IRC log for #darcs, 2008-10-27

Time	Nick	Message
00:09		beschmi left #darcs
00:29		sm joined #darcs
00:43	dons	cabal install darcs, woo
00:46	dons	http://galois.com/~dons/images/darcs.svg
00:51		kpreid_ joined #darcs
00:54		nwf joined #darcs
02:08		twb joined #darcs
02:46		dcoutts joined #darcs
02:46	dcoutts	@arr!
02:46	lambdabot	Arrr!
02:49	lispy	dcoutts: hey
02:49	lispy	did you see what I just wrote in #haskell?
02:49	* dcoutts	looks
02:50	lispy	dcoutts: I made it to link errors
02:50	lispy	dcoutts: yeah on windows
02:50	dcoutts	hmm
02:50	dcoutts	lispy: what was the problem with autogeneration?
02:50	dcoutts	which autogeneration in particular?
02:51	lispy	I don't have 'darcs'
02:51	lispy	so auto generation fails
02:51	lispy	anything that needs to call darcs
02:51	lispy	because I don't have it installed!
02:51	lispy	So I just commented out that autogen stuff...it's like ThisVersion and Context
02:51	lispy	who cares
02:51	dcoutts	ohhh
02:52	dcoutts	right, my bad
02:52	lispy	dcoutts: http://hpaste.org/11499
02:52	dcoutts	yeah, so it should be catching those
02:52	lispy	dcoutts: right it should catch those and do whatever autoconf does when darcs isn't built yet
02:52	dcoutts	yep
02:53	dcoutts	ok, so the first link error is related to hslibwww.c I presume
02:53	lispy	right
02:53	lispy	let's focus on the other one
02:53	lispy	zlib
02:53	dcoutts	aye, so are you building with external-zlib true or false?
02:53	lispy	let me check
02:54	lispy	hmm...I didn't change it, I'll re-run config with verbose
02:54	lispy	False
02:54	dcoutts	hmm it should fail by not finding -lz
02:54	dcoutts	run with build -v and check if it's passing -lz
02:55	lispy	dcoutts: I don't see a -lz
02:56	lispy	I haven't installed much on this machine
02:56	dcoutts	lispy: and you've got the recent patch to use extra-libraries: z when flag external-zlib is false?
02:56	dcoutts	if flag(external-zlib)
02:56	dcoutts	build-depends: zlib == 0.5.*
02:56	dcoutts	cpp-options: -DHAVE_HASKELL_ZLIB
02:56	dcoutts	else
02:56	dcoutts	extra-libraries: z
02:56	lispy	oh, I missed that one
02:57	dcoutts	ok, with that patch I expect it'll fail by not finding -lz
02:57	dcoutts	which is good
02:57	lispy	you're correct
02:57	dcoutts	you then need the external C zlib
02:57	dcoutts	or the zlib binding package
02:57	lispy	I have zlib package because I needed it for cabal-install
02:58	dcoutts	hmm, I wonder why it did not select it automatically
02:58	dcoutts	oh, I know, because you've got 0.4.0.4
02:58	lispy	maybe because I tweaked the .cabal file
02:58	dcoutts	and darcs now needs 0.5.x
02:58	lispy	I upgraded cabal
02:58	dcoutts	you'll need the darcs version of zlib
02:58	lispy	oooh
02:58	lispy	hmm
02:58	lispy	how do we get that?
03:02	lispy	one thing I can do is grab zlib dll from zlib.net
03:02	lispy	but I'm not sure how I make the build use it
03:03	dcoutts	lispy: I've no idea about making it use an external zlib.dll
03:03	dcoutts	lispy: but the darcs version of my zlib binding is pretty easy
03:03	dcoutts	darcs get http://code.haskell.org/zlib/
03:03	lambdabot	Title: Index of /zlib
03:03	dcoutts	cd zlib && cabal install
03:04	lispy	I don't have darcs :(
03:04	lispy	I need a zip file
03:04	sm	the sprint continues!
03:05	sm	go, portland, go!
03:05	dcoutts	lispy: well you can always get an old darcs binary
03:06	dcoutts	sm: heh
03:06		Jerri joined #darcs
03:06	* sm	got the cold shoulder from gf tonight, BACK TO WORK...
03:07	sm	lispy: yeah. and since I have strangely become windows binary emissary, here's http://wiki.darcs.net/index.html/CategoryBinaries
03:07	lambdabot	Title: CategoryBinaries - DarcsWiki
03:07	sm	I mean http://wiki.darcs.net/index.ht[…]cb62712589b392074
03:07	lambdabot	Title: CategoryBinaries - DarcsWiki, http://tinyurl.com/4wlxe2
03:08	lispy	dcoutts: nah, I tar'd it on another machine
03:10		adu joined #darcs
03:10	lispy	dcoutts: weird. I have that copy of zlib, 0.5, I run configure then I try to build and build tells me .\zlib.cabal has been changed, please re-configure.
03:11	dcoutts	hmm
03:11	lispy	my system clock is wrong, hang on
03:11	dcoutts	lispy: right I was going to ask about timestamps
03:11	lispy	yup, that fixed it
03:12	lispy	dcoutts: okay, just need libwww now
03:13	lispy	actually, I want libcurl
03:23	lispy	building libcurl is going to be hard
03:23	lispy	using curl.exe would be easy though
03:23	lispy	well, if darcs supports that I mean
03:24	dcoutts	lispy: it does doesn't it?
03:24	dcoutts	lispy: curl or wget
03:25	lispy	maybe, I'm trying to figure out how that's hooked up so I can change the build to not care about linking in libwww
03:26	lispy	maybe I should use HTTP
03:34	dcoutts	lispy: use configure -f-libwww
03:34	dcoutts	configure -f-libwww -fhttp
03:34	lispy	right, and then I added, build-depends: network, after build-depends: HTTP
03:35	lispy	now I need to fix some sort of autogeneration issue
03:35	dcoutts	lispy: oh does it need network in addition to HTTP?
