Camelia, the Perl 6 bug

IRC log for #darcs, 2013-02-18

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

All times shown according to UTC.

Time Nick Message
00:23 whaletechno joined #darcs
00:39 Thaalos joined #darcs
00:45 mizu_no_oto joined #darcs
00:57 carter joined #darcs
04:48 preflex_ joined #darcs
06:49 lelit joined #darcs
08:29 raichoo joined #darcs
08:43 gh_ joined #darcs
09:41 dixie morning :)
09:42 gh_ hi!
09:42 exlevan joined #darcs
09:46 gh_ dixie, I hope you had a nice trip back
09:46 gh_ dixie, I was skimming through the bug tracker and found this: http://bugs.darcs.net/patch254
09:47 gh_ is there something we can adapt to current darcs, or do we mark the bundle as obsoleted?
09:49 dixie gh_: hi, trip back was fine
09:49 dixie gh_: that patch is obsoleted. Probably I need take second look on that issue
09:49 gh_ dixie, ok
09:53 dixie now I have the issue correction process in my muscle memory, hopefully it goes faster :)
10:15 iago joined #darcs
10:22 alexei joined #darcs
10:48 owst joined #darcs
10:56 gh_ hi owst !
10:58 owst hi gh_
11:02 donri is the sprint over?
11:02 owst It is
11:02 donri how did it go?
11:02 owst I am back in Southampton :-)
11:02 owst It was good - I made some good progress on getting the bridge working
11:02 owst Lots of good conversations
11:03 owst There'll be a blogpost at some poing
11:03 owst point*
11:04 donri canhas bridge RSN?
11:04 owst I understand one of those words :-p
11:04 donri \o/
11:05 donri will we have working bridge real soon now? :)
11:05 owst aha!
11:05 owst Sooner than it would have been without the sprint...!
11:06 donri :D
11:06 donri have you solved the branching problem at least in theory?
11:06 owst yes
11:07 owst tags everywhere!
11:07 donri on which ends?
11:08 owst when converting a git DAG to darcs patch-sequences, add extra darcs tags, to encode the DAG structure
11:09 donri yo DAG i heard you like tags
11:09 owst heh
11:09 donri incidentally my name is Dag
11:09 owst there are indeed tags inside tags, if that's what you're wondering
11:09 owst hah!
11:09 owst Well, we're encoding you inside darcs patches
11:09 donri actually my middle name is ganesh and heffalump is a nickname of my sister
11:10 owst :-p
11:10 donri and tag is german for dag (which means day)
11:11 * marmoute maps owst to "The english guy hammering a git bridge"
11:12 owst marmoute would be correct - I'm Owen
11:18 marmoute owst: you had a fairly amused face while I was talking about evolution.
11:18 marmoute But I did not find time to ask you about it.
11:18 kmels joined #darcs
11:19 owst It was a "interesting - that's cool" kind of face :-)
11:22 marmoute Ok
11:23 owst I didn't realise that you didn't consider having conflicts between your meta-versions.
11:23 owst So maybe I was thinking "eek, that will be scary" at that point ;-)
11:24 marmoute ETOOMUCHNEGATION
11:24 owst haha
11:24 owst sorry
11:24 marmoute "having conflicts between your meta-versions" ?
11:25 owst If two people evolve the same changeset, in different ways, what happens?
11:26 marmoute We can detect that as "divergent changesets"
11:26 owst But do you try to resolve the divergence, as some sort of conflict-resolution?
11:26 marmoute And we can properly merge the two digervent version.
11:27 owst what happens if the divergent versions conflict?
11:27 marmoute Yes the steps are (1) detect and warn the user about troubled changesets (2) offer automatic resolution using "hg evolve"
11:27 marmoute If the divergent version conflict we trigger merge tool as for standard merges
11:28 * owst thinks about that
11:31 owst Ah, but you don't create a DAG of evolutions?
11:31 owst When you merge two evolutions, that's it, you don't remember their parents?
11:36 marmoute I'm not sure to understand your question
11:36 marmoute Evolution create a (sparse) dag of history of each "changes"
11:36 marmoute So divergences is a branching (A -> A' and A -> A')
11:37 marmoute But when we resolve the divergences we have a "merge" in the evolution DAG
11:38 marmoute (A -> A', A -> A", A' -> A"', A" -> A"')
11:38 marmoute And the evolution DAG rooted at A have a single head again (A"')
11:40 owst Right, but once you have A"', do you store A" and A', or can they be thrown away? What if there is another evolution of A -> B, do you take A to be the conflict base between B and A"'?
11:41 marmoute A" and A' may be throw again (or not propagated)
11:42 marmoute If there is another we do "best effort" trying to find the higher common precursor of the divergent changesets
11:42 owst I see.
11:42 owst Thanks for the explanation :-)
11:43 marmoute You are welcome
12:30 whaletechno joined #darcs
13:06 mizu_no_oto joined #darcs
13:09 gal_bolle joined #darcs
14:13 markstos joined #darcs
15:14 nomeata joined #darcs
17:12 alexei joined #darcs
17:13 alexei_ joined #darcs
17:15 Igloo joined #darcs
17:52 gh_ joined #darcs
18:03 alexei_ joined #darcs
18:04 alexei_ joined #darcs
18:06 alexei_ joined #darcs
