Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2014-04-20

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

All times shown according to UTC.

Time Nick Message
01:58 dolio joined #darcs
02:02 edwardk joined #darcs
03:56 mizu_no_oto joined #darcs
07:09 slyfox joined #darcs
07:09 slyfox joined #darcs
07:13 raichoo joined #darcs
07:21 pavonia joined #darcs
07:22 pavonia Hi! http://www.haskell.org/haskellwiki/Darcs_vs._Git states that darcs runs the test script after every record. Is there a way to set a test script but prevent it from being run automatically (but still being able to run it manually)?
07:26 Heffalump from memory, put "record no-test" in ~/.darcs/defaults
07:26 pavonia Can I set this locally too?
07:28 Heffalump yes, but I can't remember how
07:28 Heffalump one sec
07:28 Heffalump I'm not even sure darcs still does this, because I don't seem to have any settings.
07:29 Heffalump I mean I don't have any settings to disable it and they don't seem to be running automatically
07:29 Heffalump so perhaps just try first
07:29 pavonia Ah okay, I haven't yet
07:31 pavonia Oh, it seems it doesn't run it automatically
07:31 pavonia Thanks Heffalump!
07:45 pavonia For the record, to run it automatically set "record posthook darcs check --test\nrecord run-posthook" in ./_darcs/prefs/defaults
07:49 pavonia left #darcs
08:24 hvr joined #darcs
08:26 hvr is there a site offering continous one-way darcs-to-git mirroring?
08:27 hvr or alternatively, is there a documented guide how to set up such an automated mirroring? (the property I'm after is that the Git-view's master branch only does fast-forward-updates)
08:43 mornfall hvr: code exists, I have tried to get hub.darcs.net maintainers interested in having that feature, but was sorta turned down
08:43 hvr mornfall: code exists where? :-)
08:44 mornfall hvr: all over the place, sadly ... you would need to update http://hub.darcs.net/mornfall/darcs-fastconvert to build with current darcs/ghc
08:44 mornfall hvr: darcs has incorporated part of that code, but it lacks support for incremental exports
08:45 hvr well, I only need incremental exports from darcs to git :-)
08:45 mornfall yes, I understand
08:45 mornfall but darcs convert --export won't do that
08:45 mornfall you can re-import the entire repo every time
08:45 hvr woudl that be stable at least?
08:45 mornfall if it's small, it probably won't hurt too bad
08:45 hvr (i.e. I'd need old commits to retain their commit-ids)
08:46 mornfall the result should be deterministic, yes
08:46 mornfall as long as the darcs repo is append-only
08:46 mornfall (the source I mean)
08:46 mornfall if you run optimize --reorder, or obliterate patches, the conversion will fall apart
08:46 mornfall (i.e. same thing that happens when you git rebase)
08:46 hvr do Darcs users to those operations a lot?
08:46 hvr s/to/do/
08:46 mornfall locally, yes, on the server, no
08:47 hvr that gives me hope :-)
08:47 mornfall it's really the same as git rebase
08:48 mornfall it's considered a "bad practice" to remove or amend patches in public repositories
08:48 mornfall (compounded by the fact that darcs will give you hell if you do... git sort of copes)
09:12 lelit joined #darcs
10:06 alexei___ joined #darcs
10:09 * Heffalump looks at inlining hashed-storage
11:39 alexei___ joined #darcs
13:14 edwardk joined #darcs
14:48 sm mornfall: aren't you giving up a little too soon ? I think there's only been one brief chat about that feature. If you're willing to do the work and can motivate it, it's certainly interesting
14:51 sm I think the target users and value was not yet clear
15:46 gh_ joined #darcs
15:54 edwardk joined #darcs
16:05 gh_ sm, mornfall, I think it was clear and actually the majority of the small group that participated in the conversation was interested
16:13 gh_ http://blog.johantibell.com/201​4/04/announcing-cabal-120.html
16:28 sm gh_: great
16:29 gh_ cabal freeze seems interesting to reproduce bugs
18:24 IbnFirnas_ joined #darcs
19:07 thorkilnaur joined #darcs
19:20 dolio joined #darcs
19:26 Heffalump I am interested, I just want to do it right rather than encouraging people to use a slightly broken conversion
19:27 Heffalump but that might take a while
19:54 Heffalump gh_: FYI I'm working on the internalisation but the Setup.hs files will need merging so it's not entirely trivial. An alternative would be to make hashed-storage-darcs.
20:19 mornfall well, the only thing you need is the endianity check, and darcs already does that
20:19 mornfall (it only uses -DBIGENDIAN but changing the code to that effect should be trivial)
20:24 mornfall as for history of conflicts (which I assume is the slightly broken bit you refer to), I don't think it's entirely reasonable to expect that a git conversion would be able to represent that, when you can't really get that info with darcs either
20:32 Heffalump I'm sure I already described how to represent it :-)
20:32 Heffalump it's not the conflicts themselves that bother me, it's the loss of information about the actual patches that conflicted - you completely lose what it contained originally
20:34 mornfall what I am saying is that I (on a user level) can't get that info from darcs anyway
20:35 Heffalump yes you can, you just look at the patch..
20:35 mornfall it says conflictor [ ... ]
20:35 Heffalump if you have A;B;resolve_A_B then in darcs you just look at B (perhaps pull it into a fresh repo)
20:35 Heffalump after the git conversion you can never get back what was originally in B
20:37 mornfall it's not like the conversion magically consumes the darcs representation of the repository though
20:37 mornfall afterall, it's supposed to be a mirror
20:37 Heffalump I'm not sure I understand what you mean. I just don't like losing information and the real contents of B strikes me as pretty significant information.
20:40 mornfall the information is not lost as long as you keep the darcs repo around, it just isn't available in the git view
20:41 Heffalump ah, right
20:41 Heffalump yeah, I guess it depends on what you want it for
20:43 mornfall it's more of a concern for convert --export, which is more geared toward final conversion
20:44 mornfall but convert --export already exists... for read-only mirrors on hub, I don't think it's such a big deal
20:45 Heffalump how do you make a read-only git mirror? Surely by definition the result is just as writable in a clone as a --exported version?
20:46 Heffalump anyway, I was mainly just explaining why I wouldn't implement something for hub based on the current code, not actively trying to stop someone else
20:46 mornfall read-only at the hub side of things... you won't be able to push into it
22:10 whaletechno joined #darcs
22:28 edwardk joined #darcs
22:53 alexei___ joined #darcs
23:10 raichoo joined #darcs
23:52 mizu_no_oto joined #darcs

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