Stream: t-compiler/wg-polonius

Topic: State of -Zpolonius


Jake Goulding (May 19 2020 at 19:07, on Zulip):

At one point, I was told that -Zpolonius was effectively broken and allowed all sorts of bogus code. Is that still the case?

Jake Goulding (May 19 2020 at 19:07, on Zulip):

It's a useful tool to be able to say things like "this code will work in the future, b/c I tested it with -Zpolonius", but not if it doesn't work-work

Matthew Jasper (May 19 2020 at 19:09, on Zulip):

I think it hasn't been significantly broken for a long time now, although there's no CI for it.

Jake Goulding (May 19 2020 at 19:23, on Zulip):

Now, just gotta turn the polonius switch then, right? :cry:

lqd (May 19 2020 at 21:37, on Zulip):

while I wouldn't say it's broken per se, it can still be slow (esp w/ the move errors computation), OOM, contain bugs (especially dealing with higher-ranked things as those will require chalk)

Jake Goulding (May 19 2020 at 23:44, on Zulip):

@lqd it may be hard to say, but which style of bugs? Allowing code that it shouldn't or disallowing code it should allow?

lqd (May 20 2020 at 08:06, on Zulip):

unfortunately for now I'd say the former, to handle the problems described in https://smallcultfollowing.com/babysteps/blog/2019/01/21/hereditary-harrop-region-constraints/ there's still some necessary chalk-polonius interaction that doesn't exist yet. That being said, chalk has been making a lot of progress recently and the wg has recently started working on a lot of those building blocks, so fixing this might not be so far off

Last update: Jun 20 2021 at 01:15UTC