Linux, macOS, and Windows locks.
• Regenerates affected dependency locks with environment markers.
• Adds regression tests for workflow configuration and asyncio backport selection.

File	Description
.github/​workflows/​refresh-build-dependencies.yml	Enables universal compilation while excluding ODBC.
eng/​requirements-build-linux.txt	Adds preserved platform and Python markers.
eng/​requirements-test-linux.txt	Preserves conditional test dependencies.
eng/​requirements-build-macos.txt	Preserves macOS lock environment markers.
eng/​requirements-build-windows.txt	Preserves Windows lock environment markers.
tests/​test_release_build_dependencies.py	Adds static regression coverage for lock generation.

---

💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[github-actions]

📊 Code Coverage Report

🔥 Diff Coverage 100%

🎯 Overall Coverage 85%

📈 Total Lines Covered: 9445 out of 11094 📁 Project: mssql-python

---

Diff Coverage

Diff: main...HEAD, staged and unstaged changes

No lines with coverage information in this diff.

---

📋 Files Needing Attention

📉 Files with overall lowest coverage (click to expand)

mssql_python.pybind.performance_counter.hpp: 0.7%
mssql_python.pybind.logger_bridge.cpp: 57.9%
mssql_python.pybind.ddbc_bindings.h: 62.6%
mssql_python.pybind.logger_bridge.hpp: 70.8%
mssql_python.pybind.connection.connection_pool.cpp: 82.3%
mssql_python.pybind.connection.connection.cpp: 83.1%
mssql_python.logging.py: 86.2%
mssql_python.pooling.py: 90.1%
mssql_python.pybind.fetch_temporal.hpp: 92.1%
mssql_python.cursor.py: 92.5%

---

🔗 Quick Links

⚙️ Build Summary	📋 Coverage Details
View Azure DevOps Build	Browse Full Coverage Report

[Copilot]

Copilot review overview

🟡 Changes recommended

Both isolated Linux wheel-test paths omit lock files required by the new regression test, causing FileNotFoundError.

Review effort: Balanced
Findings: 1 · 2

Open (3)

• Include lock files in manylinux fixture · New
• Avoid hard-coding the version in the marker assertion
• Bound assertion to the macOS matrix item

[Copilot]

Warning

Copilot couldn't run its full agentic review because it didn't start before the timeout. Make sure your repository has a runner available, or add a copilot-code-review.yml file specifying one with the runs-on attribute. See the docs for more details.

Copilot review overview

Review effort: Lite
Findings: 1 · 4 · 1

Open (6)

• Include lock files in manylinux fixture
• next(...) will raise StopIteration with a less actionable error if the expected compile line… · New
• These DOTALL regexes can become brittle and accidentally match across unrelated YAML sections as… · New
• Avoid hard-coding the version in the marker assertion
• Bound assertion to the macOS matrix item
• This assertion includes a trailing space, so it will miss variants like cp -r /workspace/eng (no… · New

[Copilot]

Copilot review overview

🔵 Needs a closer look

The new regression tests can miss incorrect matrix configuration and unnecessarily block future dependency upgrades.

Review effort: Balanced
Findings: 4 · 1

Open (5)

• These DOTALL regexes can become brittle and accidentally match across unrelated YAML sections as…
• next(...) will raise StopIteration with a less actionable error if the expected compile line…
• Avoid hard-coding the version in the marker assertion
• Bound assertion to the macOS matrix item
• This assertion includes a trailing space, so it will miss variants like cp -r /workspace/eng (no…

Resolved since last review (1)

• Include lock files in manylinux fixture

[Copilot]

Copilot review overview

🔵 Needs a closer look

Cross-platform release-lock and production pipeline changes require final matrix validation and human approval.

Review effort: Balanced
Findings: None

Resolved since last review (6)

• Include lock files in manylinux fixture
• These DOTALL regexes can become brittle and accidentally match across unrelated YAML sections as…
• next(...) will raise StopIteration with a less actionable error if the expected compile line…
• Avoid hard-coding the version in the marker assertion
• Bound assertion to the macOS matrix item
• This assertion includes a trailing space, so it will miss variants like cp -r /workspace/eng (no…