Camelia, the Perl 6 bug

IRC log for #darcs, 2013-02-16

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

All times shown according to UTC.

Time Nick Message
01:32 markstos joined #darcs
04:50 preflex_ joined #darcs
06:10 carter joined #darcs
06:43 kowey joined #darcs
07:36 * kowey makes coffee
07:45 Heffalump morning
07:46 kowey how was that first day of sprinting?
07:46 kowey i saw lots of patches/issues fly by!
07:47 kowey I might try to fit some darcs in between laundry/etc, so if you want to point me any direction in particular…
07:48 kowey oh good, my passport arrives on Monday
07:49 kowey (good job I decided not to go, had forgotten the French consulate [well, their outsourced contractor rather] would want to hang on to my passport)
07:49 Heffalump went ok, owst is making progress with the problem of merges in the bridge that have been the bottleneck
07:49 Heffalump I've been cleaning up random bits and pieces
07:50 Heffalump can't think of anything in particular - the issue tracker is as always in need of gardening :-)
07:50 kowey oh, that sounds like the right sort of thing for somebody who doesn't have any darcs loaded in their head
07:50 kowey *ok
07:51 lelit joined #darcs
07:52 kowey darcs-bridge has enabled me to have my darcs and (with some red tape) benefit from nice github integraty things at the same time
07:52 kowey can't wait till this becomes something we can advertise
07:53 Heffalump how are you using it?
07:54 Heffalump might be good to start on some recipes for doing that
07:54 kowey using GH repo as a mirror
07:54 kowey i mean, with the understanding that people might send me pull rq
07:54 kowey which I don't think I currently have a workflow for yet
07:54 kowey but I assume I should not accept the pull req and do via darcs instead
07:54 Heffalump the bridge ought to support that, though I don't know if there are any practical problems right now
07:55 kowey basic layout is foo; foo-branches (foo-branches/foo_bridge/{foo_darcs,foo_git})
07:55 kowey and don't have it formalised yet, but starting to have conventions like
07:55 kowey foo-branches/foo-git $ git remote add bridge ... and git remote add github ...
07:56 Heffalump mght be useful to write down notes somewhere (not too public yet)
07:56 kowey workflow has (darcs) push, (bridge) sync, (git) pull, (git) push
07:56 kowey so, clunky, but workable
07:56 Heffalump we'll also need to work out how to do this in hub
07:56 kowey now if this magically worked on hub
07:56 kowey wow!
07:57 kowey [] mirror on GitHub [repo]
08:00 owst joined #darcs
08:25 xymox joined #darcs
09:03 dixie morning
09:04 gh_ joined #darcs
09:04 gh_ hi from the sprint! (day 2)
09:06 owst joined #darcs
09:11 gh_ I'm looking at dixie's patches http://bugs.darcs.net/patch1031
09:16 dixie welcome any review comment! I found already one - missing haddock comments for new functions.
09:29 gh_ dixie, there seems to be a problem with issue2257-impossible-obliterate-subset.sh
09:32 dixie ok, I'll try to look on that
09:40 gh_ dixie, also, about the reviewing process and screened vs reviewed: http://darcs.net/Development/PatchReview
09:45 Heffalump should we move the main darcs repos to hub.darcs.net?
09:47 Heffalump owst points out that the reviewed branch still has the update roundup posthook
09:50 owst Can't we have a cron job or something on darcs.net that scrapes/inspects hub?
09:50 lambdabot owst: You have 1 new message. '/msg lambdabot @messages' to read it.
09:50 amgarchIn9 joined #darcs
09:51 Heffalump I think in principle we can do something like my roundup-watch script for patches - scrape roundup for non-resolved bugs, and the history of the repo for "resolved issueXXX" patches - though I'm not sure how to deal with reopened ones in that workflow
09:52 owst Does the current roundup script handle reopened issues?
09:54 Heffalump well, it's a one-time thing, isn't it? You push a patch to resolve an issue, it resolves it. If you reopen it, nothing re-resolves it unless you push a new patch to resolve it.
09:58 owst Surely then the scraper should only perform action based on a single patch once?
09:58 owst That way, you can still re-open, but the scraper won't re-resolve until a new patch comes in
10:04 M30W joined #darcs
10:07 Heffalump how does the scraper define "new patch"?
10:09 M30W Is there a way to get only a specfic folder from a darcs repository?
10:09 M30W I.e get --lazy http://domain.tld/path/to/repo/SomeDir/
10:09 owst M30W: do you mean `darcs get` should only retrieve patches/pristine for a subdirectory?
10:09 owst Oh
10:09 owst No, that doesn't work
10:10 M30W I know it doens't. But is there a way to get the dir instead of the whole ?
10:10 M30W I.e repo/{_darcs/...,requestedDirNameHere}
10:11 owst No
10:11 M30W Damn
10:11 owst That's what I mean by "that doesn't work"
10:11 M30W I see
10:11 M30W :(
10:12 Heffalump why do you need it - is the repo really large?
10:12 M30W More ideal for modularly getting pieces instead of the whole repo. But yea, potentially.
10:12 M30W And sub-repos doesn't sound so nice with darcs?
10:12 Heffalump no, they don't really exist/work
10:13 Heffalump the problem with just having a subset of the files is that the history is weird
10:13 M30W Yea..
10:13 Heffalump we do have some long-term ideas for this stuff - it's a problem I'm quite familiar with from my day-job - but nothing imminent
10:14 kowey not that this would help for the darcs get case
10:15 Heffalump would you expect the patches to also be smaller when you just get a subdirectory?
