Camelia, the Perl 6 bug

IRC log for #darcs, 2013-03-08

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

All times shown according to UTC.

Time Nick Message
00:29 favonia joined #darcs
00:30 preflex joined #darcs
00:39 mornfall joined #darcs
00:39 sm joined #darcs
01:16 mizu_no_oto joined #darcs
02:01 mizu_no_oto joined #darcs
02:19 intripoon joined #darcs
03:16 mizu_no_oto joined #darcs
04:08 carter joined #darcs
04:54 preflex_ joined #darcs
05:44 felipe joined #darcs
06:27 delamonpansie joined #darcs
07:31 lelit joined #darcs
08:10 lelit joined #darcs
08:20 raichoo joined #darcs
09:09 amgarchIn9 joined #darcs
10:40 owst joined #darcs
11:05 iago joined #darcs
11:28 lpsmith Given that git appears to have won,  would it be possible to rewrite (parts of) darcs so that it would store (most of) the repo in git format,  but use darcs merge and as "porcelain"?
11:29 lpsmith basically, to use darcs as the UI and merge algorithm on top of git,  that's what I mean
11:30 lpsmith It appears that somebody's already written a haskell binding to libgit2, even,  although I don't know how good they are.   libgit2 is supposed to be pretty nice though. http://hackage.haskell.org/package/hlibgit2
11:30 owst lpsmith: One could do, yes.
11:31 owst As for who wants to do that, I'm not sure. David wrote something called iolaus that I think was somewhat like you describe
11:33 lpsmith yeah, I've seen that though I've never played with it
11:34 lpsmith I don't think he got as far as getting darcs merge to play with git though
11:34 lpsmith My impression was that it was mostly a darcs-like interface on the surface.
11:35 owst It's something we did talk a bit about, at the last sprint (using darcs to perform git merges), but I don't think there were any plans for anyone to work on it
11:35 owst I don't know what iolaus is/does, I've never looked further than the github project page :-)
11:37 lpsmith But to tell you the truth,  like many other darcs users I use git almost exclusively these days,  because that's what people are using,  and because github is awesome
11:37 owst Fair enough
11:37 lpsmith And that was just a passing thought about how darcs might stem the tide
11:38 owst I don't think so, but ok :-)
11:40 lpsmith Well,  it seems to me that git's conceptual structure is fairly agnostic about a lot of things,  like merge algorithms and UI.    (Which is kind unfortunate because the UI sucks.)
11:41 lpsmith I don't know enough about the internals of darcs to know what kind of special problems one might run into basically trying to have darcs be a UI and merge algorithm on top of git.
11:41 owst Yeah, it is, other than once all is said-and-done it stores complete snapshots.
11:42 lpsmith right, which might be a bit of an efficiency issue having to recompute the patches on a fresh checkout or whatever,   but is there anything really fundamentally incompatible with that?
11:48 owst I can't think of an example right now, but I think there could be a problem with information loss after a merge
11:48 lpsmith right,  yeah,  I was just thinking that people do have finer control over what a patch is than what you could necessarily compute from a set of snapshots
11:48 owst Indeed, something along the lines of those "git is inconsistent" arguments comes to mind
12:08 Heffalump lpsmith: our plan is for darcs bridge to stem tie tide
12:08 Heffalump i.e. use github etc, but a darcs repo locally
12:21 lpsmith Heffalump, but what if you then lose your local repo and want to re-download it from github?  Wouldn't the patches still need to be recomputed, possibly losing some information?
12:22 lpsmith In  my mind,  that problem isn't that much different from the one I posed.
12:38 lelit joined #darcs
12:45 Heffalump lpsmith: yeah, not ideal. I think we'll want to mitigate that by (a) linking up hub.darcs.net and github and (b) encoding all info about patches recorded on the darcs side in git metadata so you can recreate them
13:07 mizu_no_oto joined #darcs
13:18 mizu_no_oto joined #darcs
14:36 mizu_no_oto joined #darcs
14:48 schlaftier joined #darcs
15:27 donri joined #darcs
16:55 saep joined #darcs
16:58 sm g'day all
17:20 Weltraumschaf joined #darcs
17:21 Weltraumschaf hiho
17:25 Weltraumschaf someone here?
17:26 Heffalump sort of
17:34 Weltraumschaf some using my jenkins darks plugin has problems i can't reproduce. he got the error: .: openBinaryTempFile: permission denied
17:35 Weltraumschaf obviously he has a permsion error...
17:35 Weltraumschaf is there maybe another problem? he is doing a darks get http://...
17:35 Weltraumschaf for me it works with darks 2.5 and darks 2.8
17:36 Heffalump permission denied can mean something else has the file open
17:37 Heffalump is it intermittent?
17:39 Weltraumschaf he says it is permanent
17:39 Weltraumschaf https://github.com/Weltraums​chaf/darcs-plugin/issues/11
17:40 Weltraumschaf but he only tried once with this repo. but had the problem also with an other repo
17:42 Weltraumschaf i remember something about this. i think i had a related problem with openBinaryTempFile. but can't remember details. does darks need special access rights for this tmp file?
17:51 carter joined #darcs
17:52 raichoo joined #darcs
17:57 Heffalump and what happens if you try to do the darcs get from the same directory on the machine ( /var/lib/jenkins/jobs/ounit/workspace)?
17:59 Weltraumschaf i have no access on his machine
17:59 Weltraumschaf on my machines (mac os and debian) works well
18:03 Weltraumschaf oh, just tried again on my debian machine and got the same error
18:06 Weltraumschaf i got a slitty different message: darcs failed:  Not a repository: http://darcs.ocamlcore.org/repos/ocaml-fileutils  (.: openBinaryTempFile: perm...
18:08 Weltraumschaf if i do the command as the jenkins user on shell it works, but throws a other strange error: arcs: /var/lib/jenkins/.darcs/cache/patches/​0000041785-0939451f4116c706c114a7d4729​a097e7d3836711c8e9efae2b0ac4da2b63324-​new_4b653b6aa8b12df462cfae589423c377: getSymbolicLinkStatus: does not exist (No such file or directory)
18:15 Weltraumschaf i can reproduce it http://www.weltraumschaf.de​:8080/job/foofoo/1/console
18:15 Weltraumschaf but doing the command via shell works fine
18:16 Weltraumschaf strange. the jenkins plugin does nothing else, than calling a darcs get as user jenkins
18:30 Heffalump the getSymbolinkLinkStatus thing is probably harmless
18:31 Weltraumschaf i've deleted the .darcs/caches problem exists anyway
18:32 Weltraumschaf the `darks get` via java proc call must be slightly different than shell
18:53 whaletechno joined #darcs
18:58 mizu_no_oto joined #darcs
19:22 Weltraumschaf ahhh, that guy strafed the problem: "So darcs for whatever crazy reason, try to create a file in ./ (so where the program is located)."
19:22 Weltraumschaf more details https://github.com/Weltraums​chaf/darcs-plugin/issues/11 if u're interested
19:38 `nand` joined #darcs
19:53 lelit joined #darcs
20:36 amgarchIn9 joined #darcs
20:52 amgarchIn9 joined #darcs
21:53 mizu_no_oto joined #darcs
22:01 mizu_no_oto joined #darcs
22:07 schlaftier joined #darcs
22:14 amgarchIn9 joined #darcs
22:57 saep joined #darcs

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