Meta-issue: how to mark discussion status per issue about patma

[viridia]

I just added a 'rejected' label as a test.

[viridia]

(Sorry for the spam, I hadn't used GH labels before, was playing :)

[gvanrossum]

So should we use labels or something else? And what should the categories be? Be creative!

[viridia]

Of course we should use labels, all the cool kids are doing it.

The categories that you mentioned in the OP seem fine: accepted, rejected, open, postponed. Unless you want more exotic categories like 'idle speculation' and 'dilatory remarks'. :)

I'll make those labels now. The real question is, who decides which issue gets which label? Who will be the BDFtLotPEP (Benevolent Dictator for the life of this PEP)?

[viridia]

I went ahead and created labels for those 4 categories. And I attached labels to a few issues.

In some cases, the title of the issue is a question rather than a proposal ('how should we do X?') so the 'rejected' and 'accepted' labels don't really apply. We can, however, mark them as closed. If we think we need a 'resolved' label we can add that too.

One other question is whether you want to delete some of the default labels that Github creates, the list of labels is a bit long.

[viridia]

OK looking over the list of issues, I see that some of the discussions kind of die off without ever coming to a conclusion, although they do contains some interesting philosophical discussions. They aren't 'accepted' or 'rejected' because there's no concrete/actionable proposal to accept or reject. I'm not sure how to label these. Perhaps 'resolution needed'?

[gvanrossum]

"We're an autonomous collective." Anyone can add/remove labels. I don't expect any labeling wars.

I guess we need to think a bit more about the right categories.

It's also fine to edit the subjects of issues to be more consistent or appropriate.

I deleted all the default labels created by GitHub. (We can always recreate them later if we end up needing some of these.)

I wouldn't close any issues just yet -- maybe once the PEP is published in the peps repo we can close the issues we all agree on, if at least the PEP has a clear motivation based on the discussion in the PEP.

I like the idea of a 'resolution needed' label.

What to do with discussions that ended up accidentally happening in the wrong issue? (E.g. there was quite a bit of interesting stuff in the issue I created to link to Ivan's PEP.)

[gvanrossum]

One more thing. Sometimes there are things wrong with the PEP that aren't about deciding features but just clarifying what we decided in the issue. For those I propose that whoever finds these just file a PR. (For typos, spelling and simple grammar fixes you can also just commit and push to master.)

[viridia]

I am wondering if we need a new label. We currently have labels to indicate the status of group consensus - accepted, rejected, and such. However, the work does not stop there - we still need to update the PEP with the decision. How should we signal which issues are going to require PEP revisions and which are not?

A case in point, the recent decision to allow float constants as literal matches means that the PEP needs to be updated. It would be easy for a change like this to get lost in the shuffle - the only remedy is to scan the entire list of issues, read the final conclusion of each thread, and see if the PEP matches that conclusion.

So perhaps a label to indicate that this issue requires being documented in the PEP. Once the PEP is updated the label can be removed, unless there is a subsequent substantial change.

I had considered perhaps the github open/closed state - where 'closed' means 'no more work to be done' - but we've been avoiding 'closed' because these issues are part of the discussion audit trail and we don't want to hide the issues in the tracking list, unless we truly no longer wish to have the issue in question discussed.

A related problem: because this is a collaborative effort with no formal project management there is the potential for "diffusion of responsibility" syndrome - tasks languish because each person thinks that it is someone else's responsibility to work on them. However, this is not like an employment situation where you can simply tell someone to work on something, we're all volunteers who have other things to do with our time. (A good start would be to gauge everyone's level of time commitment and availability).

[gvanrossum]

I think adding a label for issues that need a PEP update is a good idea. And maybe another label to indicate "PEP update complete" (so issues without a label of this kind stick out as needing triage). I don't think there has ever been a PEP that needed explicit project management -- if you think a task is languishing, just give the tracker a shout-out!