Camelia, the Perl 6 bug

IRC log for #darcs, 2013-08-02

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

All times shown according to UTC.

Time Nick Message
00:09 dolio joined #darcs
00:22 edwardk joined #darcs
00:36 mizu_no_oto joined #darcs
01:08 kofno joined #darcs
01:26 favonia joined #darcs
02:31 mizu_no_oto joined #darcs
03:02 mizu_no_oto joined #darcs
03:47 preflex_ joined #darcs
03:49 kofno joined #darcs
04:09 jgross_ joined #darcs
04:13 idnar joined #darcs
04:13 idnar joined #darcs
05:51 adnap joined #darcs
06:29 lelit joined #darcs
06:52 mulander joined #darcs
07:55 edwardk joined #darcs
08:11 alexsuraci_ joined #darcs
08:38 amgarching joined #darcs
09:55 donri joined #darcs
10:17 owst joined #darcs
10:20 mulander joined #darcs
10:32 favonia joined #darcs
10:47 Modius_ joined #darcs
10:47 bfrank joined #darcs
11:17 bfrank_ joined #darcs
11:17 Modius joined #darcs
11:59 mizu_no_oto joined #darcs
12:00 raichoo joined #darcs
12:05 kofno joined #darcs
12:10 raichoo joined #darcs
12:13 konundra joined #darcs
12:29 bfrank__ joined #darcs
12:32 bfrank__ in darcs, when I am doing things interactively, is there a way to type a number to "skip" to?
12:48 mizu_no_oto joined #darcs
12:56 sm not really, other than echo nnnnnnnny | cat - | darcs ...  or something
14:13 mizu_no_oto joined #darcs
15:17 bfrank joined #darcs
15:19 dolio joined #darcs
15:57 amgarching joined #darcs
16:13 mizu_no_oto joined #darcs
16:54 amgarching joined #darcs
17:21 amgarching joined #darcs
17:39 konundra joined #darcs
18:07 lelit joined #darcs
18:21 mizu_no_oto joined #darcs
18:39 amgarching joined #darcs
19:24 _ilbot joined #darcs
19:24 Topic for #darcs is now http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.8.4 is out http://darcs.net/Releases/2.8
19:40 idnar joined #darcs
19:40 lambdabot joined #darcs
20:23 mizu_no_oto joined #darcs
20:23 edwardk_ joined #darcs
20:34 donri bfrank: for record there is an option to skip whole files, might help
20:34 donri answer ? for options
20:35 bfrank hmm, yeah, it just didn't have an option to skip a ton of pending records
20:35 Heffalump it sounds like something that cuold usefully be added to interactive selection
20:36 bfrank I was just curious if it had the ability
20:36 Heffalump I think it would be quite an easy code change if someone specs out the feature
20:36 sm in that situation I'll typically add file paths to limit the scope
20:36 donri how would you know how many to skip?
20:37 sm exactly, I can't imagine knowing that number
20:37 Heffalump hmm, good point. I was thinking that typing "10w" = pressing 'w' 10 times
20:37 donri i guess a number could be added to 'whatsnew' or something?
20:37 donri numbers*
20:38 bfrank I want to use a number to jump ahead, and use space to find the exact revision
20:38 donri bfrank: also, it's easy to amend patches with darcs. don't be afraid to record partial work as draft patches
20:39 bfrank yeah, I am gradually becoming more aware of the things it makes possible
20:39 bfrank one thing that is interesting, is that it is possible to pull from the wrong repo, into another repo
20:39 donri heh yeah. "set of patches" is a bit different from say the git history graph where you have to share some ancestor
20:40 bfrank mercurial queues kind of have a similar concept of patches, but it isn't "smart" enough to know which patches are dependent on one another
20:42 Heffalump darcs under-approximates dependencies, git and mercurial over-approximate them
20:42 bfrank it under-approximates them?
20:43 donri it can prevent you from doing things that would work and be useful. rebase lets you ignore that IIUC ;)
20:43 Heffalump bfrank: there might be a semantic dependency between two patches (e.g. add a method, use the method)
20:43 donri well i can't speak for what Heffalump meant
20:43 Heffalump that it won't see
20:44 donri ah
20:44 bfrank ahhh, but --use-deps is for that, though, right?
20:44 Heffalump yes
20:44 bfrank so it CAN support it, but requires a bit of intervention
20:45 bfrank then again, even if those are in seperate records, hopefully they are prefixed with the same prefix, so that they are included when you pull
20:45 bfrank I still hope one day optimize can be used to sort a repo by date
20:47 sm so, who has ever manually specified additional patch deps ?
20:48 * sm not yet
20:48 Heffalump I do occasionally
20:48 Heffalump usually with amend-record
20:49 donri sounds painful to do on principle, but i can imagine it useful on occasion
20:52 donri could be useful for github-esque pull requests with a darcs twist? "pull this one patch, the rest are dependencies"
20:52 donri with git you'd set up a feature branch and request the pull from that
20:53 Heffalump I think with darcs you just use bundles
20:54 donri if you're using darcs send, sure. i was thinking of darcshub and such
20:54 Heffalump yes, but I still think that's the right unit
