Stream: t-compiler/wg-nll

Topic: regressions in #66517


nikomatsakis (Nov 21 2019 at 15:45, on Zulip):

Hey all

nikomatsakis (Nov 21 2019 at 15:46, on Zulip):

So @lqd is working on improving the list of what's broken --

pnkfelix (Nov 21 2019 at 15:46, on Zulip):

darn it i was spending too much time trying to think of some play on "abandon hope all ye who ..."

nikomatsakis (Nov 21 2019 at 15:46, on Zulip):

lol I made no effort at humor

nikomatsakis (Nov 21 2019 at 15:46, on Zulip):

but now I wish I had!!

lqd (Nov 21 2019 at 15:48, on Zulip):

(I'm a touch busy at work rn but I'm working on this in the bg)

nikomatsakis (Nov 21 2019 at 15:49, on Zulip):

ok, my one question would be if there is some way to make a centralized "to do" list?

simulacrum (Nov 21 2019 at 15:49, on Zulip):

@lqd if you can get through ~40 or so (or w/e, doesn't really matter), I'll try to finish up this weekend

nikomatsakis (Nov 21 2019 at 15:49, on Zulip):

i.e., maybe a master hackmd that multiple people can work through?

simulacrum (Nov 21 2019 at 15:49, on Zulip):

I think we should copy my comment into a hackmd

nikomatsakis (Nov 21 2019 at 15:49, on Zulip):

(I won't have any time today myself, though)

simulacrum (Nov 21 2019 at 15:50, on Zulip):

and then start crossing out or checking ticky boxes or something

nikomatsakis (Nov 21 2019 at 15:50, on Zulip):

I think we should copy my comment into a hackmd

this is what I had in mind, :+1:

lqd (Nov 21 2019 at 16:09, on Zulip):

ok so, I'll 1) turn the comment into a todo list hackmd, 2) cross out the ones I've done, 3) keep checking out some of the others so that we at least have an idea

lqd (Nov 21 2019 at 16:10, on Zulip):

and we can then fill it little by little

lqd (Nov 21 2019 at 16:26, on Zulip):

I'm bit confused by the list btw, it seems already partially filtered ? (some roots mention X regressions but have no <details>, or less than X items in there)

lqd (Nov 21 2019 at 16:52, on Zulip):

something like this ? https://hackmd.io/lKRRzo1sSpq82ZwCHQtfVg

simulacrum (Nov 21 2019 at 17:51, on Zulip):

@lqd this looks great!

simulacrum (Nov 21 2019 at 17:52, on Zulip):

Yes, the list deletes all the non-roots (details contains the list of regressions caused by that root, we don't care about those)

lqd (Nov 21 2019 at 17:53, on Zulip):

hum some of the details exist, maybe I shouldn't have checked those ?

simulacrum (Nov 21 2019 at 17:53, on Zulip):

I think I included a few, but shouldn't have

lqd (Nov 21 2019 at 17:53, on Zulip):

in any case I added the existing ones as sublists

simulacrum (Nov 21 2019 at 17:54, on Zulip):

oh -- I think I remember -- the sublists are presumably all of only one root; we need to fix the root, but not the sublist crates

simulacrum (Nov 21 2019 at 17:54, on Zulip):

there's just not a handy link to the root itself sometimes

lqd (Nov 21 2019 at 17:55, on Zulip):

yeah that's also why I usually go with versions (I've answered to your forge PR mention, describing a bit what I meant about my own process) :)

simulacrum (Nov 21 2019 at 17:57, on Zulip):

yeah I have that in my queue :)

lqd (Nov 21 2019 at 17:57, on Zulip):

queue of things to read or queue of things to do (wrt versions) ?

lqd (Nov 21 2019 at 17:59, on Zulip):

(if it's the second, I can try to release said-tool, which is actually based on your old crater-report-merge, and is very similar to crater-generate-report)

simulacrum (Nov 21 2019 at 18:03, on Zulip):

uh

simulacrum (Nov 21 2019 at 18:03, on Zulip):

I have not yet read your comment

simulacrum (Nov 21 2019 at 18:03, on Zulip):

not sure as to the later :)

lqd (Nov 21 2019 at 19:45, on Zulip):