03:35	* dcoutts	tries
03:35	lispy	I need to set PACKAGE_VERSION
03:36	lispy	in HTTP.hs
03:36	dcoutts	ohh
03:36	dcoutts	that should really be using a different method
03:36	dcoutts	like using ThisVersion.hs and passing the user agent in as a param
03:38	lispy	So we need to do a fair bit of cleanup before this "just works" on windows, but it's not tha far really
03:39	dcoutts	lispy: pushing a patch...
03:39	lispy	oof
03:40	lispy	darcs.exe get http://darcs.net/
03:40	lambdabot	Title: darcs
03:40	lispy	darcs failed: handleToFd not supported!
03:41	dcoutts	lispy: pushed the package version and network thing
03:41	lispy	whoa
03:41	dcoutts	I've no idea about the handleToFd thing
03:41	dcoutts	src/win32/System/Posix.hs:handleToFd _ = fail "handleToFd not supported!"
03:42	lispy	dcoutts: http://hpaste.org/11499#a1
03:43	dcoutts	heh, looks like a utf16 decoding problem
03:43	lispy	it's totally hosed
03:44	lispy	dcoutts: I should stop here. I don't even know here to begin
03:45	lispy	using HTTP really should have worked
03:45	lispy	I should have just had problems with transparent proxies
03:46	dcoutts	aye
03:46	dcoutts	lispy: what does it normally use on windows ?
03:50	lispy	not sure
03:51	lispy	but, HTTP should work as far as I know
03:51		clanehin joined #darcs
03:51	lispy	http://hpaste.org/11500
03:51	lispy	those are my notes
03:52	lispy	but of course hpaste cut me off :(
03:52	lispy	stupid hpaste
03:52	lispy	with it's silly limit
03:52	* lispy	reboots
03:58	lispy	dcoutts: my hunch is that the way we detect windows doesn't like the lack of cygwin/mingw
04:00	sm	why don't we use ghc --make for darcs, would anyone know ?
04:01	sm	it seems to work, given all the extra flags
04:03	dons	sm: it can't pick packages very well
04:03	dons	of course, we could do it that way. cabal does that, after picking all the flags
04:04	lispy	dons: did you see that FastPackedStrings (now ByteStringUtil) is marked: Stability: experimental, all the more reason to remove it :)
04:04	sm	it's come up because I'm trying to rebuild automatically with sp.. something is breaking on the huge command line
04:04	lispy	sp?
04:06	sm	http://searchpath.org . It has the ability to watch all your app's modules and re-run ghc --make when anything changes. Optionally run your tests too. It's extremely useful when hacking
04:06	lambdabot	Title: SearchPath — Automatic import chasing across the Internet for Haskell modules
04:07	lispy	dcoutts: ah, the problem with that windows build was that use_mmap = True, and on windows it needs to get defined as False
04:07	sm	really shrinks the feedback loop from code change to ghc error
04:08	lispy	ah. I'm not found of those types of programs, eclipse is like that, but yeah some people really like that
04:08	lispy	sm: can it be taught to use cabal instead of ghc --make?
04:09	dons	searchpath is dead. long live cabal install!
04:09	sm	lispy, maybe
04:09	sm	I'm using searchpath strictly for this auto-recompile/run-on-change feature
04:10	sm	nothing else does that yet, afaik
04:10	lispy	dons: looks like what we need next on for windows support is industrial strength http that can be cabal-installed
04:10	lispy	dons: either, adding pipelining and transparent proxy support to HTTP or make a hackage package for libcurl
04:11	dons	or use the galois curl package
04:11	dons	http://hackage.haskell.org/cgi[…]ipts/package/curl
04:11	lambdabot	Title: HackageDB: curl-1.3.2.1
04:14	lispy	dons: excellent, any idea if that builds on windows yet?
04:15	lispy	dons: also, does that API support pipelining?
04:15	dons	a) yes, b) not sure.
04:15	dons	but it is a full binding
04:15	dons	we originally wrote it so we could get windows support
04:15	dons	its just a binding to the underlying api, and the code is online, so we could merge in the darcs stuff to it
04:16	dons	http://code.galois.com/cgi-bin[…]url.git;a=summary
04:16	lambdabot	Title: code.galois.com Git - curl.git/summary
04:17	lispy	okay
04:17	lispy	modulo the licensing
04:17	dons	bsd?
04:17	dons	oh, darcs.
04:17	lispy	yeah, gpl :(
04:18	dcoutts	lispy: hmm, that's pretty odd, take a look at Setup.hs, it's using:
04:18	dcoutts	subst "USE_MMAP"
04:18	dcoutts	\| isWindows = show False
04:18	dcoutts	\| otherwise = show True
04:19	lispy	dcoutts: I had to kill Setup.hs
04:19	lispy	and I likely killed too much of it
04:19	dcoutts	lispy: heh
04:19	lispy	(i've since rebooted so hard to test)
04:19	dcoutts	lispy: so how did you generate Autoconf.lhs ?
04:19	dcoutts	by hand?
04:20	dcoutts	lispy: and why did you need to kill it? what was not working?
04:20	lispy	I forgot that it needed to be generated so it was the one from osx
04:20	dcoutts	ah
04:20	lispy	I didn't have a darcs.exe
04:20	lispy	So I kept getting that error from runInteractiveProcess
04:20	dcoutts	ah
04:20	lispy	and I started commenting out anything that looked like autogeneration
04:20	lispy	I just wanted to see what would work
04:21	dcoutts	hmm, it should have been ok
04:21	dcoutts	oh, no, ok
04:22	lispy	where did you push your patches to?
04:22	dcoutts	lispy: it's the two uses of rawSystemStdout
04:22	lispy	ah
04:22	dcoutts	I changed it from using exceptions to using exit codes
04:23	dcoutts	but of course if darcs does not exist it still throws exceptions
04:23	dcoutts	so we should have been using that all along
04:23	dcoutts	though it'll be harder to make that work with 6.10
04:23	lispy	that sucks
04:30	lispy	dons: if we could modify that curl library to package libcurl the way the zlib package does, then it could become usable for what I want.
