Camelia, the Perl 6 bug

IRC log for #darcs, 2013-02-28

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

All times shown according to UTC.

Time Nick Message
01:45 dgpratt joined #darcs
01:46 dgpratt is darcs known to work (or not work) on Windows 8?
01:46 dgpratt when I try to 'darcs get...' it crashes
01:47 dgpratt I tried running it under Win7 and then XP compatibility mode, no dice
01:57 favonia joined #darcs
02:29 intripoon_ joined #darcs
02:46 mizu_no_oto joined #darcs
02:58 mizu_no_oto joined #darcs
04:15 mizu_no_oto joined #darcs
04:37 preflex_ joined #darcs
06:39 preflex joined #darcs
06:44 lelit joined #darcs
06:54 Igloo joined #darcs
07:12 schlaftier joined #darcs
07:16 favonia joined #darcs
07:30 carter joined #darcs
08:28 raichoo joined #darcs
10:12 gal_bolle hi everyone
10:29 iago joined #darcs
10:46 amgarchIn9 joined #darcs
10:47 owst joined #darcs
11:11 lelit` joined #darcs
12:30 C-Keen_ joined #darcs
12:33 donri joined #darcs
13:06 dgpratt joined #darcs
13:09 Heffalump dgpratt: I haven't heard any reports either way about darcs on windows 8. What goes wrong for you?
13:13 mizu_no_oto joined #darcs
13:41 dgpratt very strange, very strange indeed
13:42 dgpratt when I run it under the control of a debugger, it does not crash
13:42 dgpratt I've pasted the event log entry of the crash here:
13:42 dgpratt http://hpaste.org/83202
13:43 owst libcurl strikes again!
13:43 dgpratt owst: indeed, it appears so
13:44 Heffalump grmph - you could build -f-curl
13:44 Heffalump but the alternative (HTTP) is slower and doesn't support https
13:45 dgpratt or I could just continue to use it via the debugger :D
13:45 Heffalump :-)
14:08 gh_ joined #darcs
14:14 mizu_no_oto joined #darcs
14:45 lelit` joined #darcs
14:55 dgpratt left #darcs
15:21 mizu_no_oto joined #darcs
15:25 amgarching Hi! I do "darcs unpull -a" and some patches are left. Same when I answer "a" to "darcs unpull". Is this intended behaviour?
15:25 sm amgarching: -a stops at tags. darcs show tags will tell if you have some
15:26 sm morning all
15:27 amgarching morning, though it is more like evening here
15:29 amgarching I see "obliterate -a" works the same. Is there any difference between the two yet? So if one tags regularly there is no easy way to empty repo by darcs means?
15:31 owst amgarching: why do you want to empty the repo?
15:31 amgarching This is the first time I encounter this. My first darcs patch is dated by March 2005. Was the behaviour changed?
15:31 owst I thought obliterate was just an alias to unpull (or vice-versa)
15:31 amgarching unpull was there first
15:31 owst No, I think this was always the behaviour
15:31 owst oh yeah, of course it was
15:31 amgarching I need to keep up with the changes in behaviour
15:32 owst I'm not sure there has been any in this case
15:34 amgarching owst: empty the repo and pull only early patches is still the easiest way to examin early revisions. I use darcsweb to browse the changes. But sometimes one wants a snapshot to look at.
15:35 owst I see, I suppose normally you'd want to obliterate to an old tag (I'm assuming that darcs correctly obliterates under newer tags, if you specify --from-tag)
15:41 sm amgarching: to see an older snapshot, I think I'd more likely do darcs get --lazy . old --tag TAG or --to-patch PATCH . Might be a little slower if it's a recent snapshot and long history, but it preserves my working copy
15:42 sm but I too have wanted a way to bypass tags with -a
15:43 sm to see just one (or a few ?) files, there's also darcs show contents ...
15:44 sm but I guess you want to browse in darcsweb, so changing the working copy is quickest after all
15:47 yitz joined #darcs
15:47 sm darcsden should acquire that feature
16:20 favonia joined #darcs
16:44 stepkut joined #darcs
16:53 lelit` maybe trac-darcs may be convenient for that, although it's not as light to install...
17:03 sm I was wondering about darcsden using acid-state, no need for couchd or redis
17:03 sm (speaking of lighter installs)
17:37 Heffalump sm: as in the haskell package?
17:37 amgarchIn9 joined #darcs
17:45 sm yes, I notice eg clckwrks is using it
17:49 donri joined #darcs
17:54 sm stepkut: are you finding acid-state scalable and good to work with ? I like your cli tool
17:56 stepkut sm: I have never had an issue.. obviously, at some point RAM becomes an issue. But, that will hopefully spur the developed of the solutions to that problem ;)
17:56 stepkut the first problem is to have enough users to have the problem
17:56 sm ah right.. TAM
17:57 sm RAM even. That would be the problem for me
17:59 stepkut how big is your DB?
18:00 stepkut interestingly, many people suggest that RAM is an issue, and then they try to cram a majority of their working data into memcached servers because disk is too slow :)
