GH-48616: [GLib] Use Arrow-${MAJOR}.${MINOR}.typelib not Arrow-1.0.typelib

[kou]

Rationale for this change

Our deb/RPM GLib library packages include version information such as libarrow2300-glib.deb and arrow2300-glib-libs.rpm. It's for installing multiple versions in the same system.

But all version library packages include Arrow-1.0.typelib. If we install multiple GLib library package versions in the same system, Arrow-1.0.typelib is conflicted.

If we use Arrow-${MAJOR}.${MINOR}.typelib not Arrow-1.0.typelib, we can really install multiple GLib library package versions in the same system.

What changes are included in this PR?

• Use XXX-${MAJOR}.${MINOR}.{gir,typelib} not XXX-1.0.{gir,typelib}
• Update release script to substitute versions automatically

Are these changes tested?

Yes.

Are there any user-facing changes?

Yes.

• GitHub Issue: [GLib] Use Arrow-${MAJOR}.${MINOR}.typelib not Arrow-1.0.typelib #48616

[github-actions]

⚠️ GitHub issue #48616 has been automatically assigned in GitHub to PR creator.

[hiroyuki-sato]

+1

I confirmed a installation result filenames like Arrow-23.0.typelib not Arrow-1.0.typelib
It will install multiple verion typelib files in the same directory.

/tmp/local/lib/girepository-1.0$ ls -1
Arrow-1.0.typelib
Arrow-23.0.typelib
ArrowDataset-1.0.typelib
ArrowDataset-23.0.typelib
ArrowFlight-1.0.typelib
ArrowFlight-23.0.typelib
ArrowFlightSQL-1.0.typelib
ArrowFlightSQL-23.0.typelib
Gandiva-1.0.typelib
Gandiva-23.0.typelib
Parquet-1.0.typelib
Parquet-23.0.typelib

Just for interest.
Does gobject-introspection finds the typelib properly?

[kou]

Yes. GObject Introspection uses the latest version by default: https://gnome.pages.gitlab.gnome.org/gobject-introspection/girepository/GIRepository.html#g-irepository-require

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit 9c4faee.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 1 possible false positive for unstable benchmarks that are known to sometimes produce them.