[jll63]

(Written by Claude Code, on behalf of @jll63.)

Fixes #139 (#139).

The library declared cmake_minimum_required(VERSION 3.8...3.20), but the tests use target_precompile_headers, which needs CMake 3.16. As a result, configuring the superproject with -DBUILD_TESTING=ON failed on CMake 3.8 through 3.15 (boostorg/cmake run 41556485068).

This PR raises the declared minimum to 3.16 instead of making the tests work on older versions:

• 3.16 is what was tested. Boost.CI's minimum-CMake job ran 3.16 until Set the minimum CMake version based on the version specified in the CMakeLists.txt boost-ci#355 (2026-11-05) made it read the version from CMakeLists.txt. Since then it tries 3.8, which fails before reaching our code: CMake 4 refuses to build CMake 3.8, because CMake 3.8 declares cmake_minimum_required(VERSION 2.8.12.2).
• No supported environment needs anything older. CMake 3.8 dates from April 2017 and was never shipped by any Ubuntu release (17.04 had 3.7.2, 17.10 had 3.9.1). Ubuntu 20.04 ships 3.16.3. CI already excludes gcc ≤ 9, and every distribution that ships gcc 10 also ships CMake 3.16 or newer.
• The boostorg/cmake CI then skips openmethod below 3.16, because its script excludes a library whose declared minimum is above the CMake under test.

The change is two cmake_minimum_required lines (root and test/cmake_test).

The failing Min. CMake job

The Min. CMake, 3.16 job fails, but not because of this change: it fails before configuring anything of ours. Since ubuntu-latest moved to ubuntu-26.04, the runner's CMake 4 cannot build CMake 3.16 from source either, because it rejects the old cmake_minimum_required in 3.16's Source/kwsys/CMakeLists.txt. The same job passed on develop on 2026-11-03, when it still ran on ubuntu-24.04. The CMake cache missed in both runs, so the image is the only thing that changed. Every library using the reusable workflow will hit this until Boost.CI adds -DCMAKE_POLICY_VERSION_MINIMUM=3.5 to its CMake build flags.

Testing

• CMake 3.16.0 (the version Boost.CI's min-CMake job runs): superproject configure with -DBUILD_TESTING=ON -DBOOST_INCLUDE_LIBRARIES=openmethod, Debug, PCH generated, 142/142 openmethod tests pass.
• CMake 3.28, standalone build: 215/215 tests pass.
• CMake 3.15.7: the boostorg/cmake exclusion check reads the declared 3.16 and skips the library.

🤖 Generated with Claude Code

[cppalliance-bot]

An automated preview of the documentation is available at https://144.openmethod.prtest3.cppalliance.org/libs/openmethod/doc/html/index.html

If more commits are pushed to the pull request, the docs will rebuild at the same URL.

2026-11-07 17:22:11 UTC

[jll63]

@pdimov I have to wait Nov 12 (Boost 1.93.0 open for bug fixes) to merge into master, correct? And request permission from release managers?

[jll63]

Blocked by boostorg/boost-ci#358

[pdimov]

I have to wait Nov 12 (Boost 1.93.0 open for bug fixes) to merge into master, correct? And request permission from release managers?

Correct.

[pdimov]

While this is the correct long term solution, I would prefer if we fix that now by checking the CMake version around the target_precompile_headers calls. We will revisit this after the release, when if there are no objections, we'll raise the Boost-wide CMake minimum to 3.16 as discussed on the list.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.57%. Comparing base (b681217) to head (81db2f3).
⚠️ Report is 24 commits behind head on develop.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop     #144   +/-   ##
========================================
  Coverage    92.56%   92.57%           
========================================
  Files           29       29           
  Lines         2314     2316    +2     
  Branches       699      699           
========================================
+ Hits          2142     2144    +2     
  Misses         111      111           
  Partials        61       61           

---

Continue to review full report in Codecov by Harness.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 456d95c...81db2f3. Read the comment docs.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.