04:31	lispy	dons: but as is, the package needs autotools and it doesn't come with the curl source
04:37	dcoutts	lispy: pushing another patch to your repo
04:37	dcoutts	lispy: should make it work on windows without darcs installed
04:44	lispy	dcoutts: cool, thanks
04:49	dons	mm.
05:31	* Heffalump	appears
05:34	twb	Phew, I finally finished the basic translation of .tex to reST .txt
05:34	twb	I still haven't even STARTED doing copy editing
05:50	* sm	hacks, hacks through the jungle that is sp
05:50	sm	go twb
05:50	sm	how are you liking reST ? used it much before ?
05:51	twb	Yes, I've used reST a lot
05:51	twb	To me, it acts the way word processors claim to
05:51	sm	me too. I can't say I love it. I can't think of anything much better either
05:51	twb	i.e. it gets most the syntax out of my face, but still has a straightforward mapping to syntactic objects
05:52	twb	Compared to everything else I've used, reST is definitely less painful.
05:52	sm	have you got a repo to look at ?
05:53	twb	Yes, but I don't have push access yet :P
05:54	sm	that's ok, I'm just looking forward to seeing a big darcs doc in reST, to compare
05:55	twb	I can send you the patches in just a moment.
05:55	sm	darn.. just retired my darcs-sm repo, which would have been the place I could have pushed them to see on the web
05:57	twb	I just sent output examples to darcs-users
06:03	sm	there it is.. thx
06:04	twb	sm: what's yer address? I'll bounce you the patch bundle
06:04	sm	simonjoyful.com
06:04	twb	sm: just don't apply it to a repo you care about, because it'll fuck it up
06:04	sm	ok.. don't send it to me please :)
06:05	twb	sent
06:05	sm	aieeee
06:05	twb	Oh.
06:05	twb	When I say "fuck it up" I mean "make pdf will fail, and probably you'll get lots of merge conflicts."
06:05	sm	AIEEEEE
06:06	* twb	is embarrassed.
06:06	lispy	twb: hmm...I prefer single column
06:06	sm	don't worry.. I'm just trying to avoid more pain
06:06	twb	sm: if it's any help, From: doesn't match your address.
06:06	twb	lispy: that's easy to do
06:06	lispy	twb: overall, I think the latex pdf is prettier, like the same color bg in the commands and whatnot instead of a yellow bg
06:06	twb	lispy: I actually jumped through some hoops to get two-column, because I prefer it to having huge margins or really wide body text
06:07	twb	lispy: the good news is that both work, so it doesn't matter which we prefer
06:08	sm	actually that two column is neat.. I haven't seen that from rst before
06:09	sm	do stylesheets for rst also control the style of the pdf ? I guess not
06:10	twb	sm: you have json stylesheets for rst2pdf
06:10	sm	I see
06:10	twb	e.g. {"styles": [["base", {"fontSize": 12}]]}
06:11	twb	Ugly, but the presentation is certainly separate from the source file's markup.
06:11	sm	and for all those file includes, you'll use sphinx ?
06:12	twb	Personalyl I'm not sold about using sphinx
06:12	twb	Partly because that'll break rst2pdf and rst2odt.
06:12	sm	I see
06:13	twb	I'd rather just have .. include:: command-apply.txt
06:13	twb	And then command-%.txt: src/Darcs/Commands/% preproc
06:13	twb	\t./preproc $< >$@
06:13	sm	oh yes.. I forgot rst can do that
06:13	twb	So preproc would just generate reST instead of TeX, and the static reST files would include the generated files
06:14	sm	and how do you fit the included files within the section hierarchy ? or did you just say that ?
06:15	sm	I guess they can just all follow a consistent heading style and it will work
06:15	twb	Yep
06:24		nwf joined #darcs
06:26	sm	twb: I can see this is a big job
06:26	sm	a question, why move the .tex files to .txt, rather than just leaving them where they are ?
06:30	sm	yesss.. sp rebuilding darcs when I edit any module
06:31	sm	now, how to deploy..
06:32		clanehin joined #darcs
06:33	sm	yessss.. works in emacs compilation mode
06:33	sm	mine, anyway
06:57	sm	argh.. 3 minutes of sprint left here
06:58		balor joined #darcs
07:10	sm	patch sent. What's a few minutes between friends, clocks are so imprecise
07:12	sm	end of sprint, here. I don't know if we have anyone further west..
07:18	sm	night all
07:19	* sm	dims the lights and tiptoes out
07:26	mornfall	Hm.
07:26	mornfall	lispy: I think you want cabalised-2, not cabalised-morn.
07:26	mornfall	Gotta go anyway... see ya.
07:29		kowey joined #darcs
07:32	kowey	morning
07:33		mmorrow joined #darcs
07:42	kowey	twb: nice work on the manual
07:42	kowey	not there :-)
08:01		njbartlett joined #darcs
08:03	sm	morning eric
08:03	sm	oops, typo - "this change will be bring"
08:03	sm	envigorating -> invigorating
08:04	sm	vigorous
08:05	sm	long-term
08:05		gh_ joined #darcs
08:06	sm	"by the default" -> by default
08:06	gh_	hi
08:07	kowey	sm: thanks!
08:08	sm	thanks
08:09		njbartlett_ joined #darcs
08:14	lispy	mornfall: ah
08:14	lispy	kowey: good morning
08:14	lispy	kowey: feeling chipper?
08:15	kowey	pleased to see all that work done :-)
08:15	C-Keen	good morning
08:15	lispy	kowey: yeah, now the hard work for you (us?) starts
08:15	kowey	lots of things to set into motion
08:15	lispy	yeah
08:16	lispy	we may have to change the pace, but we need to maintain the momentum
08:16	kowey	some of the work is scattered across different branches
08:16	lispy	I've been assuming you'll stream patches to darcs-users at a rate you deem appropriate and then delegate to authors as feedback comes in?