10:15 kowey but you could maybe create a repo with the subset of patches just on the subdir
10:15 Heffalump that's the really tricky aspect
10:15 kowey oh, beat me to it
10:15 Heffalump kowey: yeah, but what if the patches cut across that subdir and others, or move stuff in and out?
10:15 kowey yup
10:15 kowey I figured that it practice that you'd still get a subset
10:15 kowey plus some other stuff
10:15 kowey but still saving some pain
10:16 kowey not sure about the move stuff in/out though
10:16 Heffalump I guess in principle, one could filter the patches to just be those that touch the subdir + dependencies
10:16 Heffalump but that might be quite a big set in practice
10:17 * owst wonders about hashing/addressing of individual prims
10:17 kowey prims v3?
10:17 Heffalump v3 is part of the idea we have for doing something - because you get more control over the mapping of files to places on disk
10:18 gal_bolle joined #darcs
10:18 gal_bolle hi all
10:18 owst hi gal_bolle
10:18 kowey hiya!
10:18 gal_bolle is there a code to enter the irill building? There seems not to be anyone in the lobby
10:19 Heffalump we should come and get you!
10:19 * Heffalump does so
10:19 gal_bolle wait
10:19 gal_bolle I'm at the public library at place d'italie
10:19 gal_bolle I'll only be there in +- 7 minutes
10:19 * kowey returns to laundromat with his stack of 20p coins
10:20 M30W Heffalump: It'd only really make sense to get the dir without the patches
10:21 Heffalump M30W: yeah, that sounds more plausible, but then what use would the repo be? Just for new patches?
10:21 M30W Lets say.. Dotfiles, You're only interested in the xmonad config and don't want the vim configs.
10:22 M30W (And other configs)
10:22 Heffalump I see the general use for subrepos/subsetting repos etc - but I'm just wondering if the cut down repo without any historic patches is actually useful for that case
10:23 M30W Need something like git's modules
10:24 gpiero1 joined #darcs
10:25 M30W _darcs/modules /path/ url
10:25 M30W ?
10:27 Heffalump M30W: yeah, maybe. git modules don't seem universally liked, so we'd like to do something better if we can.
10:28 M30W Yea
10:28 M30W I hate them too
10:28 M30W :P
10:28 M30W darcs ls-modules; darcs get-modules [modname] [...]; darcs add-module path http://url
10:29 M30W Screw versioning cause that shouldn't matter which version is current in the main repo ?
10:29 Heffalump the (very) rough idea we have is just to be able to integrate a darcs repo into a bigger one, by remapping all its files to live in a sub-directory.
10:30 M30W More issues there if there are branches
10:30 Heffalump but the details will need thinking through
10:30 M30W Wait.. does darcs have branches?
10:30 Heffalump not inside a single repo directory
10:30 Heffalump you just make another repo
10:30 M30W Yea
10:30 M30W Just a fork :P
10:30 Heffalump this keeps things simple, but annoys people who like in-repo branching - there are legitimate reasons for wanting that, but we're not sure it's worth the complexity
10:31 M30W Yea dev branch for one
10:31 M30W But merging branches is pain
10:31 M30W :|
10:32 Heffalump the two strongest arguments i've heard are (1) keeping build products etc in one working dir is much mroe convenenient and (2) having a canonical remote location for getting each of multiple branches is useful
10:32 M30W Anoying issue :(
10:38 gal_bolle joined #darcs
11:20 lelit joined #darcs
11:39 markstos joined #darcs
11:39 M30W wow.. so chatty here.
11:39 Heffalump we're talking in person :-)
11:39 * M30W goes afk and the conversation died.
11:39 M30W oh damn!
11:40 M30W that sounds almost as effective as irc!
11:47 * markstos starts working on getting his environment set up to run darcs test suite.
11:49 M30W how does darcs test work exactly? just a command and/or script to run and going by the exit code for success vs fail? can there be information for those tests i.e with quickchck and each function?
11:49 * markstos checks "cabal install --help" in hopes of findind a "-j" flag for parallel buliding. Comes up short."
11:50 M30W " - mismatched
11:50 Heffalump markstos: I think it's been done for cabal quite recently, not sure if it's in a release yet
11:50 gal_bolle markstos: just find a good reason to compile two darcs at once, and you're done
11:50 Heffalump M30W: just exit code, yes
11:50 * gal_bolle ducks
11:50 markstos M30W: I understand there are both shell tests, in bash, and unit tests in Haskell.
11:50 Heffalump M30W: what sort of thing would you like it to do?
11:50 Heffalump markstos: I think M30W is asking about 'darcs test' rather than the darcs tests, if you see what I mean.
11:50 markstos Heffalump: Yep.
11:51 Heffalump markstos: re issue2282 (pulling from a rebase-in-progress repo) - how much do you want to be able to do that?
11:52 markstos Heffalump: rarely.
11:52 Heffalump i.e. would it be ok for it to just refuse with a better error message
11:52 markstos We had a developer change and haven't needed Rebase so munch lately.
11:52 markstos Heffalump: Yes.
11:53 markstos I ran "darcs test" it seems to be compiling darcs again. :) I expected it to run a test.
11:53 gal_bolle it needs to compile a pristine darcs in order to test it
11:53 Heffalump markstos: gh_ says it's in the latest cabal, so cabal install cabal-install ought to get you a cabal install you can use -j with
11:54 Heffalump markstos: 'cabal test' is the right way to test a darcs you have just built locally
11:54 markstos Oh, sweet.
