Stream: t-compiler/wg-prioritization

Topic: What is the meaning of each priority level?


Santiago Pastorino (Apr 08 2020 at 21:49, on Zulip):

@o0Ignition0o - Jeremy Lempereur have shaped better the task you had https://hackmd.io/s9ZYktwLQteMsBBsPpKoHA?view#Define-and-document-What-is-the-meaning-of-each-priority-level

Santiago Pastorino (Apr 08 2020 at 21:50, on Zulip):

please feel free to ask questions about it

o0Ignition0o - Jeremy Lempereur (Apr 09 2020 at 08:04, on Zulip):

Super cool thanks a lot ! I'll work on it this weekend. Is there any place where I can see the "expected shape" of the final document ? maybe a former doc?

Santiago Pastorino (Apr 09 2020 at 13:11, on Zulip):

nothing in particular, just gather all the information we have during the meeting discussion, also the documentation already present in the hackmd and come up with a set of clear definitions on what's the meaning of each P level

Santiago Pastorino (Apr 09 2020 at 13:28, on Zulip):

@o0Ignition0o - Jeremy Lempereur also see #t-compiler/wg-prioritization > meaning of p-critical

o0Ignition0o - Jeremy Lempereur (Apr 09 2020 at 13:41, on Zulip):

ok will do, thanks ! :)

o0Ignition0o - Jeremy Lempereur (Apr 12 2020 at 15:50, on Zulip):

Started working on a draft. I've kept the copy pasted notes in the lower section https://hackmd.io/7NRRbq62TnaezW7-n15cDw Still quite a lot to do, Please let me know if the outline seems correct, or if I'm missing some aspects

Santiago Pastorino (Apr 13 2020 at 16:07, on Zulip):

@o0Ignition0o - Jeremy Lempereur will check it out later

o0Ignition0o - Jeremy Lempereur (Apr 13 2020 at 16:08, on Zulip):

sure, there's not too much to read _yet_, just checking in to make sure I'm in the right direction :)

Santiago Pastorino (Apr 13 2020 at 16:11, on Zulip):

so I'd say is the right direction yeah

Santiago Pastorino (Apr 13 2020 at 16:11, on Zulip):

we can start with something along those lines and then people can start making reviews and fixing things

o0Ignition0o - Jeremy Lempereur (Apr 13 2020 at 16:12, on Zulip):

Great! I'll try to fill the gaps :)

o0Ignition0o - Jeremy Lempereur (Apr 14 2020 at 16:09, on Zulip):

re https://rust-lang.zulipchat.com/#narrow/stream/227806-t-compiler.2Fwg-prioritization/topic/I-prioritize.20.2371078.20.60static.20FOO.3AFoo.3DFOO.3B.60.20doesn't.20cause.20cycl/near/193913905 should i edit the definition of A P-Critical task is blocker task that needs to be completed before any other task can be started. to incorporate the fact that there shouldn't be more than 5 P-Critical tasks at a given time ?

o0Ignition0o - Jeremy Lempereur (Apr 14 2020 at 16:10, on Zulip):

I have already mentioned that ```P-Critical tasks will be discussed at every weekly triage meeting until either:

