Re: Status of CMake for 1.15?

Den ons 22 juli 2026 kl 13:10 skrev Evgeny Kotkov via dev <
[email protected]>:

> Hi all,
>
> This came up in one of the recent threads [1]: it seems that we haven't
> agreed on the official status of CMake for 1.15.
>
> From the RM standpoint, this is something we should settle before the final
> release, or even earlier, because we would probably want to describe that
> in the release notes.
>

Yes, please lets make sure we agree on a status. I think we should also
advertise the CMake build system in the relase notes (that was my intention
behind the #cmake anchor, however life got in the way before I found time
to do this).


> So far, only one thing has been discussed and decided: we plan to deprecate
> vcxproj in 1.15, and replace it with CMake in 1.16, if all goes well [2,
> 3].
>
> That leaves a couple of open questions:
>
> 1) What do we call the current status of CMake (e.g., in the release
> notes)?
>
> 2) Do we want to mark CMake experimental for 1.15 or not?
>
> My personal opinion on (1) is that we should call CMake "supported",
> meaning
> that we commit to supporting it in the future and tell people that it's
> here
> to stay.
>
> This should encourage maintainers to use it and report issues, and might
> help
> us keep the pace on its development.  We could note somewhere that we know
> the current support is not 100% feature-complete, but that it is on our
> radar.
> At this point, we won't position it as a replacement for anything other
> than
> vcxproj.
>

+1

Please note that INSTALL indicate that CMake will be the default build
system in 1.16 (it doesn't say which platforms). I wrote this many moons
ago because I understood that to be consensus at that time. Lets figure out
how to update this before the final release.


>
> My opinion on (2) follows from the above: I think we should not declare
> CMake
> experimental, for several reasons:
>
> - The "experimental" status of a feature can alienate people and prevent
> them
>   from using it.
>
> - It can reduce the development momentum, because an experimental feature
> is
>   easy to abandon.
>
> - It can complicate the eventual exit from the experimental status due to
>   a lack of clear conditions for such exit.
>
> In short, I think that we should make at least a small commitment about the
> CMake's status so that people would actually use it, and so that we would
> have reasons to keep supporting and improving it.
>

+1

I don't think we need to have the CMake build system output "EXPERIMENTAL"
anywhere. The bug in [1] was clearly already found but unfortunately not
backported to the RC.

Cheers,
Daniel



>
> Thoughts?
>
> [1]: https://lists.apache.org/thread/g2997y8ygfr4vg0z5yrr33ojw1ysdbqh
> [2]: https://subversion.apache.org/docs/release-notes/1.15.html#vcproj
> [3]: https://lists.apache.org/thread/4bxqms1xrp30jro0v4hydkypzslxjo5v
>
>
> Thanks,
> Evgeny Kotkov
>