Camelia, the Perl 6 bug

IRC log for #darcs, 2009-11-15

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

All times shown according to UTC.

Time Nick Message
00:12 * lispy|web is waiting for HPC to finish
02:02 kpreid_ joined #darcs
02:39 nwf joined #darcs
03:05 Jerri_ joined #darcs
03:34 lispy|web joined #darcs
04:02 nwf joined #darcs
04:07 tommd joined #darcs
06:16 kowey joined #darcs
06:16 kowey morning
06:16 lambdabot kowey: You have 1 new message. '/msg lambdabot @messages' to read it.
06:19 kowey_ joined #darcs
06:20 lispy|web kowey: good morning
06:20 lispy|web kowey: I can no longer build with -ftest
06:20 lispy|web kowey: I removed commutex from the normal code, but it's still used in the unit test target, which I can't build for other reasons.
06:21 lispy|web kowey: Kind of an awkward position.
06:21 lispy|web I'm getting that iconv undefined symbols error
06:21 kowey http://web.mornfall.net/stuff​/photo/darcs-sprint-3-gp.jpg
06:21 lispy|web and cabal is warning about multiple versions of the same library
06:22 kowey hmm
06:22 lispy|web kowey: nice
06:22 lispy|web package regex-compat-0.71.0.1 requires regex-base-0.72.0.2 package regex-posix-0.72.0.3 requires regex-base-0.93.1
06:22 kowey lispy|web: I thought there was a simple way to work around the iconv undefined symbols error?
06:22 lispy|web Note: our cabal file only mentions regex-compat
06:22 kowey upgrading some package or another?
06:23 kowey or on the Mac, something about installing iconv via MacPorts?
06:24 kowey lispy|web: http://old.nabble.com/Haskeline,-pcre-ligh​t,-iconv-and-Cabal-on-OSX-td21778193.html
06:26 kowey lispy|web: apologies to the Portland sprinters who got their patches stuck in the anti-spam moderation queue
06:28 lispy|web kowey: np, we figured it out, but I don't know the current pasword so I couldn't clear up the queue
06:28 kowey oh, I never changed it
06:29 lispy|web hmm
06:29 lispy|web weird :)
06:29 kowey did Thomas make it to the sprint?
06:29 lispy|web Not yet
06:29 lispy|web But, hes' in portland
06:29 lispy|web He said he'd be there Sunday
06:30 kowey cool
06:30 lispy|web We didn't have any gumption left by the time he arrived (train was very late)
06:30 lispy|web I think this solved my iconv problem cabal install haskeline --reinstall --extra-lib-dirs=/opt/local/lib --extra-include-dirs=/opt/local/include
06:31 lispy|web I was wrong
06:31 lispy|web I still got the error
06:33 kowey @localtime lispy|web
06:33 lambdabot Local time for lispy|web is Sat Nov 14 22:33:34 2009
06:34 kowey did the other newbies have fun?
06:34 lispy|web Not sure :)
06:34 kowey we had a little trouble getting started; next sprint really need to work on more active mentoring
06:34 lispy|web I _hope_ so.  But I'm a terrible judge of such things.
06:34 kowey especially for the getting-darcs-installed phase
06:34 lispy|web ah
06:35 lispy|web So, I've ran the command I pasted above with haskelline replaced by iconv and pcre-light and I still get the failure
06:35 kowey Judah should know
06:35 lispy|web doing a cabal clean on darcs
06:35 kowey also, I think you can search the IRC logs for haskeline by kowey
06:36 lispy|web The good news is that I figured out a nice way to remove commutex.
06:36 kowey ah, here you are : http://trac.haskell.org/haskeline/ticket/74
06:36 lispy|web I made two higher order functions to convert between the commute and commutex calling conventions
06:37 lispy|web Then replaced all the calls to commutex with a direct call to commute.  The places that required the commutex type I added a call to the higher order function (only about 3 places)
06:37 lispy|web The type-witnesses make this all very safe
06:38 lispy|web The reason I'm doing this is because the existence of both commute and commutex (and having them be mutually recursive) makes it harder to debug the exponential merge issue
06:40 kowey so isn't this just replacing commute + commutex with commute + the higher order function?
06:40 kowey Johnanthan's patches are having a little trouble : http://bugs.darcs.net/patch64
06:41 lispy|web The higher order function just changes the tuple types. So it is very easy to verify that you don't get into this recursive calling of each other
06:41 lispy|web kowey: he had to use mutt as his sendmail
06:47 * kowey takes breakfast
06:47 lispy|web kowey :: Breakfast -> ??