@simulacrum ok I've triaged and tested dozens (most reg ones, barely no gh) let me know what you think (but as we expected the vast majority of active ones I've seen have since released versions which build)

simulacrum (Nov 21 2019 at 19:54, on Zulip):

Are those versions semver compatible? i.e. a cargo update away?

lqd (Nov 21 2019 at 19:57, on Zulip):

unfortunately, no

lqd (Nov 21 2019 at 19:57, on Zulip):

I've seen a couple which were published as incompatible but seemed actually compatible, but that wasn't the norm

lqd (Nov 21 2019 at 19:58, on Zulip):

because these failing versions are generally pretty old

simulacrum (Nov 21 2019 at 21:35, on Zulip):

okay we'll want to try and nag folks to publish a semver-compat version I think

lqd (Nov 28 2019 at 15:38, on Zulip):

@simulacrum I hadn't seen you had completed the todo list, awesome, thank you

simulacrum (Nov 28 2019 at 15:38, on Zulip):

ah there's a few left I think

simulacrum (Nov 28 2019 at 15:38, on Zulip):

thrussh and one or two others

lqd (Nov 28 2019 at 15:39, on Zulip):

djc has done a PR, I've done one as well (which was funny)

lqd (Nov 28 2019 at 15:39, on Zulip):

I think one'd need to install pijul for thrussh

simulacrum (Nov 28 2019 at 15:40, on Zulip):

it was unclear to me about that

simulacrum (Nov 28 2019 at 15:40, on Zulip):

maybe it's possible to at least file an issue?

lqd (Nov 28 2019 at 15:41, on Zulip):

sure I'll do so and ping pmeunier on twitter

simulacrum (Nov 28 2019 at 15:41, on Zulip):

:thumbs_up:

lqd (Nov 28 2019 at 15:42, on Zulip):

are you more satisfied with the state of things after going through the list than earlier ?

lqd (Nov 29 2019 at 13:23, on Zulip):

@simulacrum so I have a fix for thrussh 0.19.x and 0.20.x, but there's not really an issue tracker on the pijul nest. There's a kind of discussion forum. I've opened a thread there as one would a github issue, but just looking at the recent ones, this forum seems to have very low activity, and I think little to none of the recent activity did involve the crate authors. So we'll see but here it is https://nest.pijul.com/pijul_org/thrussh/discussions/42

simulacrum (Nov 29 2019 at 13:23, on Zulip):

good enough, I think

simulacrum (Nov 29 2019 at 13:24, on Zulip):

yeah, I'm definitely more satisfied -- I think we're in a pretty good place, overall

lqd (Nov 29 2019 at 13:26, on Zulip):

cool

lqd (Nov 29 2019 at 13:26, on Zulip):

I wonder if no crates depend on 0.20 since it wasn't in the doc

lqd (Nov 29 2019 at 13:27, on Zulip):

but they'd be all equally affected on 1.40 (it's the same error)

lqd (Nov 29 2019 at 13:30, on Zulip):

so I'll add this info to the hackmd and tick the box

simulacrum (Nov 29 2019 at 13:33, on Zulip):

sounds good, yeah, seems plausible

simulacrum (Nov 29 2019 at 13:33, on Zulip):

or it error'd out

lqd (Nov 29 2019 at 13:40, on Zulip):

I've ticked the last remaining couple boxes

lqd (Nov 29 2019 at 13:41, on Zulip):

we could do something with the older wgpu releases depending on the broken gfx-hal version, but I'm not sure it's worth it

lqd (Nov 29 2019 at 13:43, on Zulip):

but otherwise I think we're good

lqd (Nov 29 2019 at 13:43, on Zulip):

@simulacrum in any case let me know if you see more work to be done here

simulacrum (Nov 29 2019 at 13:44, on Zulip):

yeah, I think it's fine to leave it here for now, will let you know if we find anything else

lqd (Nov 29 2019 at 20:33, on Zulip):

@simulacrum alright we have semver-compatible updates to thrussh :)

simulacrum (Nov 29 2019 at 20:33, on Zulip):

amazing!

Last update: Dec 12 2019 at 00:50UTC