08:16	kowey	so need to get that polished off and also merged in
08:17	lispy	so, one thing that needs to happen is that we need to fix ghc6.6 support in the stuff we did
08:17	kowey	yep
08:17	lispy	so what are the logistics on that?
08:17	lispy	We need to be careful about diverging
08:17	kowey	right now I'm waiting for droundy to respond to the first salvo of patches
08:17	lispy	and it's going to be painful to amend or recreate all the patches
08:17	kowey	I'm thinking of fast-tracking twb's work
08:18	kowey	which may have conflicts with the sprint branch, will have to see
08:18	kowey	what needs to be done for fixing GHC 6.6?
08:19	lispy	two things, near as I can tell
08:19	lispy	1) Adjusting the LANGUAGE pragmas so that 6.6 stops throwing errors
08:19	kowey	(salut, gh_!)
08:20	lispy	2) Doing something about the possible Data.ByteString incompatibility on ghc6.6 (this may be fixed now due to cleaning up the api usages)
08:20	lispy	we have to fix (1) before we can check (2)
08:20	gh_	salut kowey :)
08:20	kowey	is this what you personally are working on?
08:21	lispy	I started on that, but it was a bad task
08:21	lispy	I could pick it up again
08:21	kowey	a bad task?
08:21	kowey	as in boring?
08:21	kowey	or not appropriate to the sprint
08:21	lispy	as in, it was wasting valuable sprint time
08:21	kowey	right
08:21	gh_	kowey: I have requested an account on code.haskell.org , but i'm interested in that darcshub project also
08:22	kowey	gh_: you mean as a user?
08:22	gh_	is there anything I could do to help ?
08:22	lispy	kowey: the other thing I looked at was using duncan's cabal file to build on windows (it works if you use HTTP)
08:22	lispy	kowey: but it's not polished and ready for general consumption
08:22	gh_	yes as an user, but also if I could help (a little)
08:22	gh_	i have tried github this weekend, and it's so simple...
08:22	kowey	gh_: do you think you would be willing to become an administrator?
08:23	lispy	kowey: no autoconf/make required with his cabal file :)
08:23	gh_	freehg.org (for mercurial) is better , because simpler
08:23	gh_	but also very slow
08:23	kowey	lispy: :-)
08:23	kowey	there is also bitbucket for hg
08:23	gh_	kowey: frankly, i'm not sure i will have time
08:24	gh_	yes, it's good also, much faster, but a little too complicated for my needs
08:24	* kowey	wonders if we need a darcs-hosting mailing list
08:24	kowey	because there are a lot of folks interested in this problem!
08:24	gh_	i really liked hreehg in fact
08:24	gh_	and if i remember well, the code of this website is available
08:25	gh_	*freehg
08:25	kowey	gh_ : one big missing piece of the puzzle is automatic account creation
08:25	gh_	http://www.selenic.com/mercuri[…]/HowToSetupFreeHg
08:25	gh_	kowey: on darcshub ?
08:25	lambdabot	Title: HowToSetupFreeHg - Mercurial
08:26	kowey	gh_: yes, we don't have any code for that... actually we have no code right now
08:26	kowey	gh_: the second missing piece is a Leader
08:26	kowey	we need somebody to say "I am officially in charge of the darcs hosting project and my mission is to deliver free darcs hosting to the world"
08:26	gh_	didn't you say last week that you had a front end ?
08:27	gh_	yes..
08:27	kowey	i had an offer to work on one
08:27	kowey	but we're waiting a bit to see how things turn out for that person
08:27	gh_	ok
08:27	kowey	right now, the project is waiting on code.haskell.org to release their admin scripts
08:27	kowey	which I presume to be easier to wait for than recreate
08:28	kowey	now you've done some research into other sites, which is great
08:28	kowey	maybe one thing you could do is to summarise those results for us
08:28	kowey	and let us know what you think are the critical features we need
08:28	* lispy	hopes he's being useful and not just spamming kowey :)
08:28	gh_	kowey: yes why not
08:33	lispy	kowey: we should try video conferencing next time at the sprint. But with the timezones maybe it's not worthwhile..
08:33	kowey	lispy: actually, I was kinda hoping we could do that
08:33	kowey	then I forgot about it
08:33	kowey	we both use macs, so we could iChat over jabber eg
08:34	lispy	aaah
08:34	lispy	I use xmeeting with David, ekiga is the linux version
08:49	C-Keen	kowey: what's the problem you see with using head/tail?
08:50	kowey	C-Keen: empty lists
08:50	lispy	> head []
08:51	lambdabot	* Exception: Prelude.head: empty list
08:51	lispy	> tail []
08:51	lambdabot	* Exception: Prelude.tail: empty list
08:51	C-Keen	ah I see
08:51	kowey	i mean, in many cases
08:51	C-Keen	(also I see a bug in my code)
08:51	kowey	you could (a) have code that checks if the list is non-empty and then uses head/tail
08:51	kowey	or (b) pattern match
08:51	kowey	and most of the time, (b) is more elegant
08:51	lispy	if you want to use head/tail it's almost always better to pattern match
08:52	lispy	kowey: good discussions, but it's almost 2am, so I will crash
08:52	kowey	good night!
08:52	lispy	maybe we should plan on having bi-monthly video chats though
08:52	lispy	just something like an hour where we brainstorm
08:53	kowey	yeah, it would be good
08:53	kowey	faces are useful :-)
08:53	kowey	and voices
08:53	lispy	if we could get #darcs in on it, at the same time that would be really cool
08:53	lispy	actually, some projects do regular irc meetings
08:53	lispy	but that can be hard with our timezones
08:54	lispy	anyway, g'night, have fun at work
08:54	lispy	C-Keen: right some good code :)
08:54	lispy	er write
08:54	C-Keen	lispy: will do!