06:47 mornfall Breakfast time indeed.
07:27 tommd joined #darcs
07:50 tommd @tell lispy|web At 1 meg a minute it will take another ~5 hours to finish.  It will be ready for you at 0600h - three hours before the Sunday portion of the Darcs Sprint starts.  http://community.haskell.org/~tommd/ghc.  Problems include pulling from GHC HEAD (merge non-termination) and "darcs put"ing (consumed 4GB of RAM).
07:50 lambdabot Consider it noted.
08:02 darcscommitbot joined #darcs
08:03 darcswikibot joined #darcs
08:25 kowey joined #darcs
08:27 tux_rocker joined #darcs
08:27 nomeata joined #darcs
08:27 nomeata good morning
08:28 mornfall Morgen.
08:28 tux_rocker guten morgen :-)
08:29 kowey :-)
08:29 tux_rocker any Portland people still around?
08:29 tux_rocker @seen lispy
08:29 lambdabot Unknown command, try @list
08:29 mornfall tux_rocker:  [07:47]     lispy|web | #darcs       | kowey :: Breakfast -> ??
08:32 stalker joined #darcs
08:32 kowey I think I'll do a little introductory round - what we're working on, what questions/problems we may have, also note the upcoming meetings, lunch, roadmap
08:32 kowey and hope that's just not going to waste everybody's time...
08:35 mornfall ertai_: ping?
08:35 mornfall kowey: I'd guess you could shorten that a little.
08:35 mornfall kowey: Just announce meetings.
08:35 mornfall kowey: And ask for questions.
08:36 mornfall I think we roughly know who's up to what from yesterday evening.
08:36 kowey OK
08:37 kowey err... now?
08:37 kowey right... I go and hope for the best
08:39 kowey phew
08:42 luca-darcs-sprin joined #darcs
08:42 luca-darcs-sprin Guten Morgen Alles!
08:49 kowey our patch tracker is chock full of things to review :-)
08:50 mornfall luca-darcs-sprin: Guten Morgen Herr Luca.
08:55 tux_rocker_ joined #darcs
08:58 Heffalump morning sprint!
09:02 mornfall Morning Heffalump.
09:03 tux_rocker_ does anybody know an IRC nickname for Judah?
09:05 darcscommitbot 14 Nov 16:16 - Workaround a System.Directory.createDirectory issue. (Salvatore Insalaco)
09:05 tux_rocker_ I'm going to blame the negative characters fed to Prelude.chr on Haskeline :-)
09:05 darcscommitbot 15 Nov 08:59 - Add a comment to the performGC workaround in amNotInRepository function. (Petr Rockai)
09:05 tux_rocker__ joined #darcs
09:06 kowey could this be turned into some sort of Haskeline test case?
09:09 kowey tux_rocker_: ahem... (Windows :-))
09:09 tux_rocker__ presumably, yes
09:09 Quiark joined #darcs
09:18 nikon joined #darcs
09:18 nikon http://pastebin.ca/1671962
09:20 lelit joined #darcs
09:32 mornfall nikon: http://pastebin.ca/1671970
09:36 kowey Windows meeting, done
09:36 kowey official binaries are coming to town (one day)
09:41 tux_rocker__ joined #darcs
09:47 tdanecker joined #darcs
09:57 tux_rocker__ my internet connection is very flaky
09:57 tux_rocker__ am I the only one?
09:58 mornfall tux_rocker__: Dunno, seems to work fine here.
10:13 eivuokko joined #darcs
10:16 eivuokko Just wanted to ask about info on MAPI (the issue 1688), is Ms deprecating whole MAPI, including the Simple MAPI, which darcs "uses."
10:17 kirby joined #darcs
10:18 kowey eivuokko : I just walked over and grabbed kirby_ so we could talk about this
10:18 kirby_ ok, there was a question about MAPI support? :)
10:18 kowey "Just wanted to ask about info on MAPI (the issue 1688), is Ms deprecating whole MAPI, including the Simple MAPI, which darcs "uses."
10:18 kirby_ as far as I know it, MAPI is being deprecated altogether
10:18 eivuokko Ms MAPI support.  I'm someone who has written MAPI module for Win32 and hacked on darcs mapi support some in past.
10:19 kirby_ with Exchange 2007 the primary way to interact is with WebDAV
10:19 tux_rocker__ we just decided we were going to drop MAPI support?
10:19 kirby_ BUT
10:19 kirby_ I tought about the fact that in Exchange settings where SMTP is disabled
10:19 kirby_ +
10:19 eivuokko kirby, Simple MAPI too?  It's pretty much indenpendent?
