
Great, it looks awesome. Issue template is great idea as well. (for ones who aren’t familiar: https://github.com/blog/2111-issue-and-pull-request-templates https://github.com/blog/2111-issue-and-pull-request-templates) (I’m +1 for having GitHub stuff under .github/ directory) The issue template could ask for - Cabal/cabal-install/ghc versions - ask to run cabal-install with -v2 flag and add that to the issue? with quick glance it doesn’t apply to many issues, but when it does, it would be helpful. - Oleg
On 12 Jul 2016, at 23:48, Edward Z. Yang
wrote: Excerpts from Oleg Grenrus's message of 2016-07-12 12:45:05 -0700:
On 12 Jul 2016, at 18:42, Edward Z. Yang
wrote: Excerpts from Oleg Grenrus's message of 2016-07-12 04:03:43 -0700:
Looks good indeed!
I have few questions: - what is purpose of `paging:*` labels, to help people see issues they are interested in? How it’s different from assignees (which can be multiple)?
Beyond what Mikhail stated:
- Multiple people can be paged, only one assigned Yes you can. https://github.com/blog/2178-multiple-assignees-on-issues-and-pull-requests https://github.com/blog/2178-multiple-assignees-on-issues-and-pull-requests
Haha! Learned something today.
- I can put more metadata in the tag name than assignable (to help people decide who to page) Ah, page as in ping. That meaning I always find weird (and don’t remember).
summon (someone) over a public address system, so as to pass on a message
IMHO multiple assignment would work better as it actually sends notification (if one choose to receive such).
Yet, you’re right, deciding whom to page/assign is often non-obvious. Not sure if few-word description if any helpful. (e.g. whom to contact on hackage-security problems, I’m actually unsure whether it’s edsko or dcoutts, or on something else...).
OK, I am convinced we should drop it.
Let's do this:
- To page someone, just write CC @blah in the message - We should add an issue template that requests you CC someone and explains who you might want to CC
- why “bug” has “ezyang planning to delete this tag”. I’d prefer to have “type: bug” and other “type:*” labels as: “type: discuss”, “type: enhancement”, “type: question”, and maybe more as e.g. stack has https://github.com/commercialhaskell/stack/labels https://github.com/commercialhaskell/stack/labels
OK, the tag served it's purpose! I planned to delete it because there are lots of bugs in the issues tracker and no one has been methodically tagging them bug or not, so it seemed that the tag was just not that helpful. Just look at Stack's issues list: https://github.com/commercialhaskell/stack/issues who is tagging things as bugs!
If we go for type:* tags, I’ll help with triaging all issues with type:* tag.
OK, fine. I've reorged accordingly.
discuss/enhancement/question are useful and I support tags for them. Presently we have "priority: enhancement" but we can rename that as needed.
I’d like priorities to have a total-order. high > low is obvious, but what about enhancement and user-question? I’d change latter two into type:*, and maybe later introduce third priority level if we feel we need one.
Added.
- how priority labels interact with milestones?
Agreed with Mikhail; priority within milestone.
- Should we add “pr welcome” or “awaiting pr” for issues which are discussed, but nobody have interest or time implementing right away (will help new contributors especially when combined with `meta: easy`)
Sure! I wonder, however; for tickets that are tagged this way, I feel the onus is on us to write a clear spec at the top of the bug for what is desired (even better: link to a commit with a test!) Will help contributors a lot.
Yes, we should help as much as possible. I’d tag only “clear” issues, and add a comment that I can help with it, if there are some questions (or/and assign it to myself).
Also I forgot to ask about "attention: please-merge”. What’s it purpose, to tag PRs that author thinks are amerceable? IMHO the comment is enough, and also would work for external-contributors, who **cannot** tag issues/prs. (This is the reason why I got push-rights in the first place, I’m still quite uncomfortable pushing changes myself).
Dropped it.
And what’s the idea behind “attention: regression”? How it’s different from a `type: bug` (its special case of a bug, but does it matter that much. Regressions could be critical or not, so priority tag, with type:bug would be enough?)
Regression is in here because we used to have a regression tag. I'll reclassify them.
E.g. is:open label:"priority: high" label:"bug (ezyang is planning to delete this tag)" milestone:"Cabal 1.26" filter https://github.com/haskell/cabal/issues?utf8=%E2%9C%93&q=is%3Aopen%20label%3A%22priority%3A%20high%22%20label%3A%22bug%20(ezyang%20is%20planning%20to%20delete%20this%20tag)%22%20milestone%3A%22Cabal%201.26%22%20 https://github.com/haskell/cabal/issues?utf8=%E2%9C%93&q=is:open%20label:%22priority:%20high%22%20label:%22bug%20(ezyang%20is%20planning%20to%20delete%20this%20tag)%22%20milestone:%22Cabal%201.26%22%20
shows not that many.
OTOH there are 210 open issues (is:open is:issue no:milestone) without any milestone. Should we put them all into _|_ - milestone, and then promote to version milestones, when the discussion advanced enough we know when we want to release them (the soonest, or the latest?). As Cabal-1.26 165 open issues indicates, it’s more like “the soonest”, at least at this point.
cabal-install-1.24.0.1 has 12 open issues, should we create cabal-install-1.24.1 -milestone and move them there?
I think we should try to arrange a phone call behind all the Cabal stakeholders and have a triage session to remilestone these bugs.
Edward