08:54	lispy	happy hacking and all that
08:54	C-Keen	lispy: good night!
08:55	C-Keen	kowey: ah I see the issue, thanks for explaining. Also I think pattern matching will make it more readable
08:55	kowey	see why I think hacking on darcs is a great way to learn Haskell?
08:56	kowey	actually, hacking on any project, but darcs is a treasure trove of things to do :-D
08:56	C-Keen	without a real project you never learn any language imho
08:57	C-Keen	and this is my first approach to funcitonal programming so it feels a bit awkward at times
08:57	kowey	watch the changes coming in from the sprint by the way
08:57	kowey	when darcs was first written, David was still learning Haskell too
08:58	kowey	so looking at these new patches is a very good way to compare
08:58	kowey	how a smart person learning Haskell might write something
08:58	kowey	with how a smart person experienced in Haskell would write it
08:58	kowey	so we all grow together
09:01	C-Keen	:)
09:02	C-Keen	this is all really valuable advice
09:50		kowey joined #darcs
09:51		RaceCondition joined #darcs
10:14	C-Keen	@tell kowey which skripts do you want to get rewritten?
10:14	lambdabot	Consider it noted.
10:31		kowey joined #darcs
10:33		balor joined #darcs
10:44		kowey joined #darcs
10:54	kowey	C-Keen have a look at tests/shell_harness
10:55	C-Keen	alright
11:05		njbartlett joined #darcs
11:15		njbartlett_ joined #darcs
11:20		kowey joined #darcs
11:23		malcolmw joined #darcs
11:33		njbartlett__ joined #darcs
11:48		njbartlett joined #darcs
12:34		njbartlett joined #darcs
12:54		lelit joined #darcs
13:00		RaceCondition joined #darcs
16:12		thorkilnaur_ joined #darcs
17:29		arjanb joined #darcs
17:32	kowey	arjanb: thanks for checking that windows binary last night, btw
17:32	kowey	i think those are openssl libs, and I would have to work out a way to link something statically :-)
17:32	kowey	(which is why I would have really preferred somebody else, i.e. who actually uses Windows, build this binary)
17:38		sm joined #darcs
17:40		MarcWebe3 joined #darcs
17:49		sm joined #darcs
17:58	arjanb	kowey: btw the binary works after gathering all the missing dll's
18:01	kowey	cool!
18:01	kowey	are there a lot?
18:02		beschmi joined #darcs
18:02	* beschmi	waves
18:04	kowey	hiya beschmi! exciting news on the filecache :-)
18:05	kowey	ok, first 4 patches from the sprint now applied in droundy's branch
18:05	beschmi	i was actually a bit disappointed because it's still too slow on really large repos like ghc
18:05	beschmi	because you still have to apply a large number number of patches
18:05	kowey	faster, but not fast enough?
18:05	beschmi	but otoh, chunky patches should speed that up too ;)
18:06	kowey	do you think there is anything useful for you in http://code.haskell.org/darcs/zoo ?
18:06	lambdabot	Title: Index of /darcs/zoo
18:06	kowey	for benchmarking that is
18:06	arjanb	kowey: 2 of ssl and a ms c runtime one
18:06	kowey	there is zero visualisation right now, but I hope we can follow Igloo's advice to make a shootout-like graph
18:06	kowey	thanks
18:07	kowey	maybe I can tell libcurl on windows to link statically with openssl
18:07	kowey	and that in turn would remove the need for that dll
18:07	Igloo	I think it's probably easy to do once you've picked the best tool, but I have no idea what that tool is
18:08	kowey	(on the other hand, there is also that whole illegal-to-distribute thing with openssl)
18:10	Igloo	Can you use gnutls instead? (I assume that's (L)GPL, but I don't actually know)
18:10	beschmi	kowey: yeah, looks nice. For now I'm doing mainly manual tests and I should work on having a well defined set of benchmarks.
18:10	kowey	it's really super-primitive, but if you could try using it and improving it, great!
18:10	kowey	basically we need to make this testing less ad-hoc and more reproducible
18:11	kowey	(not necessarily the numbers, just the process)
18:11	kowey	note that the small tests currently use the darcs version of maybench
18:12		gh_ joined #darcs
18:13		Nasenbaer joined #darcs
18:13		Nasenbaer left #darcs
18:20	sm	morning all
18:20	sm	great kowey, thanks for working on that stuff
18:21	kowey	all that stuff?! I just spent the sprint making tea and being scatter-brained :-)
18:22	sm	ha, not true
18:22	kowey	good thing the rest of us rolled up sleeves and got productive
18:22	dons	kowey: so droundy's branch is the head branch?
18:22	kowey	i meant his local darcs repository on his machine, before he pushes it to his staging branch on his fast machine (which then automatically pushes to unstable)
18:22	sm	somebody recommended HaskellCharts to me this morning, but otherwise gnuplot seems to be the standard tool for automated chart generation. Bit of a learning curve though
18:22	kowey	i.e. "soon-to-be-unstable"
18:23	* sm	votes HaskellCharts for more haskell community-darcs synergy
18:23	kowey	dons: so, was just speaking very loosely
18:23	dons	sm, there's also a haskell gnuplot binding, and a haskell googlecharts binding
18:23	sm	aha!
18:23	kowey	and now they're in unstable
18:23	dons	http://hackage.haskell.org/cgi[…]pts/package/Chart
18:23	sm	very nice to have choices
18:24	lambdabot	Title: HackageDB: Chart-0.8
18:24	dons	http://hackage.haskell.org/cgi[…]/package/diagrams
18:24	lambdabot	Title: HackageDB: diagrams-0.1, http://tinyurl.com/6omkcp
18:24	dons	http://hackage.haskell.org/cgi[…]s/package/gnuplot
18:24	lambdabot	Title: HackageDB: gnuplot-0.2, http://tinyurl.com/5srmed
18:24	dons	http://hackage.haskell.org/cgi[…]ckage/GoogleChart
18:24	lambdabot	Title: HackageDB: GoogleChart-0.2, http://tinyurl.com/68okj9
18:25	sm	and not forgetting hsparklines
18:25	dons	mm. yes.
