Stream: t-compiler/wg-prioritization

Topic: bikesheed libs-impl or T-libs-impl


Santiago Pastorino (May 21 2020 at 22:30, on Zulip):

kind of a bikesheed here, but I see that we have libs-impl and the rest of the "teams" are really called T-something, I wonder if we should rename this to T-libs-impl

Santiago Pastorino (May 21 2020 at 22:31, on Zulip):

I don't have a concrete need more than making more explicit that this is kind of a team which today happens to be under T-compiler control but in the way we are handling tags I see this like a team

Santiago Pastorino (May 21 2020 at 22:34, on Zulip):

at some point it could help in searches like T-* or stuff like that but what I also want to avoid is issues with T-compiler and libs-impl together as it happened today because doing that kind of breaks prioritization flow

Yuki Okushi (May 21 2020 at 23:58, on Zulip):

It's definitely useful to have T prefix for triaging, though it isn't _team_ directly. So, if we consider it as T-libs related item, then I'd +1 for this idea.

DPC (May 22 2020 at 00:03, on Zulip):

i agree T-libs-impl sounds better :slight_smile:

Santiago Pastorino (May 22 2020 at 13:05, on Zulip):

cc @pnkfelix

pnkfelix (May 27 2020 at 03:14, on Zulip):

I think T-libs-impl is confusing. We don’t have a libs-impl team.

pnkfelix (May 27 2020 at 03:14, on Zulip):

If anything, it should be, what, an “area”, like A-libs-impl?

pnkfelix (May 27 2020 at 03:15, on Zulip):

But that’s only if we think a prefix is warranted at all....

pnkfelix (May 27 2020 at 03:16, on Zulip):

I see that @Santiago Pastorino says we are treating libs-impl like a team, but I guess I would like more explanation of that?

Yuki Okushi (May 27 2020 at 10:57, on Zulip):

indeed A-libs-impl is more reasonable to me.
@pnkfelix so, my concern is that the issues labeled with libs-impl don't have any T-prefixed labels, I guess we could add T-compiler to the most issues now, right?

Santiago Pastorino (May 27 2020 at 12:47, on Zulip):

hmmm I guess it depends on how you look at this issue, we could say that we do have a libs-impl team which is actually composed by T-compiler members

Santiago Pastorino (May 27 2020 at 12:48, on Zulip):

I don't care that much about that to be honest but what @Yuki Okushi raised is the main discussion I think

Santiago Pastorino (May 27 2020 at 12:49, on Zulip):

how do we want exactly to label things and how wg-prioritization is going to adapt to the labelling system we decide to use

LeSeulArtichaut (May 27 2020 at 12:49, on Zulip):

I agree that we should add T-compiler so we can filter (for example I-nominated + T-compiler) when preparing meetings

Santiago Pastorino (May 27 2020 at 12:50, on Zulip):

well that or as we are already doing which is doing queries with I-nominated + T-compiler and I-nominated + libs-impl

LeSeulArtichaut (May 27 2020 at 12:50, on Zulip):

Right, but it complicates the queries

LeSeulArtichaut (May 27 2020 at 12:50, on Zulip):

It depends on how we want to organize the meeting I think

Santiago Pastorino (May 27 2020 at 12:51, on Zulip):

well yeah, not really complicated though :P, but one thing about splitting queries is that we can also split the sections of the agenda

LeSeulArtichaut (May 27 2020 at 12:51, on Zulip):

Whether we want to consider libs-impl separately or not

Santiago Pastorino (May 27 2020 at 12:51, on Zulip):

like let's now focus on the compiler nominations and then let's focus on the libs impl nominations

Santiago Pastorino (May 27 2020 at 12:51, on Zulip):

LeSeulArtichaut said:

Whether we want to consider libs-impl separately or not

exactly, otherwise everything is mixed up

Santiago Pastorino (May 27 2020 at 12:52, on Zulip):

we could also have both labels and split anyway, queries would be t-compiler + libs-impl and t-compiler - libs-impl

Santiago Pastorino (May 27 2020 at 12:52, on Zulip):

and we can also automatically tag with t-compiler things that are tagged with libs-impl

Santiago Pastorino (May 27 2020 at 12:52, on Zulip):

I guess we just want to define how do we want to do this

Santiago Pastorino (May 27 2020 at 12:53, on Zulip):

to be honest I'm neutral but we need to properly define this :)

Santiago Pastorino (May 27 2020 at 12:53, on Zulip):

cc @pnkfelix

pnkfelix (May 27 2020 at 13:16, on Zulip):

Unless we were to state that we have a T-libs-impl team, and set up infrastructure for it (e.g. a GitHub user group), I continue to have the same position that I posed up above. We can make the wg-prioritization queries smart enough to deal with whatever name we choose here.

pnkfelix (May 27 2020 at 13:16, on Zulip):

My concern is for the UX for people not on the wg-prioritization who are trying to understand the labels

pnkfelix (May 27 2020 at 13:16, on Zulip):

And add maintain them accordingly

Santiago Pastorino (May 27 2020 at 13:18, on Zulip):

@pnkfelix yeah sounds good!

Santiago Pastorino (May 27 2020 at 13:19, on Zulip):

let's stick with libs-impl or A-libs-impl, again, I'm neutral to that

Santiago Pastorino (May 27 2020 at 13:19, on Zulip):

now the question is ... do we want T-compiler + libs-impl or just libs-impl?

LeSeulArtichaut (May 27 2020 at 13:20, on Zulip):

I think overall T-compiler + libs-impl is more flexible

Santiago Pastorino (May 27 2020 at 13:21, on Zulip):

if @pnkfelix agrees on this I'd add to triagebot the ability to automatically add T-compiler labels to things that are libs-impl

pnkfelix (May 27 2020 at 13:23, on Zulip):

That sounds reasonable to me

pnkfelix (May 27 2020 at 13:23, on Zulip):

If I remove T-compiler, will it put it back?