10:19 kirby_ MAPI is still the only way to send mail
10:19 kirby_ eivuokko: that one is being substituted by a .NET API
10:20 eivuokko Ah, cool.
10:20 kirby_ so
10:20 eivuokko Thanks for info.
10:20 kirby_ I think that we cannot really disable that support
10:20 eivuokko And yeah, no .net support darcs could support for now...
10:20 kirby_ yes, exactly
10:20 kirby_ we could ship a .net binary
10:20 kirby_ that do the sending
10:20 kirby_ but... we will think about it for 2.5 ;)
10:21 eivuokko If you have enough hackers for it :-P
10:21 eivuokko Oh, well.  Thanks for the info and gl.  And thanks kowey.
10:22 kowey eivuokko: check out http://wiki.darcs.net/Sprints/2009-​11#windows-support-meeting-saturday too
10:22 eivuokko Can't connect -.-
10:22 kirby_ http://books.google.com/books?id=xWHZwBmW-h0C&amp​;pg=PA9&lpg=PA9&dq=simple+mapi+deprecated​&source=bl&ots=_XZYB-Tzfg&sig=k6muzlC​6Hkw7sE2iArplHmq3QkE&hl=it&ei=LNb_So7qEY3​4mgOF35T3Cw&sa=X&oi=book_result&ct=re​sult&resnum=5&ved=0CBoQ6AEwBA#v=onepage&a​mp;q=simple%20mapi%20deprecated&f=false
10:22 kirby_ more information about Simple MAPI deprecation :)
10:23 kirby_ i talks about using Extended MAPI
10:23 kirby_ but that was deprecated too, in Exchange 2007
10:24 eivuokko Well, mainly the annoyance is that there are other Simple MAPI providers, like Thunderbird
10:24 kirby_ we think to ship with a simple mail sender in Windows
10:24 kirby_ and use it to send mails
10:24 eivuokko That, in some cases, had some real use cases.
10:25 eivuokko That was brushed aside before "If users want it, they can install smtp themselves." or so.
10:25 eivuokko 3-4 years ago, or something, though.
10:25 kirby_ actually we want to improve darcs send experience for windows users in 2.5
10:25 kirby_ the objective is: darcs send should "just work" :)
10:26 kirby_ how... we have still to think about it ;)
10:26 eivuokko Well, sending mail isn't "just work" nowdays with people sending mails from different WLANs. :)
10:26 kirby_ sure, that's because we are thinking about it ;)
10:27 kirby_ actually I feel that darcs is very functional on windows for local and HTTP repository
10:27 kirby_ improving the SSH and SMTP experience is the next step
10:27 eivuokko I dunno, I just use darcs locally nowdays.
10:27 eivuokko Ssh is a bit pain because putty has no good installer (and register etc entries.)
10:27 kirby_ we want to ship an installer
10:28 kirby_ with a ssh implementation
10:28 kirby_ lots of good intentions! :)
10:28 eivuokko Lots of work and testing, that.
10:28 kirby_ we will try... we want to make Windows a first-class citizen in darcs world
10:30 eivuokko kowey, will wiki.darcs.net/darcs.net be up soonish?
10:34 eivuokko Well, I think I'm off to study some math, have fun.  :-)  I'll check the link some later time.
10:34 kowey oh my!
10:34 kowey darcs.net is down
10:35 kirby_ too many patches!
10:35 luca-darcs-sprin LOL
10:36 kowey support request filed
10:37 darcs-cache-dir oppps
10:41 tux_rocker joined #darcs
10:41 beschmi joined #darcs
10:48 mornfall [11:47]       QwertyM | #kde-devel   | taras: that and I can't ever checkout a large repo properly without git failing.. I miss amarok dailies :\ wish the tarballs came with the .git folder in them, cause
10:50 kowey eivuokko: http://hpaste.org/fastcgi/hpa​ste.fcgi/view?id=12166#a12166 (while waiting)
10:52 nikon_ joined #darcs
10:54 nikon__ joined #darcs
11:01 Heffalump re the MAPI stuff, could darcs use hsdotnet to interact with the .NET API?
11:01 tux_rocker joined #darcs
11:02 Heffalump kowey: is the lack of darcs.net a problem for the sprint?
11:05 eivuokko Heffalump, last I tried, hs-dotnet isn't anywhere near the quality needed, it also introduces some installation issues because of the way wrappers work.
11:06 Heffalump ah, I thought it was quite simple and easy to use, but I didn't get very far with it
11:06 Heffalump you're talking about Sigbjorn's untyped binding rather than the UNSW typed wrappers, right?
