Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2014-05-24

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

All times shown according to UTC.

Time Nick Message
00:32 dolio joined #darcs
00:41 edwardk joined #darcs
03:53 alexei joined #darcs
04:46 c74d joined #darcs
06:56 edwardk joined #darcs
08:03 lelit joined #darcs
08:14 edwardk joined #darcs
08:38 alexei joined #darcs
09:00 * Heffalump appears
09:47 alexei joined #darcs
09:50 mdiaz joined #darcs
09:53 Heffalump hi mdiaz
09:54 mdiaz Hi Heffalump, sorry for the delay.
09:54 Heffalump no problem
09:55 Heffalump would you like to start with an update on how things are going?
09:59 mdiaz Well, I already did the first items of my project.
09:59 mdiaz I completed the garbage collection of the respositories.
10:00 mdiaz I have a new patch for the bucketed cache.
10:00 mdiaz And also,  a patch for garbage collecting the global cache.
10:02 Heffalump ok, great - what are the next steps for that work?
10:08 mdiaz I'll send the patch for the bucketed cache, so you can review it.
10:09 Heffalump one of the parts of the plan for bucketed cache was to benchmark, right?
10:09 mdiaz Yes.
10:09 Heffalump but feel free to send the patch (or put it on hub.darcs.net) whenever you'd like feedback
10:10 mdiaz I made some benchmarks.
10:10 mdiaz But I found that there is no gain in time.
10:10 Heffalump oh :-)
10:11 Heffalump what systems are you benchmarking on? I've found that a large global cache is a serious problem on Windows
10:11 mdiaz So it would only be useful when we want to view the files.
10:11 mdiaz Like using the ls command.
10:12 mdiaz I tried it in several file systems.
10:14 Heffalump which ones?
10:16 mdiaz EXT 3, EXT 4, NTFS, FAT.
10:16 Heffalump I'm surprised by the NTFS/FAT results given my experience on Windows.
10:17 Heffalump One thing Guillaume suggested was that you (perhaps in cooperation with Ale) should revive darcs-bench
10:18 Heffalump and in general it would be good if you could produce the benchmarks in a way that's easily reproducible
10:18 Heffalump that way I can compare with my own experiences on my machine
10:19 Heffalump sorry, not revive darcs-bench, integrate benchmarking into the cabal file: http://blog.johantibell.com/2012/04/cabal-bench.html
10:19 Heffalump http://irclog.perlgeek.de/darcs/2014-05-19#i_8743516
10:20 mdiaz I tested the files systems under linux, maybe that influenced the results.
10:20 Heffalump oh, yes, that's unlikely to be realistic
10:20 Heffalump do you have any access to a Windows machine?
10:20 Heffalump and MacOS for that matter
10:21 mdiaz Currently no, I've to install windows.
10:22 Heffalump ok. Well in any case if you can make your tests easily reproducible, then we can ask other people (including me) to run them too
10:22 mdiaz Ok.
10:22 Heffalump If there's really no speedup from bucketed caches we'd have to think about whether it's worth actually merging it.
10:23 mdiaz Yes :/.
10:23 Heffalump it'd still be useful to have concrete results showing we didn't need it, and benchmarks we could rerun later if we thought things might have changed, so you work wouldn't be wasted in that regard
10:25 mdiaz Good.
10:26 Heffalump ok, so another thing I'd like to talk about briefly is logistics given that you aren't able to follow the exact GSoC timetable
10:26 Heffalump Guillaume said you'd be getting back to the project full-time on/just after June 13th, right?
10:29 mdiaz Actually, June 15.
10:29 Heffalump ok
10:31 mdiaz I just sent the patch for the bucketed cache.
10:32 Heffalump I guess that's 4 weeks after the start, so it would be good if you could get one or two more blogposts out before you get back to update on the progress you've made - doesn't need to report finished results, just how things are going
10:32 mdiaz Still I have to keep testing it. But it is better if you see it.
10:32 Heffalump great, thanks, I'll take a look at it
10:33 mdiaz Thanks.
10:34 Heffalump When would you like to meet next - would two weeks from now make sense? I'll need to work out exactly when I'll be available though. Once you're back to the project full-time let's meet weekly and aim for weekly blogposts.
10:35 Heffalump and obviously we can discuss things by email etc in the meantime, it's just useful to have a regular meeting too
10:36 mdiaz Ok, I'll do a few more posts so I can explain better the current status of the project.
10:37 mdiaz I have to go for lunch, but I'll be connected if you need me, ok?
10:37 Heffalump no problem, I think that's about all there was to discuss anyway
10:38 Heffalump if you can hang around on IRC when possible it sometimes helps with spontaneous discussions (either with me or anyone else)
10:40 mdiaz Ok.
10:53 mdiaz joined #darcs
11:30 favonia joined #darcs
15:15 edwardk joined #darcs
15:21 edwardk joined #darcs
15:54 john3213 joined #darcs
15:59 john3213 left #darcs
16:03 mizu_no_oto joined #darcs
17:36 gh_ joined #darcs
18:00 gh_ joined #darcs
18:25 alexei joined #darcs
18:51 mizu_no_oto joined #darcs
19:14 gh_ hi
19:14 c74d joined #darcs
19:33 alexei joined #darcs
19:50 edwardk joined #darcs
19:53 Heffalump hi
21:08 edwardk joined #darcs
21:19 gh_ joined #darcs
21:30 gh__ joined #darcs
21:49 gh__ joined #darcs
22:03 mizu_no_oto joined #darcs
22:06 gh__ joined #darcs
22:14 gh_ joined #darcs
22:30 mizu_no_oto joined #darcs
23:01 edwardk joined #darcs
23:04 gh_ joined #darcs
23:08 mizu_no_oto joined #darcs
23:31 c74d joined #darcs

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