11:54 Heffalump but you need to have used 'cabal configure -ftest' or 'cabal install -ftest' or it won't work.
11:55 markstos I did use "-ftest" already, so far, so good there. What's the difference between "darcs test" and "cabal test"?
11:55 gh_ darcs-test or darcs test ?
11:55 markstos My attempt to run "darcs test" failed with: "       (imported from Distribution.Simple, but defined in Distribution.Simple.UserHooks):
11:55 gh_ "darcs test" is the new name of trackdown
11:56 gh_ "darcs test --linear" " darcs test --bisect" "darcs test --backoff" ...
11:56 gal_bolle darcs test: Run test on the current recorded state of the repository.
11:56 markstos Oh.  then I'll try darcs-test.
11:56 M30W Heffalump: yes darcs test. it support haskell and quickcheck?
11:57 markstos Let's see what darcs-test does then.  That explains why "darcs test" recompiled things. :)
11:57 markstos darcs-test: darcs-test: canonicalizePath: does not exist (No such file or directory)
11:58 M30W darcs test --set 'cabal configure -fdemos; cabal build'
11:58 M30W is that valid?
11:58 owst markstos: where is darcs-test? There's some code that attempts to find a darcs binary "nearby" so that is probably what's failing, for whatever reason
11:59 * M30W can't test ideally right now due to fact on n900 :P
12:02 markstos owst, I tried ./dist/build/darcs-test/darcs-test, and that seems to be running.
12:02 M30W i miss understant darcs test?
12:03 owst markstos: yeah, that's right
12:03 dixie hmm, I'm trying to update help for enviroment variables which controls some of the behavior of darcs. Does it make a sense to put all literals related to environment to some sensible module? Maybe even applicable on all string literars in darcs, like Darcs.Token.Env, Darcs.Token.Metadata etc.
12:03 owst M30W: I'm not sure.
12:03 owst M30W: What do you mean?
12:05 M30W brb sorry stomach really hurting...
12:06 markstos does darcs-test -j run the shell tests in parallel, or just the unit tests?
12:07 Heffalump the shell tests - not sure if it actually does run the unit tests in parallel
12:10 markstos What's the invocation for running a single shell test? I tried just using "bash" and I got prompted "What is your email address?"
12:10 Heffalump cabal test <testname>
12:10 Heffalump or dist/build/darcs-test/darcs-test -t <testname>
12:11 kowey I think you can run dist/build/darcs-test/darcs-test --help for more features
12:11 kowey like using a fixed directory to dump its output to so you can sift through it later
12:13 markstos I'll my tests pass with "darcs-test". That's good.
12:14 markstos cabal test test.sh worked.
12:14 Heffalump cool
12:14 markstos However, "cabal test tests/issue1473.sh " appeared to work, too just reporting that "0 passed" and "0 failed".
12:18 markstos My luck with darcs-test -t wasn't so good. It keeps telling me it can't figure out the path, but doesn't tell what path it's attempting:
12:18 markstos darcs-test: darcs-test: canonicalizePath: does not exist (No such file or directory)
12:18 markstos Wait, I'm probably using the wrong darcs-test again.
12:19 markstos That worked: ./dist/build/darcs-test/darcs-test -t issue1473.sh
12:24 * markstos digs around for the test he promised to write. Welcomes reminders.
12:31 * markstos works on testing http://bugs.darcs.net/issue2271
12:32 M30W back
12:32 M30W owst: what is the inteded usage for darcs test?
12:33 markstos M30W: figuring out which patch broke something.
12:33 M30W broke something, tested how exactly?
12:33 markstos M30W: by testing the repo by peeling off patches and running some command until the command passes (or fails)
12:34 markstos M30W: The Manual has more detail: http://darcs.net/manual/Darcs_comman​ds.html#SECTION006104000000000000000
12:34 M30W nice section id. O.o
12:34 M30W thanks
12:36 M30W ahh ima rip out my stomach soon... why pain? D: brb
12:43 kowey yeah, I hate those section ids
12:43 kowey long term hope to convert the docs to something else
12:43 kowey and not use latex2html
12:49 M30W that is latex section id?
12:49 M30W markdown is far easier to... well and processing of. :P
12:49 kowey generated by latex2html
12:49 kowey yeah, slowly working on the conversion
12:49 kowey there's some autogenerated stuff
12:50 kowey sharing the darcs help with the manual
12:50 gpiero1 dixie: while looking at it, could you also figure a place where putting the env vars *set* by darcs?
12:50 kowey that needs to be brought over
12:50 kowey I forget where we are in the process, but converging to markdown (wiki is in md) is probably good
12:50 M30W pandoc -f tex -t markdown -o page.markdown page.tex
12:50 kowey if it were only so simple :-)
12:50 M30W pandoc <3
12:50 kowey it's definitely part of the process
12:51 kowey just a matter of some work really and gh_ has done a good chunk of it
12:52 M30W gh?
12:52 kowey guillaume
12:53 M30W whatever that is....
12:53 kowey guillaume hoffmann, a member of the darcs team
12:54 M30W ah
12:54 M30W okay
12:54 M30W whos idea was it to use bloody tex anyhow??
12:55 kowey david roundy's and it was probably a good idea at the time (2002)
12:55 M30W heh
12:55 kowey he had a nice idea of using literate programming for the manual
12:55 kowey but that proved too unweildy
12:55 kowey so we had to divorce the code and docs
12:55 M30W lhs <3
12:57 M30W lhs and really good haddoc is damn useful.