18:25	dons	that might also work
18:26	kowey	would serve as a good preview if there are a lot of tests
18:26	kowey	"whoah lots of ups and not a lot of downs there!"
18:52	beschmi	kowey: i debugged it some more ans the bottleneck doesn't seem to be applying the patches anymore, it's reading the inventory
18:54	beschmi	no, it's reading the pristine
19:08	dons	lispy: photos for your portland post, http://galois.com/~dons/images/darcs-oct-08/
19:08	lambdabot	Title: Index of /~dons/images/darcs-oct-08
19:16		RaceCondition joined #darcs
19:26	lispy	kowey: we sent in some of the sprint patches right?
19:27	kowey	lispy: hmm?
19:27	kowey	yeah, four salvos (although salvos 1 and 2 overlapping)
19:28	kowey	salvo 1 has been 2/3rds applied (the other third awaiting a brief explanation)
19:28	kowey	also sm's patches have been applied
19:28	kowey	but not yet pushed
19:28	lispy	hmm
19:28	lispy	I'm confused
19:28	lispy	Has david commented on any of the sprint stuff?
19:29	Igloo	Yes
19:29	kowey	just on the first salvo - http://lists.osuosl.org/piperm[…]tober/015112.html
19:29	lambdabot	Title: [darcs-users] darcs patch: resolve issue395: warn the user when the... (and 5 mo ..., http://tinyurl.com/6hjq63
19:29	lispy	okay. so that's in our count?
19:29	lispy	court*
19:30	lispy	why wasn't issue1017 fixed
19:30	* Heffalump	appears
19:30	kowey	ah just the right person
19:30	dons	how many patches were produced during the sprint?
19:30	Heffalump	I'll reply to that email
19:31	kowey	just on the official sprint branch, 99, I think (darcs send --dry-run \| grep @ \| wc -l)
19:31	dons	lispy: will you stick the summary of what you did on a blog post, and maybe attach the photos?
19:31	kowey	but there's more patches in sprint-slurpy and in beschmi's branch
19:31	dons	or maybe there should be a single sprint summary
19:31	dons	with all photos and details
19:32	kowey	not to mention the documentation team and kirby81's Windows fixes
19:32	dons	kowey: cool.
19:32	lispy	dons: okay, I'll try to do something or coordinate with kowey
19:32	sm	need a sprint darcswatch or two eh
19:33	lispy	a good patch tracking system would be nice yeah
19:33		b4taylor left #darcs
19:34	lispy	kowey: of the sent sprint patches, 1 bundle has been reviewed (some rejected), 1 bundle is unreviewed, and 2 bundles are unsent?
19:34	lispy	dons: I was reading the xmonad archives last night, good stuff
19:34	dons	nomeata wrote a darcs patch tracker for mailing lists.
19:34	dons	it tells us what was sent to the xmonad mailing list, but not applied yet
19:35	dons	he's published it, afaik
19:36	lispy	yeah, I don't think it's all that useful yet though
19:36	kowey	lispy: 1 reviewed (some pending inquiry), 3 unreviewed, and many patches not yet sent
19:36	lispy	kowey: ah
19:37	dons	yes, i wonder if direct pushes to branches, in a tree-like structure are more effective
19:37	dons	so things don't get lost
19:37	kowey	lispy: only 95/99 patches to go!
19:37	kowey	:-)
19:37	lispy	haha
19:37	dons	N people push to lispys branch, M to kowey's branch, they review, then merge up
19:38	kowey	(there's also 80 patches in the documentation branch to go)
19:38	Heffalump	it would be good to raise the bytestring question asap
19:38	lispy	dons: my thoughts on nometa's is just that it doesn't track discussion or status of patches, only whether they were accepted or in limbo
19:38	dons	note the increase in mailing list activity, guys, http://lists.osuosl.org/pipermail/darcs-users/
19:38	lambdabot	Title: The darcs-users Archives
19:38	dons	as active as it has been since 2003...
19:39	dons	2005.
19:39	lispy	yeah
19:39	lispy	as active as ever really
19:39	dons	worth mentioning.
19:39	lispy	true
19:39	kowey	although to be fair, this is with merged lists
19:39	dons	ah good point.
19:40	lispy	not much non-devel discussion going on though
19:40	lispy	just an email here and there
19:40	kowey	like the darcsweb question just now
19:40	dons	that's fine. works for xmonad.
19:40	dons	merged lists == more users become devs
19:40	Igloo	Size of patches and patch context is also a factor, if you're looking at mailbox size
19:40	dons	ah good piont
19:40	lispy	and compression ratio :)
19:41	kowey	i have some numbers on the number of messages in the past few months
19:41	lispy	But! it is true that darcs-users is very active
19:41	lispy	and that the development rate hasn't been this high for a long time
19:41	Igloo	Number of messages and number of unique authors would be interesting
19:41	dons	messages and unique contributors would be useful numbers
19:41	dons	also irc stats..
19:41	* kowey	digs around in his mail
19:41	dons	http://xmonad.org/statistics.html
19:41	lambdabot	Title: xmonad : project statistics
19:43	lispy	dons: can you make the scripts/instructions for generating that stuff available?
19:43	kowey	http://hpaste.org/11520
19:43	C-Keen	hm, how can I _set_ environment variables within haskell?