18:06 carter joined #darcs
18:07 raichoo joined #darcs
18:10 amgarchIn9 joined #darcs
18:10 Igloo joined #darcs
18:37 amgarchIn9 joined #darcs
18:41 * Heffalump appears
18:42 iago joined #darcs
19:02 Heffalump straw poll: I'm going to rename Darcs.Repository.FileMod to Darcs.Repository.PatchIndex. Should I rename Darcs.Patch.FileMod to Darcs.Patch.PatchIndex or Darcs.Patch.Index ?
19:03 markstos left #darcs
19:08 gh_ Heffalump, Darcs.Patch.Index
19:13 mizu_no_oto joined #darcs
19:14 Igloo joined #darcs
19:30 mizu_no_oto joined #darcs
20:11 Heffalump oops, using [..] in a patch text confuses the tracker
20:29 Weltraumschaf joined #darcs
20:32 Igloo joined #darcs
20:48 Igloo joined #darcs
21:25 lelit` joined #darcs
21:46 lelit`` joined #darcs
21:49 lelit`` joined #darcs
22:09 stepkut joined #darcs
22:51 Igloo joined #darcs
22:56 iago joined #darcs
23:05 sm hey all
23:07 sm I'm trying to build darcsden, darcs and ssh all at once with ghc. The current roadbump is that it can't see Darcs/Util/Encoding/IConv.hs, normally generated from Darcs/Util/Encoding/IConv.hsc. How do you usually tell ghc what to do with hsc files ? I don't see anything special in darcs.cabal
23:09 Heffalump what can't see it?
23:09 Heffalump i.e. what is building at the time it fails
23:10 sm Heffalump: to be precise, sp --no-exts --no-default-map -o darcsden ghc-7.4.2 WebServer.hs -DDEBUG -DHIGHLIGHTINGKATE -DSSH_FLAG -DCLOSING -Wall -fno-warn-unused-do-bind -threaded -isrc -i../ssh/src -i../darcs/src -I../darcs/src -v --run
23:10 Heffalump what's sp?
23:10 sm more essentially: ghc-7.4.2 WebServer.hs -DDEBUG -DHIGHLIGHTINGKATE -DSSH_FLAG -DCLOSING -Wall -fno-warn-unused-do-bind -threaded -isrc -i../ssh/src -i../darcs/src -I../darcs/src
23:11 sm sp is searchpath, an old tool from the happs guys which is an awesome auto-rebuilder
23:11 Heffalump is there a reason you're not using cabal?
23:11 Heffalump (cabal is what handles the hsc2hs stuff)
23:11 Heffalump if you're using another scheme/building manually you'll need to manually run hsc2hs too
23:11 sm yes, debugging across multiple packages is a big pain, too many moving parts
23:12 sm and manual steps
23:12 sm righto, I thought ghc would figure it out
23:12 Heffalump also, if you've built the darcs directory once with cabal, the .hs file will be in dist/build/something
23:13 sm thx
23:17 sm that was a good tip, adding -i../darcs/dist/build gets me further
23:23 sm I wonder what HTTP http://hub.darcs.net/darcs/darcs-scr​eened/src/Darcs/Util/Download.hs#54 is trying to import
23:24 Heffalump good question. I suspect noone is testing that scenario.
23:26 Heffalump do you need it fixed?
23:27 sm I think not urgently, I have defined HAVE_HTTP
23:28 sm now Darcs/Util/Download/HTTP.hs wants Version which is a hidden module in darcs-2.9.4. hmm, right, I need to pick 2.8 or latest darcs
23:28 sm and not both at once
23:29 Heffalump against dist/build should help you with Version.hs - it's autogenerated
23:29 sm ahh.. -i../darcs/dist/build/autogen then
23:30 sm thanks
23:32 sm Darcs.Utils got renamed to Darcs.Util, right ? not finding that easily in history for some reason
23:33 Heffalump I think so. It also got broken up quite a bit.
23:34 Heffalump ok, so there's no Darcs.Util
23:34 Heffalump the functionality all ended up in Darcs.Util.something, for various somethings
23:35 sm I thought darcs changes src/Darcs/Util -s --only or darcs changes src/Darcs/Utils -s --only would show stuff
23:35 Heffalump I think the problem might be that you need to identify a file in the current repo
23:35 sm and indeed they do, I'm just not quite sure what it is
23:36 sm ah, --only was a bad idea here
23:43 gh_ joined #darcs
23:48 sm updating darcsden for latest darcs, http://hub.darcs.net/simon/dar​csden/src/DarcsDen/Darcs.hs#25 needs to provide an extra argument (a pair of paths) when calling the command. Any idea what those should be ?
23:48 sm http://hub.darcs.net/darcs/darcs-screened/pat​ch/20130115064314-5ef8f#src/Darcs/UI/Commands.hs was the darcs change
23:49 Heffalump gh?
23:50 Heffalump they both seem to be getCurrentDirectory in darcs, but perhaps one is obtained before switching to any --repo argument
23:51 gh_ sm, I pasted you the patches that provoked the API changes a couple days ago
23:51 gh_ let me find them
23:52 Heffalump gh_: those argumentsought to be documented somewhere - e.g. in commandCommand in Darcs.UI.Commands
23:52 gh_ yes
23:53 gh_ I thought runCommand provided them?
23:53 sm maybe darcsden should be using that
23:54 sm or not..
23:54 sm but what do they mean, out of curiousity ?
23:55 gh_ I think they're the absolute paths of the first and second potential arguments of the command to be run
23:55 gh_ but I gotta check that
23:56 gh_ sorry for the change, somehow I believed darcsden was using runCommand
23:56 sm I think http://hub.darcs.net/darcs/darcs-scr​eened/src/Darcs/UI/RunCommand.hs#136 provides them

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