https://pantsbuild.org/ logo
#development
Title
# development
w

witty-crayon-22786

11/26/2018, 8:19 PM
https://internals.rust-lang.org/t/announcing-rust-2018-beta-release/8901 ... i might start this tonight, if @average-vr-56795 hasn't already?
a

average-vr-56795

11/26/2018, 9:36 PM
I last tried beta 2 IIRC
The big thing that didn't Just Work with autofixes was that extern crates got removed but the attributes attached to them didn't, which broke some things; I got rid of a lot of our crate attributes, and I assume they've fixed that bug
w

witty-crayon-22786

11/26/2018, 9:38 PM
would you be in favor of pinning to this release?
a

average-vr-56795

11/26/2018, 9:39 PM
Feel free to have a play :) I have a local branch somewhere but it was basically generated by running cargo fix a bunch of times
w

witty-crayon-22786

11/26/2018, 9:39 PM
...or, hm. is the idea that we would wait for
1.31
, and then do it?
oh. i think i get it. "help test the beta" means, upgrade to a 1.31 candidate, which represents the expected final first 2018
so maybe we don't upgrade quite yet, but we attempt the upgrade and post it somewhere to land post 1.31
a

average-vr-56795

11/26/2018, 9:42 PM
May be worth waiting until clippy and rustfmt come out of preview, but sure, works for me. IIRC it goes stable on Dec 6th, which I was intending to update on, but happy to beta for a bit
Aha your messages all came in after I sent mine! Silly undergroundness
I suspect as the upgrade is basically mechanical, and Borja and I will be making some non-trivial rust changes in the next couple of weeks, probably not worth keeping the branch around
We should definitely upgrade when it hits stable - can play around before or not. I will probably do a test update this week if you don't :)
w

witty-crayon-22786

11/26/2018, 9:45 PM
ok. i'll give it a go tonight if i have time
a

average-vr-56795

11/26/2018, 9:45 PM
Cool :)
w

witty-crayon-22786

11/26/2018, 9:45 PM
thanks
a

average-vr-56795

11/26/2018, 9:46 PM
I'm curious about both how many random scopes we can drop because of NLL, and how to identify them without just reading all our code
w

witty-crayon-22786

11/26/2018, 9:46 PM
mm, yea. good question.
there is always some other goalpost, but i'm realizing that borrowing across async boundaries is going to be the big big thing.
and that's still post async stabilization
a

average-vr-56795

11/26/2018, 9:48 PM
That's the big migration piece I have no clue about. But also, we don't need to drop the extra scopes, it would just be nice to...
w

witty-crayon-22786

11/26/2018, 9:48 PM
right
a

average-vr-56795

11/26/2018, 9:49 PM
Yeah, async borrows will be fancy. On the plus side, they're making good pace at developing that area :)
w

witty-crayon-22786

11/26/2018, 9:49 PM
yea, i don't know how many there will be. but they also shouldn't (?) affect performance
indeed! the stabilization rfcs have been lively, but seemingly converging
a

average-vr-56795

11/26/2018, 9:51 PM
Yeah, no performance, just a code simplicity thing. I believe NLL only applies to the borrow checker, and doesn't change any drop semantics (otherwise scary things would happen), just just a widening of what code is accepted
I ended up giving the migration a go; filed one bug with them (linked on the above thread)... Will push a branch for the sake of interest :)
Oh dear, I just hit an ICE :(
w

witty-crayon-22786

11/26/2018, 11:39 PM
Yay! It was all worthwhile =D