The task is complete.
The task gets deprioritized.

``` but i might wanna add that P-Critical issues must be reviewed before any release as well ?

pnkfelix (Apr 14 2020 at 19:08, on Zulip):

This hackmd has started using the word "task" where I would have expected the word "issue"

pnkfelix (Apr 14 2020 at 19:09, on Zulip):

I don't have a strong preference for "issue", but I will admit that I worry that the wording may lead to some confusion. For example, the idea that "there shoudn't be more than 5 P-critical _ at one time"

pnkfelix (Apr 14 2020 at 19:09, on Zulip):

if you plug "tasks" in the above, it sounds reasonable (sure, we can just stop putting new tasks on our collective work queue)

pnkfelix (Apr 14 2020 at 19:09, on Zulip):

but if you plug "issues" in the above, it does not sound reasonable: We cannot stop people from discovering bugs

Santiago Pastorino (Apr 14 2020 at 19:13, on Zulip):

pnkfelix said:

This hackmd has started using the word "task" where I would have expected the word "issue"

yeah, @o0Ignition0o - Jeremy Lempereur was asking for some feedback, to be honest I just gave a glance to the structure but I'm pretty sure that @o0Ignition0o - Jeremy Lempereur is more than welcome to receive the kind of feedback you're giving :slight_smile:

Santiago Pastorino (Apr 14 2020 at 19:15, on Zulip):

I think I'd go with issue, but I'd need to read the whole document

Santiago Pastorino (Apr 14 2020 at 19:15, on Zulip):

maybe @o0Ignition0o - Jeremy Lempereur once you consider it ready for review please let us know

o0Ignition0o - Jeremy Lempereur (Apr 14 2020 at 19:17, on Zulip):

I'm indeed eager for feedbacks, thanks for taking the time to read the draft !

pnkfelix (Apr 14 2020 at 19:18, on Zulip):

no problem

pnkfelix (Apr 14 2020 at 19:19, on Zulip):

unfortunately hackmd's interface is a bit limited when it comes to writing feedback inline

pnkfelix (Apr 14 2020 at 19:19, on Zulip):

I also made some edits to try to fill in some gaps

o0Ignition0o - Jeremy Lempereur (Apr 14 2020 at 19:21, on Zulip):

thanks a lot, if there's any edit that would make sense but would take too much of your time, please let me know and I'll try to figure it out

pnkfelix (Apr 14 2020 at 21:00, on Zulip):

Section Check-in frequency, subsection P-High, is just not true: We do not have time to discuss every P-high bug at every weekly T-compiler triage meeting (aka the Thursday meeting). A more realistic goal would be to say that the someone (WG-prioritization?) should check-in on the progress of every P-high bug each week, and if some seem stalled or abandoned, then those should be nominated for discussion at the T-compiler meeting.

pnkfelix (Apr 14 2020 at 21:00, on Zulip):

This all ties into the fact that our P- categorization scheme has long needed refinement (and I'm hoping the addition of P-critical will help with this.)

pnkfelix (Apr 14 2020 at 21:03, on Zulip):

By the way, when I am reading "weekly triage meeting" in the text, I am assuming that you are talking about the T-compiler triage meeting that happens each thursday. But it might be good to make that explicit somewhere, since there are other interpretations (e.g. there is separate triage done by the release team, and there is the work of WG-prioritization itself, which could be viewed as triage)

o0Ignition0o - Jeremy Lempereur (Apr 15 2020 at 07:07, on Zulip):

We do not have time to discuss every P-high bug at every weekly T-compiler triage meeting (aka the Thursday meeting)

I meant P-Critical (removing it now)

o0Ignition0o - Jeremy Lempereur (Apr 18 2020 at 09:58, on Zulip):

@Santiago Pastorino @pnkfelix
The document is now ready for review.
Sections might be elaborated on, and I'll gladly do so if you have pointers or any feedback :)

Santiago Pastorino (Apr 19 2020 at 13:34, on Zulip):

@o0Ignition0o - Jeremy Lempereur will check it out tomorrow probably

pnkfelix (Apr 21 2020 at 17:34, on Zulip):

Thank you so much for writing that up @o0Ignition0o - Jeremy Lempereur ; its so useful to get an objective attempt to infer what the heck we are doing over here. :)

Santiago Pastorino (Apr 21 2020 at 18:59, on Zulip):

@pnkfelix I need to review this but ... looks good to you?

Santiago Pastorino (Apr 21 2020 at 19:00, on Zulip):

after we do a proper review I'd publish this in forge

pnkfelix (Apr 21 2020 at 19:20, on Zulip):

@Santiago Pastorino (i wrote comments on the rendered doc with feedback; its not ready for publication yet)

Santiago Pastorino (Apr 21 2020 at 19:20, on Zulip):

but looks in a good direction?

Santiago Pastorino (Apr 21 2020 at 19:20, on Zulip):

will review ASAP :)

Santiago Pastorino (Apr 21 2020 at 19:44, on Zulip):