11:07 eivuokko hs-dotnet was the name of the package, as I recall, and yeah it was Finne's.
11:08 kowey right now, we're quite happy that we're using a distributed revision control
11:08 kowey because somebody here needs a complete copy of the repo
11:13 mornfall http://128.131.203.6/mainline/ (it has couple extra patches though, but shouldn't be much of an issue)
11:14 mornfall Ok, prolly this doesn't work here.
11:15 eivuokko What is Tortoise ssh?
11:17 Heffalump mornfall: might matter if people depend on those patches..
11:17 mornfall http://repos.mornfall.net/darcs/mainline/
11:19 mf_ joined #darcs
11:19 nikon joined #darcs
11:19 mornfall mf_: http://repos.mornfall.net/darcs/mainline/
11:19 kirby_ eivuokko: it is a fork of plink from putty that handles messages with dialog boxes
11:19 kirby_ eivuokko: it is used by TortoiseSVN
11:20 eivuokko Hm
11:20 kirby_ eivuokko: it will resolve the dreaded "please accept the remote hash" issue that locks darcs
11:21 eivuokko Dialog boxes are not good for running without desktop, but I can see the point certainly.
11:21 kirby_ eivuokko: it is not mandatory to use it, it is just a nice default :) you can use any ssh implementation
11:24 eivuokko Do you have a plan on which installer-generator you're gonna use?
11:24 eivuokko And exactly what goes into installer, and how to get it working.  Like building docs in Windows?
11:24 kirby_ we were thinking about using the one on hackage (bamse if I recall correctly)
11:24 eivuokko Yes, generting msi would be good imo.
11:25 kirby_ inside it you will get the darcs binary, documentation, required curl library
11:25 kirby_ yes, they will be msi
11:25 kirby_ and TortoiseSVN and a mail client (msmtp-like) as optional installas
11:25 kirby_ installs
11:25 kirby_ and it should add darcs to the default path
11:25 kirby_ maybe the next sprint :)
11:26 eivuokko Does documentation build in Windows?
11:26 eivuokko (Without horrible hacks and installing stuff like libpng on msys)
11:26 kirby_ I think it doesn't easily...
11:26 eivuokko And that wasn't a joke, it used to require that.
11:26 kirby_ but I didn't try recently :)
11:26 kirby_ I don't know if we switched away from TeX yet
11:27 eivuokko That'd building installer a bit annoying.
11:27 kirby_ not really: I don't want to provide an installer builder :)
11:27 eivuokko TeX isn't neccesarily problem, miktex is pretty well behaved.
11:27 kirby_ I want to provide and installer
11:27 eivuokko Yeah, well, hf generating docs on other machine...
11:27 kirby_ providing an installer that builds  from sources will be very very hard
11:27 kirby_ so I won't do it probably ;)
11:27 eivuokko As bamse *will not* work in non-windows iirc.
11:28 kirby_ yes
11:28 kirby_ it doesn't work
11:28 eivuokko I meant that to buidl an installer, you'd want to work in Windows machine, and if you have to build docs somewhere else...
11:28 kirby_ that's not a real problem for me, as I use Windows on a Virtual Machine :)
11:41 mornfall nikon: http://repos.mornfall.net/darcs/mainline/
11:45 eivuokko1 joined #darcs
13:16 arjanb joined #darcs
13:24 dcoutts joined #darcs
14:18 kowey joined #darcs
14:24 Quiark joined #darcs
14:26 kirby joined #darcs
14:26 tux_rocker joined #darcs
14:26 luca-darcs-sprin joined #darcs
14:27 luca-darcs-sprin Hello everyone
14:27 gh_ joined #darcs
14:28 tux_rocker Guten Nachmittag Herr Molteni
14:30 kowey dacrs.net is back up
14:31 kirby joined #darcs
14:32 dcoutts anyone think this bug is known/unknown ?
14:32 dcoutts darcs: bug in darcs!
14:32 dcoutts Empty filename in is_relative at src/Darcs/URL.hs:60 compiled Nov  2 2009 20:08:25
14:32 dcoutts 2.2.1
14:33 kirby_ is it reproducible?
14:33 dcoutts yes
14:33 kirby_ which platform?
14:34 dcoutts darcs push nonexistant/
14:34 dcoutts linux
14:34 kowey I think that one has been fixed
14:34 dcoutts the trailing / is necessary
14:34 dcoutts kowey: ok, I had a quick look in the bug tracker, but didn't check closed tickets
14:35 kowey http://bugs.darcs.net/issue1162 <-- is this it?