12:57 M30W oh that reminds me.. does darcs have an importable api?
12:57 M30W i'd love that.
12:57 kowey we expose our modules as a library; what we don't have is stability
12:58 kowey in other words, we're doing lots of cleanup work
12:58 kowey so things getting moved around quite a lot
12:58 kowey and I think we're getting a Darcs monad one day
12:58 markstos I need a little shell test help. I'm trying to test for an error message:
12:59 markstos darcs optimize --disable-patch-index 2>&1 | grep 'Could not delete patch index'
12:59 markstos But it still reports my test as a "failure" after this, perhaps because 'darcs optimize" failed here.
13:00 markstos Oh, looks like I need "not" in front of "darcs". Trying that.
13:02 M30W good morning world
13:02 M30W kowey: that sounds interesting.
13:02 M30W so darcs it's self is by far still in heavy development.
13:02 kowey yes
13:03 kowey it's old and young at the same time
13:03 M30W once the library is all stable and compiles with newest ghc i'm gonna be very happy. :)
13:04 M30W i have to download from darcs's website cause i can't compile it on stabl ghc. :(
13:04 M30W without patching etc
13:04 kowey think of it as as piece of software that implemented a really  cool idea, and is now being slowly transformed into something that can be worked on by a distributed team without one guy that knows *everything* about it
13:04 kowey so this transformation is part of darcs' youth
13:05 kowey we could use loads of help
13:06 kowey for example, teasing apart the documentation stuff would be one less thorn
13:07 markstos Here's new test for issue2271 for peer review: http://hub.darcs.net/markstos/darcs-​screened/patch/20130216130307-8a0ef
13:07 markstos It passes for me, I just want to make sure the test has been constructed properly. I haven't built one in a while.
13:08 M30W kowey: i'd love to help, i really would but i am not so advanced with haskell yet. note my http://hub.darcs.net/M30W/ :P
13:08 kowey markstos: is the intention to test that darcs optimize --disable-patch-index should not attempt to use the pi?
13:08 kowey (and the permissions just being a vehicle to see if it is trying to use it?)
13:09 markstos kowey: I think there used to be a bug where --disable-patch-index would claim success, even if it actually failed, due to a permissions issue.
13:09 kowey well, M30W if you're feeling ready for a ProbablyEasy task, give us a shout
13:09 kowey some of them are really quite entry level
13:09 Heffalump gh_: http://darcs.net/releases/branch-2.8/
13:10 M30W kowey: examples of things i can do from N900 after midnight?
13:11 kowey hmm, I think this might just be a matter of looking something up in the libraries http://bugs.darcs.net/issue2020
13:12 markstos I'm going to have breakfast with my family. I might be able to check-in later.
13:12 kowey this could be digging through Darcs.UI.Arguments, I think http://bugs.darcs.net/issue1789
13:12 kowey this seems like a one liner http://bugs.darcs.net/issue1549
13:12 markstos I can read the IRC logs online to catch-up.
13:13 M30W define bonus points? O.o
13:13 kowey this too looks easy (although involves searching for examples of how darcs does temp dirs, maybe http://bugs.darcs.net/issue1416)
13:14 kowey hah, just speaking informally, M30W that reviewers would probably be happiest if…
13:14 M30W xD
13:14 M30W i read as bounty at first but saw no $$ tags. :P
13:16 dixie kowey: what is your opinion on http://bugs.darcs.net/issue1579?
13:16 kowey glancing at that, dixie, I think it could be worth looking at the plans for cleaning up all the flags
13:16 kowey but I don't mean to turn one issue into a huge scary one
13:17 kowey in a way, I'm inclined to say that if we're going to change how the opts work
13:17 kowey we might as well do it right
13:17 kowey (the big overhaul; it's in the tracker somewhere)
13:17 dixie kowey: do you mean some existing plans?
13:17 kowey and anything less than that would just add to confusion
13:18 kowey dixie: yeah, well a design that mornfall sketched out
13:19 kowey dixie: I'm relying on old memories, as have been out of the loop for a long time
13:19 M30W-N900 slab'o'text
13:19 M30W-N900 D:
13:20 M30W-N900 ty maemo... what's that now? 27 windows opened on @4yr old phone... true multitasking <3
13:21 kowey dixie: http://bugs.darcs.net/issue1457
13:21 kowey this may need something like a http://darcs.net/Ideas/Flags
13:21 Heffalump markstos: I think you offered a shell test for issue2272, btw
13:22 kowey dixie: sorry to throw that thing at you; I don't mean to distract
13:23 M30W-N900 so many bugs!!!
13:23 M30W-N900 kowey: link is dead
13:24 dixie kowey: no problem :) It is good that there is a good discussion on that topic.
13:27 * M30W-N900 likes --no-foo
13:31 M30W-N900 kowey: how about i contribute to darcs by pulling in people and advertising the vcs. sounds good. :D
13:32 M30W-N900 (already doing it :P)
13:33 M30W-N900 can darcs push to more than one host?
13:34 gh_ simultaneously ? no
13:35 M30W-N900 didn't think so.
13:36 gh_ M30W-N900, in the code that implements the darcs commands like push, pull, etc., you can see that there is only one remote repostory.
13:36 M30W-N900 gh_: that could be map'd over no?
13:37 gh_ yeah but.. then what happens with interactive patch selection ? I guess you should go through the SelectChanges interface once per remote repository
13:37 M30W-N900 pull only needs one remote tho
13:37 gh_ it raises certain issues
13:38 M30W-N900 unless you feel like say torrenting your darcs xD
13:38 M30W-N900 gh_: yea or an option or configuration setting to determine how that goes.