@o0Ignition0o - Jeremy Lempereur just reviewed and it looks good, check out the comments from pnkfelix, and I've also made some edits, check them out. In particular, sorry for changing prioritisation to prioritization :), the thing is that the working group is named using a z instead of an s so to avoid confusion I've changed everything in the doc to reflect that.

o0Ignition0o - Jeremy Lempereur (Apr 21 2020 at 20:43, on Zulip):

Cool thanks a lot both of you :D I’ll have a read at the comments and try to address them tomorrow :)

Santiago Pastorino (Apr 30 2020 at 14:22, on Zulip):

@o0Ignition0o - Jeremy Lempereur any news on this?

o0Ignition0o - Jeremy Lempereur (Apr 30 2020 at 14:34, on Zulip):

Yes ! I have completed the missing sentence, and switched uses of "task" to "issue". I think we can iterate from here, maybe by expanding a bit on the I-Prioritize label. There was also a discussion about eventual N-<Priority> labels that would nominate an issue for prioritization and hinting about what priority seems relevant to the author ?

Santiago Pastorino (May 04 2020 at 20:18, on Zulip):

o0Ignition0o - Jeremy Lempereur said:

Yes ! I have completed the missing sentence, and switched uses of "task" to "issue". I think we can iterate from here, maybe by expanding a bit on the I-Prioritize label. There was also a discussion about eventual N-<Priority> labels that would nominate an issue for prioritization and hinting about what priority seems relevant to the author ?

hey, was going to read again the document

Santiago Pastorino (May 04 2020 at 20:19, on Zulip):

where that discussion about N-<Priority> happened?

Santiago Pastorino (May 04 2020 at 20:19, on Zulip):

I think it would be cool if more people suggest priorities

Santiago Pastorino (May 04 2020 at 20:20, on Zulip):

I think the outcome would be better if the author and the ones that tag with I-prioritize participate in this process

Santiago Pastorino (May 04 2020 at 20:20, on Zulip):

but on the other hand we can assume I-prioritize + P-medium means please wg prioritize this but I feel that this is P-medium

Santiago Pastorino (May 04 2020 at 20:21, on Zulip):

and people can always add a comment explaining the tags they are using

Santiago Pastorino (May 04 2020 at 20:21, on Zulip):

cc @pnkfelix

Santiago Pastorino (May 04 2020 at 20:22, on Zulip):

also a bit off topic of what we're talking about here, would love if people that tag with I-nominated explain what for and why are they nominating the issue, same with beta nominations

Santiago Pastorino (May 04 2020 at 21:13, on Zulip):

and also I'd just open a PR to include the information in the document as part of forge if @pnkfelix agrees on that

Santiago Pastorino (May 04 2020 at 21:13, on Zulip):

cc @o0Ignition0o - Jeremy Lempereur

o0Ignition0o - Jeremy Lempereur (May 05 2020 at 07:00, on Zulip):

I can't find mentions of eventual hints anymore. so I'll remove it for now. I really like the I-Prioritize + P-* idea which would serve the purpose perfectly. I wonder when we could discuss that so I can eventually add it to the docs. maybe tonight ?

Santiago Pastorino (May 05 2020 at 10:43, on Zulip):

if @pnkfelix agrees that something like that is a good idea I'd just add it

Santiago Pastorino (May 05 2020 at 10:43, on Zulip):

but yeah, people can always add comments on github saying, this seems P-high, P-medium or whatever

Santiago Pastorino (May 05 2020 at 10:44, on Zulip):

I guess the discussion is if we want to encourage that and in particular if we want to use a system like I-prioritize + P-something, I'm in favor of whatever along those lines

pnkfelix (May 05 2020 at 13:51, on Zulip):

I ... am conflicted

pnkfelix (May 05 2020 at 13:52, on Zulip):

I suppose my main issue is that labels alone are a very limited channel for communicating intent

pnkfelix (May 05 2020 at 13:53, on Zulip):

sorry, I'm still working out my thoughts as I write

pnkfelix (May 05 2020 at 13:53, on Zulip):

so lets look at the hypothetical scenario: Someone sees an issue, wants it prioritized, but has a suggestion for what the priority should be

pnkfelix (May 05 2020 at 13:54, on Zulip):

the options that I see there are either:

pnkfelix (May 05 2020 at 13:54, on Zulip):

tag with I-prioritize and also a P-label

pnkfelix (May 05 2020 at 13:54, on Zulip):

or tag with I-prioritize and a comment with the suggested priority

pnkfelix (May 05 2020 at 13:55, on Zulip):

the comment form has one main advantage that I can think of: It provides an immediate place for someone to write why they think the issue has that priority

pnkfelix (May 05 2020 at 13:55, on Zulip):

(whether or not they chose to do so is up to the person doing the tagging, of course. But at least the option is immediately there, with no friction.)

pnkfelix (May 05 2020 at 13:56, on Zulip):