20:54 sm I'm not understanding dnori
20:54 edwardk_ joined #darcs
20:54 donri which arguably should support send/bundles in some way but anyway :)
20:55 Heffalump sm: donri is suggesting having a pull request be represented by a single patch that depends on all the others that should be included in the pull request
20:55 Heffalump donri: aditya is working on that
20:55 donri cool
20:55 sm bsrk's latest bundles-support code is currently running at http://hub.darcs.net:8901
20:55 sm it looks like you can upload a bundle when you report an issue
20:55 sm and then.. merge the bundle from the issue page, right Heffalump ?
20:56 donri random thought: could also be nice if the patch pull checkboxes on the branches tab would automatically check+disable dependencies. probably not trivial work though...
20:56 donri (by disable i mean, you can't uncheck)
20:57 sm donri: agreed! and/or show deps without clicking. alexsuraci partly implemented that IIRC, it's currently disabled
20:57 donri aha
20:57 sm draws lines between depending patches
20:58 amgarching joined #darcs
20:59 sm speaking of hub, has anyone else been logging in with google and seeing intermittent errors ?
20:59 donri yes, i only got errors when i tried it, both github and google
20:59 sm :(
20:59 donri i should've reported that...
21:00 sm donri: both should be working right now
21:00 sm except I get an error from google login once in a while, then it succeeds on page reload
21:02 Heffalump joined #darcs
21:02 donri sm: account synced, i think that was google. now with github i got FailedConnectionException "api.github.com" 443
21:03 donri would be nice if the UI reflected what accounts have been linked successfully. and maybe option to unlink?
21:03 sm yes please
21:03 sm so both google and github are failing for you ? every time ?
21:03 sm both work for me
21:04 donri now with google ResponseTimeout
21:04 sm I think that's the one I get once in a while
21:04 donri now github synced
21:04 donri google still ResponseTimeout
21:06 sm donri: to be clear, you're seeing these problems when logging in ? or when clicking link with X in settings ?
21:06 donri settings
21:06 donri haven't even tried login yet
21:07 sm ok thanks. I haven't been able to reproduce yet. Aditya has some nice functional tests for this stuff, we should get him to stress test it
21:07 donri now google synced too
21:07 donri gonna try login
21:07 sm maybe we need a longer timeout or retry or something
21:08 sm hub's max request time is 30s
21:08 dcoutts joined #darcs
21:08 donri google failed login, github worked
21:08 donri it feels like it fails faster than 30 seconds though
21:08 sm it certainly should..
21:09 sm do you have packet loss or other net trouble right now ?
21:09 donri now google worked i think
21:09 sm when it works you'll see green
21:09 sm what browser ?
21:10 donri i don't think i've ever experienced packet loss o_O not that i've noticed
21:10 donri firefox
21:10 sm ping will tell for sure
21:10 donri yes i know it worked, i just wasn't sure i actually clicked google :p
21:11 donri tried again, google, first failed then worked
21:12 sm I don't see this level of failure, and haven't heard of it from anyone else, so I'm wondering why you see it
21:12 donri i'm in sweden; latencies?
21:12 Heffalump yeah, it's the dog slow internet connections in Sweden!
21:12 donri where are you?
21:13 sm ha. Can you run ping hub.darcs.net for a little just to rule out latency/packet loss
21:13 sm also to github and google
21:13 sm those well known slowpokes (actually github does have a lot of outages, but not right now)
21:14 sm I'm in los angeles, and hub is in fremont (a few hours north)
21:14 donri 20 packets transmitted, 20 received, 0% packet loss, time 19025ms
21:14 donri rtt min/avg/max/mdev = 206.989/208.654/213.466/1.642 ms
21:15 sm hub.darcs.net, github.com and accounts.google.com
21:15 donri that was hub
21:15 donri --- github.com ping statistics ---
21:15 donri 17 packets transmitted, 17 received, 0% packet loss, time 16020ms
21:15 donri rtt min/avg/max/mdev = 140.988/145.009/153.714/3.588 ms
21:16 donri --- accounts.l.google.com ping statistics ---
21:16 donri 15 packets transmitted, 15 received, 0% packet loss, time 14019ms
21:16 donri rtt min/avg/max/mdev = 44.441/45.424/49.830/1.328 ms
21:16 sm good, so that's not the problem
21:17 donri Heffalump: it is pretty crappy, at best some 30ms latency and 16 mbit downstream. i'm used to 100/100 mbit full duplex with about 2-3ms latency :p
21:19 alexsuraci_ joined #darcs
21:20 sm can anyone else reproduce these failures ?
21:21 sm ie a failure when clicking settings->link with X, or login->sign in using X
21:21 donri it seems to work every time in epiphany
21:22 donri maybe it's one of my many firefox extensions that block shit
21:22 sm interesting
21:23 sm I have firefox 22 with adblock lite/firebug/duckduckgo plus