14:35 kirby_ I just tried on 2.3.1, it seems to work
14:36 dcoutts kowey: looks very similar, yes
15:00 eivuokko joined #darcs
15:14 kowey joined #darcs
15:16 luca-darcs-sprin where are yesterday's photos?
15:17 mornfall luca-darcs-sprin: On my CF.
15:19 * kowey broke the buildbot
15:21 tux_rocker joined #darcs
15:23 luca-darcs-sprin mornfall:  do you want to see our implementation of bucketing?
15:27 mornfall luca-darcs-sprin: I'll look at the patches. :)
15:35 nomeata To test darcswatch/roundup-integration, can someone please review and apply http://bugs.darcs.net/patch66, without closing the ticket manually?
15:40 luca-darcs-sprin this macosx "instrument", based on dtrace, is the most buggy thing ever
15:41 ertai_ mornfall: pong
15:41 * kowey is on it, nomeata
15:43 kowey done
15:44 mornfall ertai_: Hi, could you please install latest hashed-storage? Thanks.
15:44 mornfall ertai_: Also, the slave previously died with strange linker errors.
15:44 ertai_ yep I have a custom timeout on my linker :)
15:45 ertai_ mornfall: I had problems with ld running my machine completly (allocating Gigs of memory)
15:46 ertai_ mornfall: hashed-storage-0.4.3 installed
15:50 tux_rocker_ joined #darcs
15:52 mornfall ertai_: (.data+0x78): undefined reference to `networkzm2zi2zi1zi4_NetworkziURI_a59_closure'
15:52 mornfall ertai_: And a bunch of similar.
15:52 mornfall ertai_: : - (
15:52 mornfall ertai_: Maybe cabal install HTTP --reinstall is in order?
15:53 mornfall ertai_: Or maybe even network
15:53 luca-darcs-sprin maybe buketing is working...
15:53 ertai_ mornfall: hum this look indeed a different issue
15:55 ertai_ mornfall: HTTP and network installed (a newer version maybe, I didn't needed --reinstall)
15:56 mornfall forced, let's see
15:57 mornfall ertai_: Looks better, thanks!
15:57 tux_rocker_ joined #darcs
16:05 luca-darcs-sprin ARGH @--no-test
16:07 mornfall luca-darcs-sprin: : - P
16:07 kami joined #darcs
16:08 kami left #darcs
16:08 luca-darcs-sprin does anyone remember the number of the bucket cache?
16:08 tux_rocker__ joined #darcs
16:12 kowey 1624
16:12 Heffalump should I be including bugs@darcs.net when I reply to submitted patch?
16:12 Heffalump It happens by default with reply-all
16:13 luca-darcs-sprin kowey: thanks
16:13 sm joined #darcs
16:17 * Heffalump definitely likes darcswatch auto-handling the acceptance of patches but is less sure about it sending emails to tell us it's watching submitted patches
16:18 kowey Heffalump: there's a ticket for that :-)
16:18 Heffalump I see the review team is going to be busy for a while post-sprint :-)
16:19 kowey http://bugs.darcs.net/issue1689
16:20 darcscommitbot joined #darcs
16:20 darcswikibot joined #darcs
16:21 sm bots were out because my server was restarted due to excessive swap, sorry about that
16:25 darcswikibot 15 Nov 16:20 - update msmtprc instructions (david)
16:40 luca-darcs-sprin I'm going home folks
16:40 luca-darcs-sprin thank for all...
16:41 luca-darcs-sprin BYE :D
16:42 gwern Heffalump: I really just wants buttons to click accepted rejected obsolete
16:44 Heffalump as opposed to dropdowns?
16:45 darcswikibot 15 Nov 16:44 - making it better (kowey)
16:47 tux_rocker__ joined #darcs
17:01 nomeata gwern: talking about darcswatch? This feature is now a bit more likely to be implemented. If the sprint would go on for another few hours, I’d probably do it.
17:03 lispy|web joined #darcs
17:06 lispy|web Hello!
17:06 lambdabot lispy|web: You have 1 new message. '/msg lambdabot @messages' to read it.
17:06 lispy|web ?messages
17:06 lambdabot tommd said 9h 16m 2s ago: At 1 meg a minute it will take another ~5 hours to finish. It will be ready for you at 0600h - three hours before the Sunday portion of the Darcs Sprint starts. http://
17:06 lambdabot community.haskell.org/~tommd/ghc. Problems include pulling from GHC HEAD (merge non-termination) and "darcs put"ing (consumed 4GB of RAM).