13:38 M30W-N900 once vs everytime
13:39 Heffalump how about: for repo in ... ; do darcs push -a $repo ; done
13:39 Heffalump but interactive selection is indeed tricky
13:40 M30W-N900 meh shell side
13:40 Heffalump I don't think it would save much work to do it inside darcs
13:40 Heffalump for each remote repo, it has to compute the patches that aren't already there, and send them across and execute darcs apply.
13:41 M30W-N900 set default multiple. darcs push. yay. okay... there any aliasing config ability?
13:41 Heffalump So at best it would really save a bit of parsing and a bit of your typing (and familiarity with shell loops)
13:41 Heffalump funny you should mention that :-)
13:41 M30W-N900 i.e set alias host1 to push to that host
13:41 Heffalump http://bugs.darcs.net/patch948
13:42 Heffalump so not yet, but likely soon
13:42 M30W-N900 that post needs moar newlines -_- scroll sideways!!!!!
13:43 M30W-N900 awesome. :D
13:43 Heffalump :-0
13:45 M30W-N900 so would that be like for example 'darcs hub' == 'darcs push http://hub.darcs.net/M30W/%r' ?
13:46 kowey M30W-N900: ha, well we're kind of laying low these days, has to be a word of mouth cult following kind of thing, people who get there's something there deep down underneath all the incompleteness
13:46 M30W-N900 or actural litteral and no variable cause that's kinda useless unless you have it in your darcs template dir.
13:47 M30W-N900 kowey: it works perfectly for me. no bugs detected. just wishlist. :P
13:47 Heffalump M30W-N900: I think it would be darcs push hub
13:47 Heffalump we want 'darcs hub' to pop up a webbrowser pointing to a local copy of the hub :-)
13:47 M30W-N900 hence this happened... https://github.com/archeydevil/dotfiles
13:47 M30W-N900 xD
13:47 Heffalump [well, maybe: we certainly want local itnegration with the hub, but not sure about how it would work precisely]
13:48 M30W-N900 Heffalump: ahh nice
13:48 M30W-N900 so host aliasing
13:48 M30W-N900 like in ssh/config. hehe i 'ssh germany' :P
13:49 M30W-N900 except this manages path etc.
13:49 M30W-N900 sweet.
13:49 Heffalump if you can be bothered, try the patches and see how it goes
13:50 M30W-N900 i'm on my N900 right now. and with ghc 1.7.... so i can't even compile darcs
13:50 M30W-N900 The Glorious Glasgow Haskell Compilation System, version 7.6.2
13:50 M30W-N900 yea that version :P
13:51 Heffalump :-)
13:51 Heffalump you should be able to compile darcs with that version, as of recently
13:51 M30W-N900 oh yay
13:51 Heffalump both the 2.8 branch and screened should support GHC 7.6.*
13:51 M30W-N900 cabal updating
13:51 M30W-N900 yay
13:52 M30W-N900 i am still on my phone so yea.... it can wait :P
13:53 M30W-N900 it compiles!!! :D
13:53 kmels joined #darcs
13:54 M30W-N900 compiling English? O.o
13:54 kowey :-)
13:55 kowey it helps us with plurals etc
13:55 kowey could be nice to replace some of that with the minimorph package
13:55 M30W-N900 plurals?
13:55 kowey patch -> patches
13:55 kowey 1 patch, 0 patches, 2 patches
13:55 M30W-N900 so... parsec with autocorrections not good enough?
13:55 kowey 1 user, 2 users,
13:55 M30W-N900 ohh
13:56 M30W-N900 sweet
13:56 M30W-N900 i read it as user input
13:56 kowey it's imprecise, just something we threw together
13:56 kowey this patch => these patches
13:56 kowey that sort of thing
13:56 M30W-N900 i might look in to it. :D
13:56 M30W-N900 tomorrow maybe
13:57 M30W-N900 could improve my blog (written in hakyll) http://jamesm.com.au/ :D
13:57 M30W-N900 i.e n posts for month M
13:58 kowey heh, well it's pretty minimalistic (the module); nice setup you have
13:58 M30W-N900 kowey: you darcs devs make english?
13:59 kowey we cabal install it perhaps
13:59 M30W-N900 i need to update the hub darcs online repo version.. ima fix that up in a min
13:59 M30W-N900 what? hakyll?
13:59 M30W-N900 it's awesome. :D
13:59 kowey english
13:59 kowey nevermind, just being silly
14:00 kowey I like hakyll
14:00 M30W-N900 :D
14:00 kowey need to convert my old blogspot and wordpress stuff over
14:01 M30W-N900 heh. i migrated my 27 line php blog cms to hakyll and redesigned the whole thing from scratch in only 3 minutes. :D then a little rage getting my head around some hakyll things.
14:01 M30W-N900 templating was piece of cake. :)
14:04 kowey on twitter, git user trying darcs observed that easy config stuff would be nice, eg. git config color.ui = auto
14:04 kowey (how do we do it in darcs? I forget the set of env vars we need, know we talked about this on list)
14:05 kowey always something to learn from others
14:05 M30W-N900 darcs knows colour?
14:05 kowey oh ugh, sorry, preachiness on my part, bad habit
14:05 gal_bolle it's ok
14:05 kowey http://lists.osuosl.org/pipermail/d​arcs-users/2013-January/026754.html
14:19 kowey darn, darcs-test doesn't offer a way to get at the default test-framework args
14:19 kowey particularly the one that lets you set the seed
14:27 kowey gpiero1: interested in sending issue2275 as a standard darcs test case?
