Publish releases for tags only

[danicheg]

That's previous behavior, so I think we should go with it right now.

[armanbilge]

Do you mind if I ask why? I think publishing snapshots is a trivially small cost, and extremely useful:

1. Contributors can use their changes as soon as PRs are merged.
2. It verifies that publishing is working. (I just caught a regression in sbt-typelevel v0.4.2 b/c your snapshot publish failed).
3. With all Typelevel projects publishing snapshots, we can one day do a distributed typelevel community build. See discussion in https://discord.com/channels/632277896739946517/920140670394589224/930847161330061323

Of course, it is finally up to the maintainers what makes sense for their project, but if there's anything I can do in sbt-typelevel to improve the experience I would like to know :)

[danicheg]

@armanbilge Mouse is a small library, and we release every new functionality, without waiting when they will accumulate to something 'weightful'. And it's just about preserving our previous behavior. Probably, it's not so bad to publish snapshots. But I haven't faced the problems with getting functionality by users after merging because of our as-soon-as-possible release policy.

[armanbilge]

Thanks for explaining, I wish that every project has that as-soon-as-possible release policy :)

If snapshots are not harmful to you, they are helpful to me to make sure sbt-typelevel is working (and they are helpful to you, so you don't burn a tag when a release unexpectedly fails). But I also understand the goal to preserve previous behavior as much as possible 👍

[armanbilge]

tl;dr: publishing snapshots is like running CI for publishing. It checks that nothing is broken and it lets us catch mistakes sooner.

[benhutchison]

FWIW I have no objection to reenable publishing snapshots if someone wants them, or for example, users complain that we arent releasing stuff quickly enough for them.