mozilla :: #rust-infra

17 May 2017
00:41jonhoosimulacrum acrichto: I'm playing around with my Rust Nightly twitter bot and integration with Any chance we could land Without that the API is pretty useless from Rust :/
00:50jonhooI put a private struct in my code and use mem::transmute for now, so no rush, but just figured I'd ping you anyway
00:50jonhoocome to think of it, that was probably unnecessary :p sorry!
00:51aturonjonhoo: btw, if you're interested in hacking on rust infrastructure in general, this is a great place to hang out -- and we have IRC meetings on fridays at 2:30 pacific
00:51aturoni mention this because we've long wanted a twitter bot for monitoring and other infra notifications
00:51aturonlots of great potential problems to hack on!
00:52* misdreavus perks up at the mention of twitter bots
00:52jonhooaturon: tempting! I'll see if I can make this Friday :)
00:52jonhoomisdreavus: mine is at, though I know there are two others at least
00:52jonhoofollowing some prodding from nrc
00:54misdreavusyeah, i know two others, one got rt'd by @rustlang, and the other name-dropped me for the twitter library
00:55jonhoomisdreavus: mine will soon report nightly performance changes from perf.rlo too ;)
00:55misdreavusthat sounds awesome!
01:16jonhoowe'll see how well it works when the next nightly is out
01:16jonhooI've pushed and deployed the changes -- had to compress the text a fair amount to make it fit in 140 chars :(
01:16jonhoobut in theory it'll print something like:
01:16jonhoo2017-05-15 nightly released.
01:16jonhoorust diff:
01:16jonhooperf down 0.7%:
01:17jonhooif cargo has also changed, it'll include a "cargo diff" line with a GitHub compare link for cargo
01:17jonhooa bit sad that the bot now itself requires nightly due to its dependency on rustc-perf
01:18jonhooi also worry that perf.rlo may not have any relevant benchmark results by the time the bot runs
01:30simulacrumjonhoo: Sorry I missed you, but yeah, we can land it :)
01:30simulacrumI left a comment though
01:55jonhoosimulacrum: no worries -- I kept an eye on
02:01simulacrumjonhoo: Merged!
14:17frewsxcvbrson_: when people ask you to do crater runs to find regressions, which cargobomb "list" do you run it on? e.g. top-100
14:24carols10centsfrewsxcv: i think brson is out this week?
14:32frewsxcvwell, if anyone knows the answer to that question, i'd be happy to fill in for him
14:32frewsxcvi figure it's not being run on all crates, otherwise that'd take like a week
14:34steveklabniki thought it was
14:34steveklabnikhe said it takes ~72 hours to do the run
14:34frewsxcvmaybe my machine is just very slow
14:34frewsxcvactually, i'm already at 'm', and it's only bene running for a couple days, so it's probably not as slow as i thought
14:40carols10centssounds like it would be awesome if it ran in popularity order, so you could see results for more commonly used/substantial crates first and possibly stop the run if you get a bunch of regressions early
14:41carols10centsunless the point is to see what percentage of the whole ecosystem fails
14:45frewsxcvlooks like there's a "hot" list which is populated by the crates with the most reverse dependencies
14:45carols10centsfrewsxcv: so does it run "hot" first? or "top" first? or?
14:46frewsxcvthere's the code for running on all the crates, which is what i'm doing
14:47frewsxcvit's just combining all of the lists in a set, then sorting, presumably alphabetically
14:48frewsxcvthese are the four different modes cargobomb can be run in
14:48frewsxcvas far as which set of crates to test on
16:59onurHi guys. Looks like LeaseWeb (hosting provider of will stop providing DNS service for third party domains at the end of this month. It would be great if you could add to EasyDNS account that is used for all rust related domains.
17:00carols10centsacrichto ^^ edunham ^^
17:00* steveklabnik waves at onur
17:01acrichtoonur: sure yeah I can look into that
17:01acrichtoonur: what's the timeline here?
17:01acrichtooh 2 weks
17:01acrichtoI'l look at it right now
17:01onuryes acrichto
17:02onurwe are not in rush :-) but thanks
17:05acrichtoonur: ok I'll give you an update by the end of the day
17:06onurthanks acrichto
19:01tedacrichto: BTW, I love the SHA-512 switch with ring, debug sccache builds are actually usable now!
19:01acrichtoted: :D
19:01acrichtonot exactly an intended side effect
19:01acrichtobut hey if it works it works
19:01tedi tried building firefox with a debug sccache a while ago and i gave up, it was awful
19:02acrichtoheh I could believe it
19:02tedas you saw, hash generation was the hot path, and using a non-optimized sha-1 was just unusable
19:03acrichtoted: btw in true "alex uses sccache as a testing ground for everything" fashion, I've been playing around with async/await in sccache --
19:03acrichtoit's working out great so far
19:03acrichtoonce you get around all the ICEs
19:03acrichtothis is like super far future
19:03acrichtosuper extra nightly
19:04tedare there docs or a writeup on it anywhere?
19:04acrichtolol not yet
19:04acrichtoit's not even in rustc itself
19:04tedit's not coroutines, is it?
19:04acrichtoI'm toying around here --
19:04acrichtonah yeah it's all built on coroutines
19:04acrichtothis commit --
19:04tedi really miss having generators in rust, having written a lot of Python over the years
19:05acrichtoI&#39;ve been at this for like < 5 hours
19:05acrichtobut damn this is a big payoff
19:05tedthat diff does look nice!
19:06tedsyntactic sugar makes so much difference for async programming
19:06acrichtoyeah you&#39;re not kidding
19:10teddoes that generator stuff have an RFC attached?
20:13frewsxcvshould we mark all rustdoc issues from now on as T-dev-tools?
20:52tededdyb: thanks for your help, I think I fixed that sccache-induced rustc panic
20:53eddybwoo o/
21:03steveklabnikfrewsxcv: nrc acrichto aturon ^
21:06nrcfrewsxcv: yeah, go for it, thanks!
22:56simulacrumaturon: Sorry I haven&#39;t done anything with the irlo post, been busy
22:56simulacrumCould you give me a quick summary of what we want in it?
23:00aturonsimulacrum: oh no worries! actually, as it happens, we&#39;re probably going to announce a couple of other new subteams at the same time, so we can just roll it up
23:01simulacrumWell, I can make part of it at least if you&#39;d like; just let me know what the broad points should be
23:06aturonsimulacrum: sure! gimme a sec
23:37aturonsimulacrum: argh got distracted, will follow up later
23:38simulacrumno problem
18 May 2017
No messages
Last message: 127 days and 10 hours ago