21:23 sm could be some firefox preference too ?
21:23 sm but it's intermittent
21:24 donri seems to be working fine if i whitelist hub.darcs.net in disconnect
21:24 donri https://disconnect.me/
21:25 donri odd that it works sometimes though
21:25 donri hm google still failing even without disconnect
21:27 sm disconnect looks good, thanks
21:27 donri it's like ghostery but open source and not selling your data :p
21:28 sm \o/
21:29 sm nothing unusual in FF options -> advanced -> encryption I guess
21:30 sm maybe disabling all extensions is worth a try
21:30 kofno joined #darcs
21:31 donri ok google not working in epiphany either, only tried github before
21:31 amgarching joined #darcs
21:31 donri no extensions in epiphany
21:31 sm weird
21:35 sm I must eat, thanks for the troubleshooting donri
21:36 sm interested in anything else you discover. Maybe watching it in firebug net pane can give a clue
21:37 donri tried the firefox developer tools, no particular luck
22:31 dolio joined #darcs
22:41 bfrank_ does david ever join this channel?
22:43 sm no, bfrank_
22:43 Heffalump he's stopped working on darcs
22:45 bfrank_ hmm
22:45 bfrank_ he doesn't even stop by?
22:45 Heffalump 'fraid not
22:46 bfrank_ for something he put so much time and energy into, seems sort of sad that he walked away
22:46 bfrank_ did he leave on good terms?
22:46 sm probably needs to conserve brain space for Science
22:47 sm I think so, but it was long ago. You could check the list archive
22:48 bfrank_ it just seems like he would show up every so often, just to see how things have progressed since he left
22:48 Heffalump he actually plated around with git for a while - a project called iolaus
22:48 Heffalump s/plated/played/
22:48 sm bfrank_: you could ask him
22:49 Heffalump but I think he abandoned that after a while
22:50 bfrank_ I was just curious, I mean, I know for the things I am passionate about, it would be hard to completely walk away
22:50 bfrank_ I guess it is a testament to his trust in the people he left in charge
22:51 bfrank_ but darcs without david is basically arcs
22:52 sm http://physics.oregonstate.edu/~roundyd/codes.php , though it looks about a year old, seems to imply continued use of darcs
22:52 sm bfrank_: I think he felt pessimistic about making darcs correct and scalable
22:53 bfrank_ if I could run screened, with the patch index, and it worked as good as it sounds, it really sounds like a lot of the performance issues are getting addressed
22:53 sm I really would like to hear him interviewed about it, like on FLOSS Weekly except somewhere better
22:53 bfrank_ I saw an interview he did in 2004 or something, about darcs
22:54 bfrank_ or, I should rephrase that, I read an interview he did
22:55 bfrank_ I do think, mercurial queues have the advantage of applying patches faster than darcs does, though. Hopefully that will improve with time too
22:55 sm droundy was interested in a sound theoretical basis. darcs is still the only vcs which attempts that, apparently
22:57 Heffalump David actually had a quite intuitive approach to the theory - he wasn't too keen on formal maths or anything.
22:57 sm oh
22:58 sm someone needs to start the darcsminutes.com podcast and invite him as first guest :)
22:59 Heffalump http://darcs.net/Theory/MergersDocumentatiois david's writing
22:59 Heffalump at least mainly
22:59 Heffalump sorry, mangled that
22:59 bfrank_ ha
22:59 Heffalump http://darcs.net/Theory/MergersDocumentation is David's writing
22:59 sm that's good to know, I must reread that
23:00 sm refresh my darcs fundamentals
23:02 Heffalump don't read too much about how merges are actually performned, it's the darcs 1 stuff
23:03 Heffalump TBH the whole quantum physics analogy is quite unhelpful, IMO
23:03 bfrank_ hmm
23:04 sm Heffalump: righto, thanks
23:04 * sm stalks droundy's commits.. those darcs repos all stopped being updated in 2009
23:05 bfrank_ how much stress testing is darcs put through/
23:06 bfrank_ for example, has anyone tried to load the linux kernel tree into darcs just to see how it would perform?
23:07 sm wow, the "david roundy github" search is dramatically worse on duckduckgo than google
23:09 sm bfrank_: I don't think anyone has tried that for a couple of years
23:09 sm good idea
23:10 sm GHC's darcs repo is the largest known public one
23:10 Heffalump we need to get the bridge going properly and then use that as a basis for that kind of benchmark
23:11 sm why ?
23:11 sm oh to get it converted to darcs. darcs-fastconvert will probably do it though
23:12 sm right mornfall ?
23:12 sm no problemo
23:12 Heffalump yes, but I'd like to benchmark the bridge as well
23:12 sm you speak of the bridge like it's a real thing :)
23:13 sm owst has well and truly scared me away from it by now
23:13 Heffalump hence "going properly"

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