17:06 lispy|web ?tell tommd thanks!
17:06 lambdabot Consider it noted.
17:10 tux_rocker joined #darcs
17:11 gwern Heffalump: dropdowns? you mark rejected or obsolete by emailing darcswatch and praying your email is just right
17:13 mf_ joined #darcs
17:14 lispy|web darcs get is really slow
17:14 lispy|web I'm now scping this copy of the ghc repo because get looked like it would take a decade
17:14 * lispy|web notes he doesn't have a decade to spare today
17:15 mornfall lispy|web: We have semi-designed an interim fix for that.
17:15 lispy|web mornfall: semi-explain :)
17:15 gwern doesn't ghc provide a tarball of the repo up to a certain date?
17:15 lispy|web gwern: this is tommd's copy that exhibits a nasty merge issue
17:15 mornfall It involves creating a pair of tarballs upon tags in public repositories (you'd need to enable that in the defaults file on the remote repo, I suppose).
17:16 lispy|web Heffalump: I was a little confused by your reply to my commutex patch.  Is the patch accepted and you're pointing out future work?
17:16 lispy|web Heffalump: or was the patch rejected?
17:17 lispy|web mornfall: ah
17:17 lispy|web mornfall: yeah, that has "not ideal" written all over it :)
17:18 tux_rocker_ joined #darcs
17:19 mornfall lispy|web: Sure. But it's likely to work and fix the issue at hand.
17:19 mornfall lispy|web: And it's doable in the 2.4 horizon.
17:21 lispy|web :)
17:25 gwern the mailing list and bugtracker activity is getting a little irritating...
17:28 kowey gwern: it'll settle down over the week, sprint's winding up in Vienna
17:28 JaffaCake joined #darcs
17:29 kowey and also we're working on improving the automation so that we get fewer admin messages on the mailing list
17:32 lispy|web gwern: thanks for the feedback though!
17:33 gwern :)
17:33 gwern the annoying thing is that gmails' mute isn't working well
17:33 gwern I guess all sorts of the same-issue or same-bug are showing up as different conversations
17:35 tux_rocker_ joined #darcs
17:38 lispy|web gwern: I don't think gmail's mute ever works in a meaningful way :
17:38 lispy|web :)
18:01 lispy|web joined #darcs
18:01 Heffalump joined #darcs
18:01 Korusef joined #darcs
18:01 kolmodin joined #darcs
18:02 darcswikibot joined #darcs
18:02 stalker joined #darcs
18:02 amuck joined #darcs
18:02 Igloo joined #darcs
18:02 Jaak joined #darcs
18:02 lisppaste2 joined #darcs
18:02 mjrosenb joined #darcs
18:02 dleverton joined #darcs
18:02 faheem joined #darcs
18:02 lambdabot joined #darcs
18:02 karlpinc joined #darcs
18:02 mornfall joined #darcs
18:02 mle joined #darcs
18:02 PHP-Pierre_ joined #darcs
18:02 Riastradh joined #darcs
18:02 gwern joined #darcs
18:02 dixiecko joined #darcs
18:02 gbeshers joined #darcs
18:02 waern joined #darcs
18:02 thorkilnaur joined #darcs
18:02 kpreid_ joined #darcs
18:02 Jerri_ joined #darcs
18:02 nwf joined #darcs
18:02 tommd joined #darcs
18:02 nomeata joined #darcs
18:02 tdanecker joined #darcs
18:02 arjanb joined #darcs
18:02 dcoutts joined #darcs
18:02 gh_ joined #darcs
18:02 eivuokko joined #darcs
18:02 sm joined #darcs
18:02 darcscommitbot joined #darcs
18:02 mf_ joined #darcs
18:02 JaffaCake joined #darcs
18:02 Liskni_si joined #darcs
18:02 kosmikus joined #darcs
18:02 tux_rocker_ joined #darcs
18:03 kowey joined #darcs
18:03 ertai joined #darcs
18:03 Quiark joined #darcs
18:03 tmug joined #darcs
18:03 idnar joined #darcs
18:08 kwallmark joined #darcs
18:14 iago joined #darcs
18:20 darcswikibot 15 Nov 18:19 - Add all demands for 2.4 release to roadmap (ReinierLamers)
18:20 dons lispy|web: i'm sorry, looking at my work this week, i don't think i'll have time to day to hack. got to get a bunch done before weds, so won't be in today to work on darcs. :/
18:22 lispy|web dons: okay.  No worries
18:22 cygnus joined #darcs
18:23 tux_rocker_ joined #darcs
18:24 felipe joined #darcs
18:25 darcscommitbot 15 Nov 14:45 - hlint error: Use first instead of *** (Joachim Breitner)
18:27 tux_rocker__ joined #darcs
18:31 j3h joined #darcs
18:32 lispy|web j3h:     http://bugs.darcs.net/patch39
18:34 j3h joined #darcs
18:34 lispy|web j3h: http://bugs.darcs.net/patch39
18:36 lispy|web j3h: http://lists.osuosl.org/pipermail/da​rcs-users/2009-November/022062.html
18:38 kowey sorry for the noise, folks :-)
18:39 kowey there's a bit of construction work in the patch tracker
18:39 * lispy|web laments the size of the GHC repo
18:39 kowey but the result is a lot less admin overhead
18:46 * tux_rocker__ yawns
18:52 gh_ hi sprinter
18:52 gh_ +s
18:52 kowey hey, hey!