the main downside to the comment form is that there is a friction, up front, imposed on everyone; namely, that they need to add the label and the comment

pnkfelix (May 05 2020 at 13:57, on Zulip):

but we already have a bot that can add the label, right? as in, the I-prioritize and the note can all be bundled into a single comment form, if a contributor wants to do that?

pnkfelix (May 05 2020 at 13:57, on Zulip):

so I guess I'm trying to understand what the main benefit of having the suggested priority be encoded as a P-label

pnkfelix (May 05 2020 at 13:58, on Zulip):

the main benefit I can see there, is that since prioritization is a concurrent process, it will enable people doing issue review to see immediate results on what the set of P-priority issues are (and it will just potentially include issues with suggested priorities)

pnkfelix (May 05 2020 at 13:58, on Zulip):

without having to wait for the wg-prioritization to get around to it

pnkfelix (May 05 2020 at 13:59, on Zulip):

and that does seem like a potentially big win, depending on how many people are actually using the P-priority labels in the first place in their attempts to traverse issue database when looking for things to fix

pnkfelix (May 05 2020 at 14:00, on Zulip):

so yeah, I guess this is not obvious to me which protocol we want to officially recommend

pnkfelix (May 05 2020 at 14:00, on Zulip):

but my mind keeps coming back to the question of how limited labels alone are as a medium for communication

pnkfelix (May 05 2020 at 14:01, on Zulip):

will wg-prioritization end up in a situation where its doing work trying to determine why a priority was suggested?

pnkfelix (May 05 2020 at 14:01, on Zulip):

(and in those cases, would it have helped anything to encourage the comment form instead.... because people could easily just write a comment saying nothing but "I-prioritize P-medium " and we'd be in the same boat)

Santiago Pastorino (May 05 2020 at 17:59, on Zulip):

yeah, I agree with what you've said, maybe we should just recommend people to add comments explaining things

Santiago Pastorino (May 05 2020 at 17:59, on Zulip):

and in general is a thing I'd try to encourage more

Santiago Pastorino (May 05 2020 at 18:00, on Zulip):

as I've said, I've seen things I-nominated and end wondering why is this nominated?

Santiago Pastorino (May 05 2020 at 18:00, on Zulip):

or also, same has happened during thursday's triage meetings with beta nominations, that are not clear why they are nominated

Santiago Pastorino (May 05 2020 at 18:01, on Zulip):

so I guess it would be nice if as a general procedure we encourage to explain why people add those tags

Santiago Pastorino (May 05 2020 at 18:02, on Zulip):

maybe the same applies here, you go and tag with I-prioritize and then comment something like ... hey I'm requesting prioritization on this issue because X and Y but in my opinion this is P-something because Z

LeSeulArtichaut (May 11 2020 at 16:55, on Zulip):

@lcnr Have you read this document by o0Ignition0o - Jeremy Lempereur ? You probably have, but I thought you may have missed it :slight_smile:

lcnr (May 11 2020 at 16:56, on Zulip):

I did not. Thanks :heart:

LeSeulArtichaut (May 11 2020 at 16:56, on Zulip):

You're welcome ;)

LeSeulArtichaut (May 11 2020 at 16:56, on Zulip):

I don't know if it's listed somewhere (e.g. in the master document)? If not we should add it

LeSeulArtichaut (May 11 2020 at 16:57, on Zulip):

Oh it is

lcnr (May 11 2020 at 16:57, on Zulip):

what is the master document?

LeSeulArtichaut (May 11 2020 at 16:58, on Zulip):

:D

LeSeulArtichaut (May 11 2020 at 16:58, on Zulip):

https://hackmd.io/P3yvagSSS7yAI2QxhLfPtQ?view

LeSeulArtichaut (May 11 2020 at 16:59, on Zulip):

People hid so many things to you :eyes: :laughing:

lcnr (May 11 2020 at 17:00, on Zulip):

it's a terrible day for rain

Santiago Pastorino (May 11 2020 at 20:17, on Zulip):

btw, I guess we should open a PR with this and continue discussions on Github cc @o0Ignition0o - Jeremy Lempereur, I'd say open a PR with the document to the forge repo and cc me and pnkfelix

o0Ignition0o - Jeremy Lempereur (May 12 2020 at 06:52, on Zulip):

Oh yeah sure it's a good idea !

Santiago Pastorino (May 12 2020 at 12:38, on Zulip):

:+1:

o0Ignition0o - Jeremy Lempereur (May 13 2020 at 06:52, on Zulip):

Ok I'm going to create a PR and mention the discussion we're having here. I just need to figure out where to create it and what "the forge" is (sounds like some star wars / revan space weapon related stuff, i like it!)

o0Ignition0o - Jeremy Lempereur (May 13 2020 at 06:52, on Zulip):

(sorry for the noob questions)

o0Ignition0o - Jeremy Lempereur (May 13 2020 at 11:04, on Zulip):

Ok dpc sent me a link to the forge, i'll create a PR today :)

Last update: Jun 05 2020 at 23:20UTC