14:28 Heffalump kowey: judging whether issues should be for 2.10 or not would be useful too, btw
14:28 kowey gpiero1: http://darcs.net/Development/RegressionTests (I'm skimming bugs, but superficially it seems like sometimes-follow-links-sometimes-not is no good)
14:28 kowey what's 2.10? rebase + pi?
14:29 Heffalump yeah
14:29 Heffalump and last regrets
14:29 kowey oh, last regrets :-)
14:29 * kowey thinks he's gotten used to it by now
14:29 kowey or maybe just says “yes” without thinking
14:29 Heffalump we need to discuss if we have any regrets about last regrets
14:30 kowey ok, may not feel confident about making that kind of judgement call, but will see what i can percolate up
14:30 kowey am just trying to reduce amount of unknown status bugs
14:30 Heffalump I think it's fine to be conservative and mark lots of stuff for 2.10, for a future pass to weed out
14:30 kowey mmm, yes
14:33 kowey we doing optimize --http for 2.10?
14:36 Heffalump good question
14:36 Heffalump any volunteers to investigate/fix it?
14:37 iago joined #darcs
14:37 M30W-N900 optimize http?
14:38 kowey so part of the reason people (rightly) think darcs is slow, M30W-N900 is that when they first encounter darcs (via darcs get some repo)
14:38 kowey it takes a while to fetch the repo: it's downloading lots of individual patch files, etc
14:39 kowey optimize --http makes a big tarball that they can download in one go
14:39 M30W-N900 sweet
14:39 M30W-N900 sounds useful for lots
14:39 kowey we, err, keep having trouble hammering out the corner cases
14:39 M30W-N900 and if they --lazy?
14:39 kowey this has been around since 2010? frustrating
14:39 kowey we have two tarballs, actually, one for pristine and one for patches
14:40 kowey I don't remember what happens for lazy, but I think in either case you can fetch the pristine tarball
14:40 kowey maybe you skip the patch one if lazy
14:40 kowey *finishing* jobs is really hard :-/
14:41 M30W-N900 okay
14:41 kowey like I say, we need lots and lots of help :-)
14:42 Heffalump this one was apparently finished, but had a last minute bug :-(
14:42 M30W-N900 haha yea. <3
14:43 iago kowey, I hardly can minimize issue2046
14:43 dixie gh_:
14:43 gpiero1 kowey: re test for issue2275, currently busy but will look at it later or tomorrow
14:43 dixie oops
14:44 kowey thanks, gpiero1; iago, I think giving up is the best thing we can do anyway, cut our darcs-1 losses, eh?
14:44 iago kowey, but in this case, SmartCheck may be useful ;-)
14:44 kowey yes!
14:45 kowey current attitude to bugs is that we need to be more honest with ourselves about what we can realistically get to
14:45 kowey and if heat death of universe is likely timeframe, well…
14:47 owst iago: how much work is it to turn "use" SmallCheck on this bug?
14:47 owst s/turn//
14:47 iago owst, Sm*art*Check
14:47 kowey https://github.com/leepike/SmartCheck
14:48 owst We also need to *understand* the bug, which is the harder part, even given a minimal testcase.
14:48 owst iago: hah! Yes
14:48 iago SmartCheck is basically QuickCheck + "advanced" shrinking
14:48 owst iago: same question for SmartCheck :-p
14:48 iago if we have a Read instance it should be pretty straightforward
14:49 kowey if the Haskell community had lots of prominent Czechs it'd make for a great photo meme thing
14:49 iago (I don't know why, but SmartCheck reads counterexamples in that way :-P)
14:49 kowey quick, small, smart, lazy…
14:50 kowey or we could take pictures of somebody in various poses
14:52 iago damn it, I hate those "small" functions with HUGE (even nested) where clauses
14:52 kowey what are you looking at, iago?
14:53 markstos joined #darcs
14:53 Heffalump welcome back
14:54 markstos Thanks, Heffalump. if you missed the post earlier, the test for RT#2271 is ready now: http://hub.darcs.net/markstos/darcs-​screened/patch/20130216130307-8a0ef
14:55 Heffalump cool
14:55 iago kowey, SmartCheck code
14:55 Heffalump 2272 next? :-)
14:55 kowey ah
14:57 markstos :) I'm multitasking with my family today. We'll see how the day goes.
14:57 markstos There's a Trello board!
14:57 gh_ ah yes
14:57 kowey there is!
14:57 kowey dunno if anybody else wants to use it
14:57 Heffalump oh, your test case shows me how to both check for failure and for a specific error message. Thanks! (I just wrote a test that does it much more laboriously)
14:57 kowey thought it could be good for global view
14:57 gh_ kowey, I wondered if there were some things that were not very up to date on it
14:58 gh_ like the darcs send thing
14:58 markstos Heffalump: I grep'ed the test suite for prior art.
14:58 kowey heh, by now at least everything I was working on should probably go back to some other pile
14:58 Heffalump I couldn't think what to grep for, or I'd have done the same
14:58 markstos Is there is a section or page on the wiki about how to contribute to the test suite? I don't see one linked from here: http://darcs.net/Development  I'm thinking of creating one.
14:58 Heffalump I don't think there is.