LeSeulArtichaut (May 27 2020 at 13:24, on Zulip):

Seems feasible to me

pnkfelix (May 27 2020 at 13:24, on Zulip):

I’m namely thinking of the (rare) scenario where I dont want T-compiler on it. Eg if we want to nominate it for T-libs discussion

pnkfelix (May 27 2020 at 13:24, on Zulip):

I wasn’t clear: I don’t want the bot to put it back. :)

Santiago Pastorino (May 27 2020 at 13:24, on Zulip):

yeah it won't

LeSeulArtichaut (May 27 2020 at 13:25, on Zulip):

Well we can do both :D

Santiago Pastorino (May 27 2020 at 13:25, on Zulip):

what both means?

LeSeulArtichaut (May 27 2020 at 13:26, on Zulip):

I meant that we can decide how we want to implement it

LeSeulArtichaut (May 27 2020 at 13:26, on Zulip):

Sorry, that was unclear

Santiago Pastorino (May 27 2020 at 13:26, on Zulip):

I still don't get what you mean :)

Santiago Pastorino (May 27 2020 at 13:26, on Zulip):

you meant, if the bot keeps adding the label regardless Felix wants to remove it?

LeSeulArtichaut (May 27 2020 at 13:26, on Zulip):

:sweat_smile:

Santiago Pastorino (May 27 2020 at 13:27, on Zulip):

if it was april 1st we could totally do if pnkfelix then add_label_back :joy: :smiling_devil:

LeSeulArtichaut (May 27 2020 at 13:27, on Zulip):

I wanted to say that we could choose either to add it back or not

Santiago Pastorino (May 27 2020 at 13:28, on Zulip):

yeah what I was trying to say is that in the usual way we are implementing this, the label is not re-added

Santiago Pastorino (May 27 2020 at 13:28, on Zulip):

we have this for I-prioritize and regressions already

LeSeulArtichaut (May 27 2020 at 13:28, on Zulip):

Right, but I-prioritize is a bit special

Santiago Pastorino (May 27 2020 at 13:28, on Zulip):

right

LeSeulArtichaut (May 27 2020 at 13:29, on Zulip):

We might want a general functionality to add labels when certain labels are added

LeSeulArtichaut (May 27 2020 at 13:29, on Zulip):

To extract this functionality from the prioritize functionality

Santiago Pastorino (May 27 2020 at 13:29, on Zulip):

agreed

LeSeulArtichaut (May 27 2020 at 13:29, on Zulip):

cc @simulacrum

Santiago Pastorino (May 27 2020 at 13:30, on Zulip):

we have already a bunch and I also want to do I-unsound and not requires-nightly then add I-prioritize

Santiago Pastorino (May 27 2020 at 13:31, on Zulip):

I'd say we need a mapping from a list of patterns of labels to one or maybe a set of labels

LeSeulArtichaut (May 27 2020 at 13:31, on Zulip):

:+1:

Santiago Pastorino (May 27 2020 at 13:32, on Zulip):

@LeSeulArtichaut do you want to implement this or do I want to implement this?

Santiago Pastorino (May 27 2020 at 13:32, on Zulip):

:P

Santiago Pastorino (May 27 2020 at 13:41, on Zulip):

more ideas ... we need to ping cleanup whenever E-needs-bisection or E-needs-mcve are added

Santiago Pastorino (May 27 2020 at 13:41, on Zulip):

probably something similar exists for LLVM

Santiago Pastorino (May 27 2020 at 13:41, on Zulip):

basically I guess we could get rid of the command and just have a label doing the job

LeSeulArtichaut (May 27 2020 at 13:41, on Zulip):

We could use the Cleanup label then :shrug:

Santiago Pastorino (May 27 2020 at 13:42, on Zulip):

we could totally unify most of our commands under labels

Santiago Pastorino (May 27 2020 at 13:42, on Zulip):

LeSeulArtichaut said:

We could use the Cleanup label then :shrug:

right, but then we would want E-needs-bisection and E-needs-mcve to add Cleanup label too :P

Santiago Pastorino (May 27 2020 at 13:42, on Zulip):

those labels will add Cleanup and Cleanup addition will fire the command, hehe

LeSeulArtichaut (May 27 2020 at 13:42, on Zulip):

This might be the best solution

Santiago Pastorino (May 27 2020 at 13:43, on Zulip):

:+1:

LeSeulArtichaut (May 27 2020 at 13:44, on Zulip):

Else we might trigger the command twice due to adding E-needs-bisection then E-needs-mcve

LeSeulArtichaut (May 27 2020 at 13:44, on Zulip):

Anyway

LeSeulArtichaut (May 27 2020 at 13:44, on Zulip):

:bike:

Santiago Pastorino (May 27 2020 at 13:47, on Zulip):

we would need to check and fire once I guess :)

LeSeulArtichaut (May 27 2020 at 13:47, on Zulip):

If we fire when the label is added we won't add the label if it is already there, right?

LeSeulArtichaut (May 27 2020 at 13:48, on Zulip):

So GitHub's label system would prevent the mistake :slight_smile:

Santiago Pastorino (May 27 2020 at 14:02, on Zulip):

hmm I think it kind of re-adds it, not 100% sure though

Santiago Pastorino (May 27 2020 at 14:02, on Zulip):

I'd prevert this to happen from our code

LeSeulArtichaut (May 27 2020 at 14:02, on Zulip):

Of course :big_smile:

Last update: Jun 05 2020 at 23:15UTC