Camelia, the Perl 6 bug

IRC log for #darcs, 2012-12-29

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

All times shown according to UTC.

Time Nick Message
00:08 mizu_no_oto joined #darcs
00:38 owst joined #darcs
00:45 Heffalump @tell markstos http://hpaste.org/79935
00:45 lambdabot Consider it noted.
00:47 owst Sweet, my clean-ups are doing wonders there :-)
00:48 Heffalump I'm not really sure how I got to the top
00:48 Heffalump I've refactored a lot of code, I guess.
00:48 owst Heh, winner!
00:52 sm Heffalump: pretty cool :)
00:53 Heffalump the main goal of the darcs-utils project is to give the API a work-out - ideally all the utilities in there will have no copy-and-paste from darcs
00:53 Heffalump but also have useful utilities, of course :-)
00:54 sm great idea
00:57 favonia joined #darcs
01:00 mizu_no_oto joined #darcs
01:10 mizu_no_oto joined #darcs
01:24 mizu_no_oto joined #darcs
01:40 markstos joined #darcs
01:41 Heffalump markstos: did the @tell work? http://hpaste.org/79935, anyway
01:42 markstos Heffalump: Probably not. Nothing notified me that I had a message waiting, if that's what you mean.
01:42 lambdabot markstos: You have 1 new message. '/msg lambdabot @messages' to read it.
01:42 markstos There it is. :)
01:42 Heffalump ah, it doesn't notice joins, just talking :-)
01:44 markstos Neat! Thanks!  I'm going to bring up "darcs show authors" and stare at the differences.
01:44 Heffalump it would be easy enough to rework to produce intermediate data, but I'm not sure how to expose that usefully
01:47 markstos A flag could be used to specify an output file for it.
01:47 markstos A flag could be used to specify an output file for it.
01:51 Heffalump do you definitely want that? If so what format would you suggest?
01:51 * markstos thinks
01:53 markstos maybe a line-based report with:  "$path $author $no_lines_attributed_to_this_author_in_this_file" ?
01:53 markstos I think it would enable other interesting reports and graphs, like attribution-by-file-type or attribution-by-directory.
01:55 markstos I'm waiting on a "darcs patch index" build to complete. It's using only 4% of  my CPU, which makes me suspect it's I/O bound, despite having an SSD on this laptop.
01:58 markstos I'm not completely confident in my reading of the "iostat" output, but I would guess it's not bottlenecking on I/O. I'll just be patient.
01:58 Heffalump ok - I think $author should be last because it could have lots of random characters in it. What separator is sensible, given that $path might have spaces.
01:58 markstos Heffalump: Thanks for your work on this tool-- it's exciting. I look forward to testing it more.
01:59 Heffalump I could use show on a Haskell String, which would produce "./wibble/file with spaces"
01:59 * markstos thinks.... CSV, JSON, XML....  or what's a character that's not valid in a file path?
02:00 Heffalump I like CSV.
02:01 markstos Does that mean pulling in a CSV library, or did you have in mind hand-rolling it?
02:01 Heffalump the former
02:02 markstos It's certainly easy when using a CSV library.
02:02 markstos Using a standard format will avoid strange parsing problems later.
02:10 Heffalump do you care about ordering?
02:10 markstos No.
02:10 markstos It's easy to run it through a "sort" function later if that's desired.
02:21 Heffalump ok, pushed
02:21 Heffalump I'll probably get bored of it soon, so get your feature requests in quick :-)
02:23 markstos Thanks! I just saw in the source there's some parallelism. How does it know "how parallel" it can run? (Ie: how many cores?)
02:24 markstos Found it: http://hub.darcs.net/ganesh/darcs​-utils/patch/20121229022037-81bb2  Thanks!
02:25 Heffalump +RTS -N will use as many threads as your machine has cores (GHC works that out, can't remember how off the top of my head)
