GH-44384: [C++] Use CMAKE_LIBTOOL on macOS

[tomjakubowski]

When a builder sets CMAKE_LIBTOOL, use that as the program to bundle dependencies. This matches the behavior of the Windows build.

Also make a nitpicky minor update to the error message when a non-Apple libtool is detected.

Are these changes tested?

I did not add a test case for this build configuration. I confirmed the behavior when CMAKE_LIBTOOL is set, using this script on a Mac. It creates an ad-hoc libtool wrapper and then passes it as CMAKE_LIBTOOL.

#!/bin/bash

mylibtool=$(mktemp -d)/libtool.sh
cat << 'EOF' > "$mylibtool"
#!/bin/bash
say -v Tom -- "voice-enhanced libtool:" "$@"
/usr/bin/libtool "$@"
EOF
chmod +x "$mylibtool"

cmake -GNinja -DARROW_DEPENDENCY_SOURCE=BUNDLED -DCMAKE_LIBTOOL="$mylibtool" -S cpp -B build/
cmake --build build/

I am not sure how to best add a test for this to the repository.

Are there any user-facing changes?

This changes how bundled builds behave when an Apple build-user sets CMAKE_LIBTOOL in their CMake invocation. I think the change is a sensible one and wouldn't be surprising to users, but it may break existing builds in environments where /usr/bin/libtool is Apple's libtool, and CMAKE_LIBTOOL is set to a non-Apple one.

• GitHub Issue: [C++] Bundled Mac C++ build fails when /usr/bin/libtool is an invalid libtool #44384

[github-actions]

⚠️ GitHub issue #44384 has been automatically assigned in GitHub to PR creator.

[kou]

+1

Could you fix a lint error by pre-commit run --show-diff-on-failure --color=always --all?

[tomjakubowski]

Updated the formatting and got myself set up with pre-commit. Thanks for the pointer!

[tomjakubowski]

C GLib & Ruby / AMD64 Ubuntu 22.04 GLib & Ruby (pull_request) Failing after 10m

  connect to 140.82.113.4 port 443 failed: Connection timed out

  Failed to connect to github.com port 443 after 135169 ms: Connection timed
  out

looks like a network outage maybe?

R / Check minimum supported Arrow C++ Version (15.0.2) (pull_request) Failing after 25s

E: Version '15.0.2-1' for 'libarrow-dev' was not found
E: Version '15.0.2-1' for 'libarrow-acero-dev' was not found
E: Version '15.0.2-1' for 'libparquet-dev' was not found
E: Version '15.0.2-1' for 'libarrow-dataset-dev' was not found

Not sure what that one is about.

[tomjakubowski]

two other build failures also seem unrelated (they have the same error):

C++ / AMD64 Conda C++ AVX2 (pull_request) Failing after 22s

C++ / AMD64 Ubuntu 22.04 C++ ASAN UBSAN (pull_request)

Reading state information...
python3 is already the newest version (3.12.3-0ubuntu2).
python3-dev is already the newest version (3.12.3-0ubuntu2).
python3-pip is already the newest version (24.0+dfsg-1ubuntu1).
0 upgraded, 0 newly installed, 0 to remove and 32 not upgraded.
error: externally-managed-environment

× This environment is externally managed
╰─> To install Python packages system-wide, try apt install
    python3-xyz, where xyz is the package you are trying to
    install.
    
    If you wish to install a non-Debian-packaged Python package,
    create a virtual environment using python3 -m venv path/to/venv.
    Then use path/to/venv/bin/python and path/to/venv/bin/pip. Make
    sure you have python3-full installed.
    
    If you wish to install a non-Debian packaged Python application,
    it may be easiest to use pipx install xyz, which will manage a
    virtual environment for you. Make sure you have pipx installed.
    
    See /usr/share/doc/python3.12/README.venv for more information.

note: If you believe this is a mistake, please contact your Python installation or OS distribution provider. You can override this, at the risk of breaking your Python installation or OS, by passing --break-system-packages.

[kou]

The GLib failure is just a network error. It'll be fixed by re-run.

Other failures are already fixed on main.

[kou]

+1

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 25a2440.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 19 possible false positives for unstable benchmarks that are known to sometimes produce them.