14:58 kowey http://darcs.net/Development/RegressionTests
14:59 Heffalump ...but then I don't know much about darcs :-)
14:59 gh_ markstos, there's also ./tests/README.test_maintainers.txt
14:59 gh_ it may or may not be up to date
14:59 gh_ last patch that touched it: 2010
14:59 markstos Thanks kowey and gh_, I'll review both and see how they might be be improved.
15:00 kowey I'm for killing duplication, eg. by making the latter point to the wiki
15:00 gh_ yeah
15:00 markstos If you just want one, I would vote for the wiki as well.
15:00 gh_ I think this file dates from our "pre wiki-as-website" era
15:01 markstos Since the wiki is also a repo of text files at it's core, it could even be formatted to be shipped for reading offline as well.
15:01 Heffalump you can darcs get it
15:01 markstos gh_: I think I created the file around darcs 1.0...
15:02 markstos On a Sunday in 2008, it appears.
15:02 Heffalump that's positively modern
15:03 kowey yup darcs get --lazy http://darcs.net/darcs-wiki
15:03 iago unfortunately, it *seems* that there is no Read for V1 patches
15:04 Heffalump is there a Show?
15:04 iago yep
15:05 iago ok wait, my bad, there is a ReadPatch instance
15:11 iago ok I'm confused, the bug is in the _ -> alternative of the case expression of http://hackage.haskell.org/packages/arc​hive/darcs/2.8.4/doc/html/src/Darcs-Pat​ch-V1-Commute.html#reconcileUnwindings
15:12 iago (_:_) -> OR [] -> OR _ -> bug
15:12 iago ?
15:13 markstos So after I grab a copy of the wiki and make an edit is there an address to "darcs push" to, or do I "darcs send" ?
15:13 owst iago: hmm, that's strange
15:14 Heffalump iago: it doesn't match NilRL lists in the (_:_) case
15:15 Heffalump markstos: darcs send, as I don't think you have push access  to the wiki - though it currently goes to darcs-users; kowey, should we change that/
15:16 markstos I wouldn't want to bother a mailing list with small doc edits. I could just use web UI instead.
15:16 iago Heffalump, but the producers are  of the form (_:<:_) <- ...
15:16 gh_ markstos, yes, for this the web UI works well
15:18 Heffalump markstos: we could just give you push access. But agreed, the mailing list is the wrong target.
15:18 markstos Is that a matter of providing public SSH key(s)?
15:19 Heffalump yep
15:20 owst iago: are you sure that it is the bug case not bugDoc that's actually hit?
15:20 Heffalump oh, you actually have an account on darcs.net, do you still have access to it?
15:21 * markstos checks
15:21 kowey markstos: yeah, I can definitely add a key for you
15:22 Heffalump kowey: in theory putting markstos in wiki-author group would be the right thing
15:22 iago owst, do'h :-)
15:22 Heffalump but we haven't tested the permissions etc if we get people doing that
15:22 owst iago: Heffalump spotted it ;-)
15:22 markstos Yes, turns out I can still log into my darcs.net account.
15:22 kowey now's our chance
15:22 * kowey still doesn't trust this author/group stuff
15:23 Heffalump ok; so kowey puts markstos in wiki-author, then markstos tries push via ssh to /home/wiki-author/wikidata
15:24 kowey (done)
15:24 Heffalump oh, I could have done that, I forgot I have sudo. Anyway, let's see what happens :-)
15:24 iago owst, ok I guess it should be a easy task to try to shrink the patch using SmartCheck, even though I'm not sure if it's gonna work well
15:26 iago kowey, I was taking a look to https://github.com/leepike/SmartChec​k/blob/master/src/Test/SmartCheck.hs
15:26 markstos OK,  I had a successful wiki push via SSH. Thanks team.
15:26 drostie joined #darcs
15:27 kowey iago: as a generally useful thing for darcs and its generated patch properties tests?
15:27 Heffalump markstos: yay
15:27 kowey or more of a hey-i-wonder-how-its-smart-shrinking works?
15:28 kowey *-works
15:28 Heffalump markstos: and I think the group permissions are still ok
15:28 Heffalump https://twitter.com/lolaisbeauty/​status/302778198851203072/photo/1
15:28 kowey heh
15:30 markstos I saw this, I presume it's a post hook output: Removing cache entry: /var/lib/gitit/cache/./Devel​opment/RegressionTests.page
15:30 Heffalump yes, I think so
15:30 kowey yeah, the cache is a good thing because of robots
15:30 kowey but we need(ed) a way to notify gitit to expire the thing
15:30 kowey maybe obsolete nowadays, dunno
15:31 iago kowey, the former case, I was taking a look to the interface it provides; even though I guess that taking some time to improve patch shrinking could be more productive
15:31 markstos I'm going to take another shot at having breakfast. Back after a while.
15:33 gh_ Heffalump, so what do with http://bugs.darcs.net/patch993 ?
15:33 gh_ accept as it is? (I'm ok with that)
15:40 Heffalump I sent a followup and declared both accepted
15:44 Heffalump kowey: I'm going to un-duplicate 2276/2269, if that's ok: I think the bug will still be there once we move the prompt from 'unsuspend' to 'suspend'
15:45 kowey oh, ok
15:48 * Heffalump is currently trying to convince the typechcker to let him fix issue2227
16:01 dixie aaa
16:01 dixie ops, wrong window
16:02 lelit joined #darcs
16:51 * sm checks/cleans up darcs hub a bit
16:52 Heffalump what's the status of the lock issue?