19:43	dons	:hoogle setEnv
19:43	dons	?hoogle setEnv
19:43	lambdabot	No results found
19:43	dons	mm
19:43	C-Keen	I tried that
19:43	C-Keen	and google
19:44	C-Keen	but I always get referred to System.Environment but that is for reading only
19:44	dons	oh, silly hoogle
19:44	dons	doesn't know about the unix package
19:44	dons	for political reasons
19:44	dons	:t System.Posix.Env.setEnv
19:44	lambdabot	String -> String -> Bool -> IO ()
19:44	kowey	note that 'patch' refers to messages that have 'darcs patch:' in their title
19:44	C-Keen	thanks dons
19:44	lispy	C-Keen: I already asked these questions (I assume your trying to convert shell_harness)
19:45	lispy	C-Keen: the way to di it is on the mailing list :)
19:45	lispy	C-Keen: you'l want to avoid Posix so that the harness can work on windows
19:45	C-Keen	lispy: alright. Any particular reason why you did not succeed other than lack of time?
19:45	kowey	although having it working on unix is already a very good start
19:46	lispy	C-Keen: http://lists.osuosl.org/piperm[…]tober/014881.html
19:46	lambdabot	Title: [darcs-users] Help converting Perl to Haskell, http://tinyurl.com/5z2fgl
19:46	lispy	C-Keen: lack of time yeah
19:47	C-Keen	lispy: I will have a try
19:47	lispy	C-Keen: on windows you fetch the env via getEnv in System.Environment
19:48	lispy	C-Keen: then you append the env you want to use to that and pass the whole thing on to runProcess
19:49	C-Keen	lispy: ok, I will do a unix version first, as I don't have a windows system to test on
19:49	beschmi	kowey: I did some test related to http://bugs.darcs.net/issue1068
19:49	lambdabot	Title: Issue 1068: annotate stack overflow - Darcs bug tracker
19:49	beschmi	http://hpaste.org/11523
19:50	* sm	notes that gmane's stats http://dir.gmane.org/gmane.com[…]ontrol.darcs.user are stale, someone should bug them
19:50	lambdabot	Title: Gmane -- Mail To News And Back Again
19:51	kowey	beschmi: are those two links related? (i do like the idea of being able to do darcs changes in 5 seconds, and not 2 minutes, though)
19:52	lispy	C-Keen: well, the functions you'll want to use on windows will work on unix too. So even if you can't test it, you can write it in an anticipating way (you only have to write 1 new function that is pretty simple)
19:52	kowey	beschmi: any idea why a stack overflow in particular?
19:52	C-Keen	ok
19:53	beschmi	it's the same file, Annotate shouldn't take much more time but I haven't changed it yet.
19:54	beschmi	kowey: the code will change a lot, so I wont spend much time changing stack overflows in old code
19:54	kowey	ok!
19:54	kowey	:-)
19:55	kowey	Heffalump: did you say you fixed your sprint-slurpy bugs?
19:55	beschmi	when I filtered the patchset for annotate, annotate also finished in 8M for me
19:55	kowey	compared to... never?
19:56	* kowey	thinks 8 minutes is a terrific alternative to never
19:56	kowey	he says vapidly
19:56	beschmi	I meant with stack set to 8 Megabyte
19:56	kowey	hehe
19:56	* sm	pokes gmane.discuss
19:57	kowey	oh, I see now, rereading the bugtracker
19:58		sm left #darcs
19:58		sm joined #darcs
20:04	lispy	and so what is left wrt to bytestring? We need to see if our bytestring can work on 6.6?
20:05	dons	yes, tidy things up so the parts of bytestring we use are also usable with 6.6's bytestring
20:05	dons	api should be much the same, modulo the move of two modules (Internal/Unsafe from Base)
20:05	lispy	so the first hurdle there is our LANGUAGE usage
20:05	dons	and finishing off that.
20:05	lispy	dons: iirc we fixed that
20:05	lispy	we removed the unsafe/internal
20:05	lispy	I thought
20:06	dons	all uses?
20:06	dons	we removed some. i'm not sure we removed all.
20:06	lispy	I think so
20:06	dons	that'd be nifty
20:06	dons	there's not that much of bytestring actually used
20:06	dons	so it seems doable
20:07	beschmi	anyone here familiar with Repository.Internal?
20:09	Heffalump	beschmi: not familiar, but I've looked at it a bit
20:10	beschmi	darcs changes file spends a lot of time checking if file has been renamed in unrecorded patches
20:11	beschmi	it calls "get_unrecorded_unsorted repository", but I think it should suffice to look into pending instead of checking the complete working directory
20:12		InkBottle joined #darcs
20:12	Heffalump	renames need to be explicitly signalled, right? So that sounds right to me.
20:14	beschmi	yes, I'll have to look at apply_to_filepaths which is used like "files' = apply_to_filepaths (invert unrec) files" to be sure
20:14		mcxx joined #darcs
20:15	beschmi	I'm wondering if there is a comparable function to get_unrecorded_unsorted that just returns the patches in pending
20:17		dcoutts joined #darcs
20:17	beschmi	read_pending seems to be right
20:18	C-Keen	beschmi: there is
20:19	C-Keen	beschmi: yes read_pending it is
20:19	* C-Keen	has already spend a while on Repository.Internal today
20:19	beschmi	ok, so I just have to handle the Sealed and I'll be fine. Thanks.
20:21	beschmi	C-Keen: ah, what have you been looking at?
20:21	C-Keen	beschmi: maybe you want to use slurp_pending?
20:21	C-Keen	beschmi: just printing a warning when you did darcs whatsnew on nonexistant or nonrecorded files
20:22	C-Keen	beschmi: but I needed a way to tell whether I know that FilePath already
20:22	beschmi	i see.
20:23	C-Keen	(nothing that interesting)
20:25	beschmi	C-Keen: unsafeUnseal `fmap` read_pending seems to fit the bill here.
20:25	C-Keen	beschmi: unsafeUnseal?
20:25	C-Keen	beschmi: and here: in your case?
20:27	beschmi	yes, in my case. fits the bill as in: it compiles and passes simple tests.
20:30	C-Keen	:)
20:30	C-Keen	good night
20:34		InkBottle joined #darcs
20:40	dons	anyone looked at haskeline?