18:53 kowey looks like we're going to be able to put away our jackhammers
18:53 kowey now to amend patches you only need to do darcs send --subject '[patchNNN]' ; no crazy stuff
18:54 kowey woo!
18:55 kowey have fun, lispy|web and friends!
19:00 lispy|web kowey: oh, interesting
19:02 kowey phew, recap is darcs-users is nosy, so everything is in one place
19:02 kowey darcswatch now automatically sets status=accepted so no need to do crazy things in the subject
19:02 tux_rocker mornfall proposed that we just "use" hackage's lack of authentication to release a new version of utf8-string that fixes my unit tests :-)
19:03 Heffalump kowey: do amendments replace the old ones so it's very clear which is "current"?
19:03 kowey and due to darcs-users being nosy, just setting --subject='[patchXX]' works
19:03 nomeata tux_rocker: that was me, don’t say bad things about mornfall :-)
19:03 kowey if you look on the web interface, it appears the attachments with same name are sorted by date
19:03 kowey and finally due to nice mornfall procmail hacking, you don't need to say "darcs patch:" in the subject line
19:03 kowey just the fact the mail is directed at patches@ means that the tracker will pick it up
19:04 lispy|web oh, you don't have to hack procmail manually if you use autoproc (on hackage)
19:04 lispy|web it provides a eDSL for writing procmail
19:04 kowey SO it should all be little smoother, with fewer hoops to jump through
19:04 kowey but we're all tired and things may be broken
19:04 kowey tests seem encouraging
19:06 kowey oh and darcs-users now treats darcswatch mail as spam so this shouldn't generate so much admin noise on list
19:06 Heffalump cunning :-)
19:06 Heffalump though it'd be nice to see the "accepted" mails
19:06 kowey ^^ mornfall :-)
19:10 kowey ok, spam filter now refined to only treat X-Roundup-patch-status: needs-review as spam
19:18 lispy|web http://queue.acm.org/detail.cfm?id=1595636
19:19 tommd lispy|web: still copying?
19:19 lambdabot tommd: You have 1 new message. '/msg lambdabot @messages' to read it.
19:19 tommd It would probably have been faster to run over to monk with a USB thumb drive :-)
19:19 tommd @messages
19:19 lambdabot lispy|web said 2h 12m 43s ago: thanks!
19:23 kowey err, there is this faster server you can darcs get the GHC repo from
19:23 kowey by kili...
19:24 kowey http://www.mail-archive.com/darc​s-users@darcs.net/msg06092.html
19:24 lispy|web tommd: just finished about 1 minute ago
19:28 lispy|web tommd: running profiling the same way you were now
19:33 tommd1 joined #darcs
19:34 tommd1 kowey: If you were talking to me, this is a special repo that has hit an exponential or infinite merge when pulling (and consumes over 4GB or RAM on 'darcs put').  Useful for testing.
19:34 kowey oh ok
19:35 * kowey wasn't paying enough attention
19:35 tux_rocker joined #darcs
19:42 darcscommitbot 15 Nov 17:20 - Remove does not fail only warns on already removed or not added files. ()
19:42 darcscommitbot 15 Nov 19:23 - Resolve issue540: darcs remove --recursive ()
19:42 mornfall tommd1: http://bugs.darcs.net/issue1540
19:43 lispy|web We spend a lot of time in Darcs.Patch.Prim (==)
19:43 lispy|web Which isn't very helpful
19:44 tommd1 Oh, lispy|web.  Do you want or need the profiles I ran yesterday?  You can reproduce in 10 minutes, but if you want my copy then let me know.
