Stream: t-compiler/wg-prioritization/alerts

Topic: I-prioritize #74384 Compiler doesn't terminate with --releas


triagebot (Jul 16 2020 at 12:53, on Zulip):

@WG-prioritization/alerts issue #74384 has been requested for prioritization.

Procedure

Santiago Pastorino (Jul 16 2020 at 12:54, on Zulip):

lcnr said:

I guess P-high :thinking: not sure how easy this is to hit, so it might even be P-critical

:+1:

Santiago Pastorino (Jul 16 2020 at 12:54, on Zulip):

to be honest unsure between those

Santiago Pastorino (Jul 16 2020 at 12:54, on Zulip):

to me in general this kind of issues are like P-critical for sure

Santiago Pastorino (Jul 16 2020 at 12:55, on Zulip):

but I've seen some of those already where we don't typically tag like so

Santiago Pastorino (Jul 22 2020 at 15:34, on Zulip):

@WG-prioritization this one is the last unprioritized issue and an important one, given that could be P-high or P-critical

Santiago Pastorino (Jul 22 2020 at 15:34, on Zulip):

checking it again but please feel free to share thoughts here

Santiago Pastorino (Jul 22 2020 at 15:35, on Zulip):

I'd label as P-critical I think, at least for now

Santiago Pastorino (Jul 22 2020 at 15:36, on Zulip):

cc @lcnr given that you have commented about it too :)

DPC (Jul 22 2020 at 15:36, on Zulip):

p-critical seems fine to me

triagebot (Jul 22 2020 at 15:48, on Zulip):

Issue #74384's prioritization request has been removed.

simulacrum (Jul 22 2020 at 15:49, on Zulip):

I am confused by the critical label here, I would have put it no higher than P-medium, maybe even P-low.

simulacrum (Jul 22 2020 at 15:50, on Zulip):

It seems "ungreat" of course but if this isn't a regression or anything it doesn't seem that bad, certainly not worse than any of the other hanging issues we have.

simulacrum (Jul 22 2020 at 15:50, on Zulip):

(and fundamentally a case of "throwing too much code at LLVM, I guess)

Yuki Okushi (Jul 22 2020 at 15:52, on Zulip):

hm reasonable

Yuki Okushi (Jul 22 2020 at 15:55, on Zulip):

Santiago Pastorino said:

but I've seen some of those already where we don't typically tag like so

so, that point explains why we've been doing it so far

lcnr (Jul 22 2020 at 15:56, on Zulip):

I really dislike hangs, as esp in bigger projects it is often very hard to debug/analyse for users.

I also think this often is something worse than just "throwing too much code at LLVM".

Note: I underestimated how many I-hang we already have, so in this case P-high is probably more appropriate for issues like this

Santiago Pastorino (Jul 22 2020 at 16:09, on Zulip):

simulacrum said:

It seems "ungreat" of course but if this isn't a regression or anything it doesn't seem that bad, certainly not worse than any of the other hanging issues we have.

well seems reasonable, let's downgrade to P-medium then?

Santiago Pastorino (Jul 22 2020 at 16:35, on Zulip):

ok, gonna change to P-medium, if someone thinks it should be a different priority please speak up

lcnr (Jul 22 2020 at 16:38, on Zulip):

I think this is worse than other I-hang issues and would prefer P-high

lcnr (Jul 22 2020 at 16:42, on Zulip):

This is code which really should just compile but currently just hangs, using Box<[Foo; 125000]> finishes in less than a second

lcnr (Jul 22 2020 at 16:44, on Zulip):

ah well, we already assigned this P-medium :sweat_smile: so let's just keep it this way

simulacrum (Jul 22 2020 at 16:53, on Zulip):

The rarity of trying to create huge arrays on the stack is why my opinion is what it is, to be clear

simulacrum (Jul 22 2020 at 16:53, on Zulip):

It's probably a missed case somewhere

lcnr (Jul 22 2020 at 16:58, on Zulip):

The rarity of trying to create huge arrays on the stack is why my opinion is what it is, to be clear

The rarity of trying to create huge arrays on the stack is why I want P-high :laughing: I personally rank priority using unsound > hang > ICE without error > ICE after error.

lcnr (Jul 22 2020 at 16:58, on Zulip):

But yeah, I don't think P-medium is the end of the world here, so I am fine with how this is

Last update: Apr 11 2021 at 17:45UTC