Can we create a Gitlab issue label like “release blocker” or something similar? Just wanted to poll the other maintainers and get some agreement/visibility on this before doing it. We have some issues like #1830 that I believe absolutely must be fixed before cutting a new release, and there’s no easy way of currently searching for them.
I’m in favor of creating such a tag. All other tags describe what kind of issue the issue is so I think the new tag should also by a kind. Perhaps “release blocker” fulfills that criteria. I cannot come up with something better ATM anyway.
That said, I’m wondering if #1830 is such an issue. In order to qualify for that, I think an issue must be a regression since the latest release and I’m not sure that that one is.
In my mind a release can always be made as long as it’s “better” that the previous one, even if it’s really, really bad.
I don’t completely follow what you mean by “kind.” Labels are not mutually exclusive, so to me it does not seem they all need to exist in the same taxonomy. E.g. we could have a “windows” tag for issues that are specific to Microsoft Windows or a “hard” tag for something we’ve investigated and has unknown cause or we know is hard to fix.
I see your point about blockers, and I agree that a release merely needs to be not a step backwards from the last one. I think we need to at least root cause #1830 and assess its impact before cutting a new release though. I’m concerned we shipped a broken or somehow incomplete tarball last time and never realized it.
1 Like
I think ‘releases should not regress’ is a good test to apply. Having a release-blocker tag SGTM.
Your right. I had some bad luck when thinking 
. I agree with you on both points. My hunch is that the problem is very old, though.