19:44 lispy|web tommd1: I have all the setup locally to reproduce it
19:44 lispy|web tommd1: plus I removed commutex
19:44 lispy|web tommd1: now I only have commute
19:44 tommd1 ohhh
19:44 lispy|web so my profile is more helpful that way
19:45 tommd1 And it still hangs?
19:45 tux_rocker lispy|web: most of the time is in comparing hunks iirc
19:45 lispy|web tommd1: yeah
19:45 lispy|web tommd1: I guess, I only waited 10 minutes :)
19:45 tommd1 Halting problem.  10 is enough.
19:46 lispy|web tux_rocker: Well, the question I'm after is _why_ is it doing so much work to perform a merge
19:46 tux_rocker lispy|web: did you consider manually marking cost centers?
19:47 lispy|web tux_rocker: The problem understanding the control flow
19:47 lispy|web but yeah, we've done that
19:47 lispy|web The profile isn't as helpful as I thought it would be fore infering control flow
19:48 tux_rocker there's also the assertConsistent function
19:48 lispy|web What does that do?
19:48 tux_rocker arjanb once thought that that had some problems of its own
19:48 tux_rocker it is called at various places in the merge code
19:49 tux_rocker for checking that a certain created patch is consistent
19:49 tux_rocker but arjanb said that it was called many times on the same data
19:50 lispy|web I wonder where this == function is actually implemented also
19:51 lispy|web I guess it's called by unsafeCompare
19:52 * arjanb digs in his memory of the previous sprint
19:52 tux_rocker hi arjanb :-)
19:56 * tux_rocker thinks it is time to go
20:01 arjanb the == you see in profile is probably the default instance for FilePatchType
20:04 trofi joined #darcs
20:04 arjanb i remember trying to improve the == but that didn't help much because the problem is in the amount of times it's called
20:05 trofi hia. we are slowly preparing for ghc-6.12 release and i've tried to build current darcs-darcs: http://dpaste.com/120812/
20:06 trofi appied http://dpaste.com/120813/
20:06 trofi Maybe it's worth releasing 2.3.2 with http://dpaste.com/120813/ ?
20:07 trofi btw, darcs-2.3.1 with http://dpaste.com/120813/ builds and wirksforme
20:08 Heffalump those changes aren't enough, you also need to fix a few other things to make the tests and things build properly
20:08 Heffalump we've got them submitted already, but held off applying because of not wanting to cause conflicts
20:08 Heffalump I guess a 2.3.2 might make sense
20:09 trofi ah, ok
20:11 Heffalump the bit you pasted doesn't actually cause conflicts so perhaps we should apply that bit right away
20:36 gwern joined #darcs
20:56 gh__ joined #darcs
21:12 mf_ joined #darcs
21:25 dixiecko joined #darcs
21:41 patch-tag joined #darcs
21:41 patch-tag build problem, what do I do? http://hpaste.org/fastcgi/hpa​ste.fcgi/view?id=12183#a12183
21:43 Heffalump hack around it locally I suggest
21:43 tommd joined #darcs
21:43 Heffalump and ask on the list
21:46 eivuokko joined #darcs
21:51 visq joined #darcs
22:06 patch-tag working now
22:09 cygnus joined #darcs
22:12 cygnus So I'm looking at adding a command to darcs to set the "_darcs/prefs/email" value, since it looks like that command doesn't exist.  What do people think about this?  I don't think we should use "setpref", because this is a preference that doesn't travel with the repository.
22:14 Heffalump is writing into the file onerous or confusing?
22:15 cygnus Manually, you mean?
22:23 cygnus If you mean manually, then yes; it seems reasonable to me to say, nobody should need to edit anything in _darcs by hand for any reason, ever.  I mean, making _darcs's representation human-readable is nice, but nobody should ever need to know about it.
22:24 cygnus I'm thinking of something like "darcs repopref email <...>".
22:25 cygnus or "setdefault".
22:28 kwallmark darcs localpref?
22:33 cygnus Yeah, that works
22:43 kwallmark patch-tag: http://bugs.darcs.net/issue1627
23:13 gwern joined #darcs
23:21 dons_ joined #darcs
23:21 faheem_ joined #darcs
23:21 mjrosenb_ joined #darcs
23:23 * tommd is sad that he skipped todays hacking and hopes it went well
23:39 twb joined #darcs
23:46 visq joined #darcs
23:47 lispy|web cygnus: http://lists.osuosl.org/pipermail/da​rcs-users/2009-November/022115.html
23:48 Heffalump I tend to agree that not having to edit files in _darcs is a good design principle.

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