16:55 sm unfixed, http://hub.darcs.net/simon/darcsden/issue/46
16:56 sm undiagnosed too, as in is the real bug in darcsden or darcs
17:28 stepkut joined #darcs
18:01 * sm afks
18:06 markstos joined #darcs
18:17 markstos Should I resubmit my 2271 test patch the patch tracker? I guess there's not a "merge pull request" work flow for the Hub yet. http://hub.darcs.net/markstos/darcs-​screened/patch/20130216130307-8a0ef
18:18 Heffalump up to you, if you just keep an eye that we don't miss stuff, that's probably enough
18:19 Heffalump do you know if there's a way of grabbing a .dpatch given that URL?
18:20 markstos I think I may be better off using the official workflow. The Hub doesn't allow commenting, so I'm not sure of the status of the patches I put there back in December: http://hub.darcs.net/darcs/darcs-screened/patches The patch tracker allows commenting, which is an important feature.
18:21 Heffalump ok.
18:21 Heffalump yeah, I just noticed those other patches, and I remember there being some discussion of them btu not the outcome
18:21 * gh_ finds out about darcshive...
18:25 gh_ http://lists.osuosl.org/pipermail/da​rcs-devel/2004-November/000683.html
18:25 gh_ in the sprint room, we are reading commit messages in the darcs repo of exactly 10 years ago
18:26 gh_ looks like we missed the opportunity to celebrate the 10th anniversary of the first patch ever of the repository
18:28 gh_ original announcement of darcs: http://www.haskell.org/pipermail/h​askell-cafe/2003-April/004139.html
18:33 gh_ end of sprint day 2
18:33 gal_bolle see you tomorrow
19:29 markstos When I asked about running tests earlier, I was told to use "darcs-test" and "cabal test". However, tests/README.test_maintainers.txt documents a third syntax:  "runghc Setup.lhs test tests"  Are they really 3 advised still, or is this document out of date?
19:36 markstos "darcs-test" looks most useful to me, because it contains the most useful set of option flags. I'm going to update docs to recommend it.
19:38 markstos When I run darcs-test --help, it spits out "Usage: darcstest [OPTIONS]". Where can this string be updated to say "darcs-test" in the Usage line?
19:40 kowey cabal test replaces the runghc Setup.lhs one I think
19:41 kowey it used to be that cabal-install did not have a test command
19:41 markstos Thanks, kowey, but the output of darcs-test --help looks a good deal more useful than that of  cabal test --help
19:42 kowey I think the latter is generic
19:42 markstos That was my assessment.
19:42 kowey and the former exposes the full set of darcs-specific options
19:42 kowey so cabal test --help is the same you'd get for all haskell packages that have a test suite
19:42 kowey we need to change the cabal file so that our non-standard -ftest flag is just --enable-test
19:42 * markstos nods
19:43 kowey details, details
19:44 markstos I recently learned that it's OK if the hard drive falls out of my laptop, as long as the laptop is suspended, and the drive is returned before unsuspend. It even smartly re-prompts me for the HD password. I should really get the broken HD door replaced.
19:45 kowey still kind of alarming!
19:46 markstos The worst is sitting down to work, only to realize the HD never made it out of my shoulder bag.
20:19 markstos I want to add some Help text here, but I'm not sure of the syntax:
20:19 markstos threads %> Default (1 :: Int) %+ short 'j' ]
20:19 markstos I'm not familiar with the %> and %+.
20:20 kowey ooh, yeah I think that comes from mornfall's cmdlib
20:20 kowey err, think you can make sense of http://hackage.haskell.org/packages/archive/cm​dlib/0.3.5/doc/html/System-Console-CmdLib.html ?
20:20 markstos There are examples nearby which also set "Default" and "short", none that do both like "threads" does.
20:21 * markstos reads
20:22 kowey it may be more realistic to go with the majority here and just use cmdargs (a more commonly (?) used lib)
20:22 kowey cmdlib was written to address some shortcomings with cmdargs, if I recall correctly
20:22 markstos I think I can guess from this line how it might work:
20:22 markstos enable <% option1 +% option2 +% option3 %% disable <% option4
20:23 markstos Although there, the "%+" and "=%" and "<%" and "%>" are inverted..
20:32 markstos I figured it out. Apparently " %+" works like a feature in Perl known as the comma.
20:34 markstos kowey: thanks for the doc link. That was helpful.
20:35 donri joined #darcs
20:38 mizu_no_oto joined #darcs
20:41 markstos joined #darcs
20:54 markstos It would nice if after building, the build process did this on platforms that support symlinks: ln -s $PWD/dist/build/darcs-test/darcs-test ; ln -s $PWD/dist/build/darcs/darcs .
20:54 markstos for repeated testing, ./dist/build/darcs-test/darcs-test gets tiring to type.
21:17 donri markstos: why not "make test" or "cabal test"
21:18 donri also you can add dist/build/darcs-test to $PATH
21:19 donri although relative paths in $PATH is a potential security risk. not big but still :p
21:23 mizu_no_oto joined #darcs
21:29 markstos joined #darcs
21:31 markstos The Trello board mentions that GHC 7.6 support is in progress, but I thought I read earlier here that it is done: https://trello.com/board/dar​cs/50018818943578851e95b534
21:33 markstos I have completed merging the test suite docs on the wiki and at tests/README.test_maintainers.txt. The new combined doc is here: http://darcs.net/Development/RegressionTests  The other patches I created today are here: http://hub.darcs.net/darcs/darcs-screened/patches
21:33 markstos Enjoy the sprint, I'll catch you later.
22:43 gh_ joined #darcs
22:55 owst joined #darcs
23:43 mizu_no_oto joined #darcs

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