Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2015-09-29

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
02:50 IcyFoxy sm: If I'm doing things right - so far, yes.
02:50 IcyFoxy sm: Or was that for the multi thread? Haven't tried that yet
03:06 sm IcyFoxy: it was re your darcs patch theory in rust paste
03:07 IcyFoxy sm: Ah :)
03:07 IcyFoxy So yes, so far (so long that I'm doing things right); which ... I think I am :)
05:20 dolio sm: Did all repos that use your travis key get deactivated at some point?
05:21 dolio Apparently we were using it for primitive and vector (for some reason) and those both got turned off.
05:24 IcyFoxy travis supports darcs now? Or are these repos on git?
05:29 dolio github
05:30 IcyFoxy Right
05:31 IcyFoxy I was hoping darcsden for a moment there :P
05:31 dolio I doubt travis will ever do that.
05:32 dolio Considering they don't even cover bitbucket.
05:32 IcyFoxy Oh?
05:32 IcyFoxy I thought it worked with any git repo that it can reach? Hooks are just further integration
05:32 dolio I think. There is other CI for bitbucket, but I don't think travis is an option.
05:33 IcyFoxy Noted.
05:33 dolio I think it's github only.
05:33 * IcyFoxy checks docs
05:33 dolio I've been pondering setting up my own CI for darcs stuff, but I don't think anyone offers anything automatic.
05:34 dolio So you have to do something like rent a VPS and set up buildbot.
05:37 IcyFoxy Speaking of CI, I might end up writing a RCS-tracked build logger for my darcs-rs, that is the Ci build results would be stored in the repository (or rather, a dedicated branch). likely also reproducable builds.
05:45 dolio That'd be interesting.
08:18 IcyFoxy ^ Will likely be integrated into the package manager I'll also be working on. Which I won't be limiting to just Rust. Structuring it much like emerge and using directory structure to compile libraries/binaries/examples/tests, etc.
11:04 mizu_no_oto joined #darcs
12:12 IcyFoxy Is coalesce anything more than an optimization?
12:54 IcyFoxy Current rust implementation is at 150 sloc (excluding tests).
13:04 IcyFoxy https://gist.github.com/james-darkfox/b5188b0ad8457114ea7b
13:19 Riastradh joined #darcs
15:01 siel joined #darcs
16:59 Heffalump IcyFoxy: it's quite important for the UI of amend-record
16:59 Heffalump otherwise you'll end up with some quite weird patches
17:33 Heffalump IcyFoxy: I'd like to integrate issue tracking into the VCS, integrating CI too sounds interesting
18:38 sm dolio: not sure what you mean.. I use travis in one repo, no problems recently
18:42 Heffalump IcyFoxy: fair enough re parallelizing, but I think in practice you'd find dependencies would limit the available parallelism
18:45 dolio sm: Dunno. hvr says he's seen travis turn off with no explanation before, so I guess it's just something that happens.
18:47 dolio By turn off, I mean no new pull requests or commmits with automatically trigger builds.
18:48 dolio Will, even.
18:52 sm I haven't seen that yet
18:52 dolio Count yourself lucky, I guess. :)
18:52 dolio You are simonmichael, though, right?
18:53 dolio On github/travis.
18:53 sm I know there's a switch you can toggle in github settings, also a commit tag that disables it for a commit ([ci skip] or something)
18:53 sm yup
18:54 dolio I wonder why the primitive and vector projects were set up to use your credentials to build.
18:54 dolio You didn't use to maintain them, did you?
18:54 sm eh, certainly not.. got a link ?
18:54 dolio Well, they're not using your credentials anymore.
18:56 sm I'm a member of the haskell libraries team, which is their maintainer...
18:56 sm I just used travis for the first time a few weeks ago
18:57 dolio That's even odder, then. :)
18:57 sm I guess this is a bit off topic for #darcs
18:57 dolio Of course, I have very little idea how this all works.
18:57 sm magic!
19:25 gh_ joined #darcs
20:30 gh_ interesting talk https://github.com/meiersi/HaskellerZ/tree/master/meetups/20150531-ZuriHac2015_Johan_Tibell-Performance
20:30 gh_ I wonder how much of this can apply to darcs code
20:50 dolio Yeah, actually, I've been wondering how much anyone has looked into performance optimization of darcs.
20:50 dolio Beyond high level algorithm stuff.
20:51 dolio That's why I built with profiling the other day, to see where time was spent. But it was mostly compression.
20:54 gh_ dolio, can we improve something about it?
20:55 dolio About compression? I haven't looked at all.
20:55 gh_ dolio, I remember "darcs diff" had a big memory use
20:55 dolio I don't even know how the compression in darcs is implemented.
20:55 gh_ dolio, we depend on the zlib haskell module
20:57 dolio Anyhow, that was just for a get of darcs.
20:57 dolio So I think it got a huge pack, and spent most of its time processing that.
20:58 dolio So it's probably not representative of a bunch of other stuff.
21:01 gh_ oh, yeah it would be more interesting to get a repo without packs
21:01 gh_ also, big local merges would be interesting
21:02 dolio Oh, there's --no-packs.
21:20 dolio Well, that's quite different.
21:21 dolio Most time spent in waitNextUrl, which I assume is just downloading. But then after that is catchNonSignal.
21:21 dolio Which also does the most allocation.
21:22 dolio I guess that's misleading, though, because it's stuff umderneath it that's doing the actual stuff.
21:22 dolio Like file copying.
22:41 mizu_no__ joined #darcs
23:32 dino-_ joined #darcs

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary