abort builds with [skip ci]/etc

[djsutherland]

Even though CircleCI supposedly supports [skip ci], I still usually see builds go through when I add that; not sure what I'm doing wrong, if they dropped it, or if something is misconfigured on our end. But we could work around that by changing the build scripts, or maybe ff_ci_pr_build.py, to check the commit message and abort if it sees [skip ci].

Also, since CircleCI doesn't support [skip circle] and don't seem to want to (https://github.com/conda-forge/conda-forge.github.io/issues/45#issuecomment-198060065), we could more-or-less implement that here. Same for [skip travis] (https://github.com/travis-ci/travis-ci/issues/5032). Or we could implement [skip mac] or something to still work when we use CircleCI for Mac builds also, or similar.

Drawback: it would prevent you from manually running builds of commits that shouldn't actually have been skipped, unless we add something to do that.

[ocefpaf]

Drawback: it would prevent you from manually running builds of commits that shouldn't actually have been skipped, unless we add something to do that.

Not a big deal IMO, the advantages do outweigh this disadvantage.

I'm not sure how hard would be to implement that but I like the idea of a [skip mac] and [skip linux].

[djsutherland]

I think all we need to do is have the build script run a git command to check the current commit message. Shouldn't be too difficult; I might give it a shot this weekend.

[croth1]

This will become even more interesting once we switch to azure as the provider for many platforms.

[kynan]

Any update on this? Skipping CI for Azure does not seem to work.

[jaimergp]

I added:

• https://github.com/conda-forge/staged-recipes/pull/27765
• https://github.com/conda-forge/conda-smithy/pull/2077

With those merged and released we can close this.