18:00 sm I'm not sure how to measure it, but not big. But I'm always short of ram on this vps
18:01 sm it seems pretty likely acid-state will use more than couchdb, now or later. But better to try it of course
18:02 stepkut happstack.com, clckwrks.com, seereason.com, and a bunch of other sites all run on a VPS with 2GB of ram
18:02 stepkut most of those sites require almost no RAM, but one of the ones I didn't mention is actually a real web application with a bunch of data
18:04 * sm nods
18:04 stepkut patch-tag used acid-state and RAM was never an issue, AFAIK
18:04 stepkut the problems with patch-tag had far more to do with bad coding than the framework :)
18:05 sm I run a lot of things on this vps. darcs hub should get its own if it grows though
18:05 stepkut any more thoughts on using the POST feature of darcs send?
18:05 stepkut on the server-side, you just get what would have gone in the email body as the POST body
18:06 sm I guess patch-tag had to do a lot of stuff in an earlier, more bleeding-edge time
18:06 sm it sounds good, I'd like POST to do something useful too
18:07 stepkut sm: i think it had more to do with hartman still being a novice haskell programmer at the time
18:07 sm yes probably
18:07 sm also
18:07 stepkut I've seen the code :)
18:08 mizu_no_oto joined #darcs
18:08 stepkut though, a lot of the errors are the same ones that hub faces -- darcs/ssh going south
18:09 sm yes, a herd of darcsen is not an easy thing to put a reliable web interface on
18:10 stepkut yeah.. I'm letting you do all the dirty work :)
18:10 stepkut I have some (crappy) darcs and bugs plugins for clckwrks
18:11 stepkut once you get the low-level code working, I can try to rip it out and make it into a plugin ;)
18:12 sm well, darcsden and libdarcs already talk to each other pretty well, I'm just chipping away at corner cases and usability problems
18:12 sm and pretty infrequently right now. I'd like to make it more enticing to others to hack on
18:13 stepkut last time I tried to use libdarcs I ran into the problem that it wanted to modify the CWD of the server
18:13 stepkut is that still the case?
18:13 sm darcs itself is also a moving target
18:13 sm IIRC that's still the case, but darcs hub works around it
18:14 stepkut how ?
18:15 sm I'm having a hard time locating the issue.. one sec
18:19 sm ah.. basically by running without -threaded
18:19 sm cf http://hub.darcs.net/simon/darc​sden/patch/20120806224259-3c3f9 and http://hub.darcs.net/simon/darc​sden/patch/20120805230035-3c3f9
18:19 stepkut so.. only one external process at a time can run?
18:20 amgarchIn9 joined #darcs
18:20 sm I haven't verified, but that sounds likely
18:20 stepkut though.. libdarcs is not an external process ..
18:20 sm well, I just don't know. Right, it's not. Only the ssh server calls external processes.. and I think I've seen multiple of those active
18:20 stepkut also.. i find calling external executables hangs all the time if I don't use -threaded :-/
18:21 sm oh I guess the ssh server *is* built threaded
18:21 sm just not the web app
18:21 stepkut maybe hub does have problems with multiple overlapping calls to libdarcs.. but that just doesn't happen very often?
18:22 sm I'm fairly confident that was tested and is not a problem
18:22 stepkut :)
18:22 sm it's not fresh in my mind :)
18:22 sm but I did investigate that pretty closely and have no open issues that seem of that sort
18:23 stepkut k
18:24 sm I shouldn't say more about it, I could well be confused. I guess what I want to say is I believe current darcs hub is pretty robust, with only deterministic failures
18:24 sm all noted in the issue tracker
18:39 schlaftier joined #darcs
18:43 raichoo joined #darcs
19:05 lelit` joined #darcs
19:06 gbeshers joined #darcs
19:31 stepkut dice 3d5+3
19:31 stepkut oops
19:31 stepkut wrong channel
19:51 epta Installing darcs-2.8.4 failed https://drone.io/github.com​/biegunka/biegunka-darcs/8
19:51 epta How can I fix it?
20:09 Heffalump epta: install the dev libraries for libcurl
20:09 Heffalump there's some weird interaction we haven't figured out yet
20:11 epta libcurl4-gnutls-dev ?
20:11 Heffalump I think that depends on what libcurl4 package you have
20:12 Heffalump but that sounds plausible, it's what i have
20:12 Heffalump actually i have libcurl3-gnutls and libcurl4-gnutls-dev, which is a bit confusing :-)
20:14 mizu_no_oto joined #darcs
21:02 stepkut joined #darcs
21:46 sm joined #darcs
21:50 favonia joined #darcs
21:52 sm joined #darcs
21:55 sm belated: darcs hub will be down for a 7m service window to resize disk
22:15 mizu_no_oto joined #darcs
22:45 iago joined #darcs

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