GOV: adopt EffVer

[tacaswell]

EffVer very closely articulates what we are already doing.

This was discussed on the developer call last week and was positively received.

I used git grep to find all '* version' and * release referring to our releases (I left "minor version" in most places in the minimum dependency document when referring to other projects, but expanded the definition at the top).

---

I would like approval from 2 other steering council members before we merge.

[timhoffm]

Approved with some small changes still needed.

[WeatherGod]

A thought. Given our commitment towards backwards compatibility, we typically don't have releases that break things for many users or require significant amounts of effort to update their codes to use newer versions of matplotlib. Instead, our focus tends to be bugfixes and new features that complements the existing feature set. The minor version number change has always indicated to me the growing set of features this library has, rather than it being an indication of how much work it would be to upgrade my code to it. Has there even been indications from our users that they are confused by our current versioning system?

…

On Thu, Jan 25, 2024 at 1:59 PM Thomas A Caswell ***@***.***> wrote: EffVer very closely articulates what we are already doing. This was discussed on the developer call last week and was positively received. I used git grep to find all '* version' and * release referring to our releases (I left "minor version" in most places in the minimum dependency document when referring to other projects, but expanded the definition at the top). ------------------------------ I would like approval from 2 other steering council members before we merge. ------------------------------ You can view, comment on, or merge this pull request online at: #27702 Commit Summary - 1faa935 <1faa935> GOV: adopt EffVer File Changes (10 files <https://github.com/matplotlib/matplotlib/pull/27702/files>) - *M* doc/conf.py <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-e170e9a7d787c21095c6c11bb25f0f1ff0294a42a46d45ba6fb5ed794e457624> (2) - *M* doc/devel/MEP/MEP13.rst <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-5a2bb663ebb0477b4314d302ebcc654a0dbe401cb5c2f7fa36bb961333997f20> (2) - *M* doc/devel/api_changes.rst <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-d3658c3a2a97d8a37fde88708a9bf3208620f7ea8a1128a264c2eb16115ca404> (4) - *M* doc/devel/min_dep_policy.rst <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-a908cd6e87c8e5edf5abe9e5526e47d3517f839541e99bcccc466ee21152cca0> (9) - *M* doc/devel/pr_guide.rst <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-49b0e5f1188e20843061c77284b1610369283c20f25b03edfb7db39825721938> (18) - *M* doc/devel/release_guide.rst <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-4ca57718a84f31a4b2a993ffd096debd80f3c31f72cfc22d8c131a668492ea11> (73) - *M* doc/devel/triage.rst <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-54ab909f213f3bf9cabc13ad7e91351ce9b75722a0df5603f19849ee50c92734> (2) - *M* lib/matplotlib/_api/deprecation.py <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-bd9e804b0bba997a0a496a274e7a0ba9132190d5f5bb23190ef0b890a6a5c66b> (2) - *M* src/ft2font_wrapper.cpp <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-5a4638a8d55cba783f8c2dfea1847d86fb17267bc9cf044afc66a51270dd9879> (4) - *M* tools/github_stats.py <https://github.com/matplotlib/matplotlib/pull/27702/files#diff-284a1b462cd74af2e6791e783037c3c1cd6c141508b2f16716ebb3153aff6b4b> (2) Patch Links: - https://github.com/matplotlib/matplotlib/pull/27702.patch - https://github.com/matplotlib/matplotlib/pull/27702.diff — Reply to this email directly, view it on GitHub <#27702>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACHF6C6WJW4PGRSFHN3CNDYQKTQJAVCNFSM6AAAAABCLAQY32VHI2DSMVQWIX3LMV43ASLTON2WKOZSGEYDAOJZGEYDMNY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[timhoffm]

IMHO the point is that we haven't really made any commitment to what our version numbers mean. It's definitively not SemVer because we make breaking changes in minor releases. Even though that happens with a defensive compatibility policy (deprecation periods and not too disruptive changes), these would require a major release in SemVer.

You can only encode one kind of information in the version number, not both new features and degree of compatibility. And chosing between these two compatibility is the more helpful one. I think in practice, we'll continue to have both new features and minor API changes in the minor/meso releases.

While we could define our own meaning for the version numbers, it's better to use a common semantic. Since SemVer is out and EffVer essentially matches our compatibility policy, it's a good choice.

[efiring]

One typo, and several places where it looks like
"major" should be "macro" for consistency. Otherwise, fine with me.

[tacaswell]

We have 2 positive SC reviews so I'm going to go ahead and merge this.