02:25 Heffalump +RTS -N<num> will use <num> threads
02:26 markstos Nice.
02:28 markstos It does re-order the top contributors significantly when compared to "darcs show authors" for the darcs projects.
02:29 markstos i can see it provides a better sense of who the active contributors are, balanced with people who've made contributions that have survived over time.
04:07 alexei_ joined #darcs
06:28 mizu_no_oto joined #darcs
11:42 amgarchIn9 joined #darcs
13:46 owst joined #darcs
13:51 epta Is there any haskell lib with darcs bindings?
14:01 owst epta: you mean to "top level" darcs commands?
14:01 owst darcs itself can be used as a library
14:03 donri joined #darcs
14:22 epta Why there is no haddocs? http://hackage.haskell.org/package/darcs-2.8.3
15:08 owst I'm not sure, I don't know what it is about the package that means they're not generated
15:12 favonia joined #darcs
15:35 sm__ joined #darcs
16:11 iago joined #darcs
16:21 * owst wonders if there's any way to tell cabal that it doesn't need to rebuild everything if we just do a `cabal configure -ftest; cabal build`, having done a full configure/build run without -ftest just before
16:52 Heffalump ghc 7.6.1 gives some more clues about why it's actually rebuilding - something to do with "flags changed"
16:52 Heffalump if we can track that down it might be fixable
17:47 sm__ joined #darcs
18:00 owst joined #darcs
18:15 sm__ joined #darcs
19:37 amgarchIn9 joined #darcs
19:48 mizu_no_oto joined #darcs
20:02 mizu_no_oto joined #darcs
20:41 shirt joined #darcs
21:00 favonia joined #darcs
21:11 mizu_no_oto joined #darcs
22:34 owst Heffalump: shall I do a follow up for the pattern match error message, along the lines of "expecting one of"?
22:34 Heffalump can't hurt
22:35 owst I think "expressions over" was just me being pedantic and thinking of equational theories ;-)
22:35 Heffalump I'm a relatively theoretical compsci and it still took me exteranl context to figure out what you meant :-)
22:35 owst heh
22:35 owst I wanted short and snappy, but instead got short and confusing
22:36 Heffalump anyway, I accepted the patch as is because it was clearly an improvement even if left like that, but certainly further improvements are worthwhile
22:36 owst Sure
22:36 * Heffalump has just wasted an hour or so on a Cabal bug :-(
22:36 owst Deep joy
22:36 Heffalump researching it did mean I got to read the commit messages here: https://github.com/haskell/cabal/commits/master​/Cabal/Distribution/Simple/Build/PathsModule.hs
22:38 owst Revert Revert Revert!
22:38 Heffalump anyway, the moral of the story is, stay away from Cabal 1.6.0.2 (perhaps only on Windows)
22:38 Heffalump and if you've got a cabal-install built against it, pray
22:39 owst That's an old version of Cabal?
22:39 owst I'm on 1.16 apparently
22:40 Heffalump sorry, 1.16.0.2 - .1 and .3 are ok though.
22:40 owst Ah
23:05 owst Heffalump: re: your simple rebase refactoring: does rebaseJob do moveRebaseToEnd and checkSuspendedStatus?
23:06 Heffalump I hope so :-)
23:06 owst Oh, nevermind, yes it does
23:06 Heffalump it's a pain not being able to easily see more context when reviewing
23:06 owst Yeah
23:06 Heffalump need to beef up hub to help
23:07 owst I have a simple script that grabs darcs.net and applies the patch, so I can look at the repo and the diff, but it is annoying
23:07 owst Oh, yeah, it really is obvious - right above the function in question! :-)
23:08 mizu_no_oto joined #darcs
23:28 owst Heffalump: in the look nearby patch, when does this ever work " [path </> "darcs" ++ exeSuffix]"
23:28 owst ?
23:40 owst joined #darcs
23:44 shirt if i do "darcs get" is there any way to also copy the _darcs/prefs/defaults file?
23:44 owst shirt: I don't think so

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