Stream: t-compiler/wg-prioritization

Topic: I-prioritize #68010 Large ZST arrays take a looong time to c


triagebot (May 21 2020 at 03:14, on Zulip):

@WG-prioritization issue #68010 has been requested for prioritization.

lcnr (May 21 2020 at 05:51, on Zulip):

Afaict there isn't really a reason to use this and it was undetected for 10 version, P-low?

LeSeulArtichaut (May 21 2020 at 13:10, on Zulip):

Mhh, I think it's worth bisecting a bit

LeSeulArtichaut (May 21 2020 at 13:10, on Zulip):

Can we profile the compiler to see where that additional comilation time comes from?

LeSeulArtichaut (May 21 2020 at 17:53, on Zulip):

Did some bisection, this seems like a const-eval bug. I pinged RalfJung on the issue

LeSeulArtichaut (May 21 2020 at 17:54, on Zulip):

Depending on his answer I think we can assign P-medium or P-low

LeSeulArtichaut (May 21 2020 at 21:43, on Zulip):

RalfJ said:

Okay so this is interning walking every single field of that array... and taking forever to do that, obviously.

So the problem comes from the big size of the array. I think this means we can tag P-low?

LeSeulArtichaut (May 21 2020 at 21:47, on Zulip):

1.31.0 is the introduction of const evaluation, right? So this isn't much of a regression, but rather a cost induced by const evaluation.

Santiago Pastorino (May 21 2020 at 22:20, on Zulip):

@LeSeulArtichaut do you want to go ahead and tag P-low and remove I-prioritize?

LeSeulArtichaut (May 21 2020 at 22:21, on Zulip):

Ok

triagebot (May 21 2020 at 22:22, on Zulip):

Issue #68010's prioritization request has been removed.

Last update: Jun 05 2020 at 23:15UTC