20:40	dons	i can't build it, and it seems to require lots of things
20:40	dons	surely something like editline would be a simpler alternative? (bundled with ghc , across platforms)
20:40	dons	Igloo: esp. if editline ends up on hackage...
20:41	kowey	maybe Judah hangs out on IRC and can comment?
20:41	Heffalump	would someone else mind testing out issue1017 on latest darcs?
20:41	Heffalump	latest unstable that is
20:42	kowey	just pulling the two patches, right?
20:43	Igloo	Just one patch
20:43	Heffalump	it should be fixed with whatever is currently in the unstable tree
20:43	Igloo	Not the 20char one
20:43	kowey	oh
20:43	kowey	ok, I'll just pull the mv patch and not the 20char one
20:44	Igloo	I wonder if David test the issue in a compiler without the other patches he was pulling
20:44	Igloo	What did the actual issue closing patch do? Just add something to a changelog or something?
20:44	Heffalump	move the test over from bugs to tests
20:45	Igloo	OK, so he might have tested in his current compiler as he was deciding which patches from teh bundle to pull
20:45	Heffalump	ah, right
20:46	Heffalump	annoyingly, my slurpy stuff depends on don's LANGUAGE pragmas
20:46	Heffalump	if David rejects stuff low down the dependency tree of the sprint work, we're going to have a problem.
20:46	Igloo	No, you just have to implement darcs transplant :-)
20:47	Heffalump	that's a hack that still causes dependency problems
20:47	dons	i think it is unhealthy for all the hackathon workers to be worrying about whether the president will veto our work
20:47	Heffalump	well, not dependency problems, problems with different people having different patches
20:47	Igloo	It doesn't cause dependency problems if it does what I think it does
20:47	dons	it makes it scary/hard to contribute
20:47	Igloo	Right, agreed, but people who have patches from non-official repos need to expect that sort of problem
20:47	dons	and not how open source should be. so we either need to ignore this stressor, or do somehting else.
20:47	Heffalump	dons: s/David rejects stuff/there's a problem with stuff/ if you like
20:48	C-Keen	I wonder whether we should use one of the available haskel shells for scripting... I feel like reinventing the wheel with writing a backtick and "which" replacement
20:48	dons	i just had to put out of my mind the idea that things would be rejected, otherwise i could do no work
20:48	dons	makes me feel sick about the project, if hours and hours of new contributor work would be rejected.
20:48	dons	so... sigh.
20:49	Igloo	Rejecting things isn't necessarily bad. It's just particularly painful at something like a sprint, when lots of entangled patches are generated before early ones have a chance to be rejected
20:49	Heffalump	Igloo: right.
20:49	kowey	folks, please give us a week or so to let the dust settle
20:49	dons	biggest challenge facing darcs => working out how to make decisions when the key veto power is held by someone not involved in those decisoins
20:49	kowey	Heffalump: passes
20:49	Heffalump	kowey: ok, thanks
20:49	Heffalump	let's hope Igloo's explanation is right
20:50	kowey	well one thing is that we need to increase david's involvement in key decisions
20:50	kowey	by doing a better job reporting back to him
20:50	Heffalump	dons: I have to say, that while that kind of attitude is great for getting stuff done, it's also the kind of thing that causes the kind of package dependency mess we now have with GHC
20:51	kowey	or by working on increasing the amount of delegation of responsibility
20:51	Heffalump	kowey: except that with sprints, unless he's around, we're stuffed, for the reason Igloo says
20:51	kowey	ok, sprints is a technical problem (a nice one to have)
20:52	kowey	and we do need to find a way to get these patches in, I agree!
20:52	Heffalump	I think we just have to insist that David accepts a slightly messy patch history (i.e. a manual rollback of the ones he doesn't want, if ones he does want depend on them)
20:52	dons	its a broken project management model currently, we must admit. either david reduces his influence (to be just patch theory reviewer) , or he participates in the decisions
20:52	kowey	Heffalump, yes
20:52	Heffalump	He does participate in the decisions.
20:52	dons	after the fact.
20:52	kowey	no amending of sprint patches, too difficult
20:53	Heffalump	so your complaint is that he wasn't available over the weekend.
20:53	dons	someone will have to explain, e.g. why bytestring support was added, or cabal support
20:53	Igloo	You could have had a discussion before the sprint to get an agreement in principle
20:53	Heffalump	dons: yes, that needs to be explained in any case.
20:53	Igloo	The details may still need to be amended, which makes reviewing a pain as you can't amend the patches, but it would be better than having things rejected afterwards (if that's what happens)
20:54	dons	we burn a lot of cycles trying to massage things through david's filter.
20:54	Heffalump	Because his goals differ from your cavalier "just do it" approach.
20:54	Heffalump	Igloo: I don't understand your last point.
20:54	dons	sure. and you can see it in the code base.
20:54	* kowey	likes the very agile approach, but would like to have David's conservatism around as a countervailing force :-)
20:55	Igloo	dons: But taking things slowly can have advantages, e.g. darcs' move to bytestring would have been easier if bytestring hadn't been put into base too early
20:55	kowey	but we just have to find a way to keep things healthy
20:55	dons	its very old-feeling. and effort is spent on weird projects, like franchise
20:55	Heffalump	Nobody who doesn't want to is spending effort on franchise.
20:55	Igloo	Heffalump: Not important
20:55	mornfall	Heffalump: You are wrong here.
20:55	dons	conservatism is key. xmonad has the same requirements.
20:56	Heffalump	xmonad is much newer and smaller
20:56	dons	but we still encourage refactoring, documentation and clean up patches. and the decision authortiy people participate in dicussions.
20:56	dons	Heffalump: not much smaller now. about 10k loc
20:56	Heffalump	and the maintainers happen to be much more available than David is
20:56	Heffalump	that's a factor of 3 or so
20:56	dons	but unlike darcs, the 'extensions' are not in the core.
20:56	dons	so its easier to review thiiiings
20:56		balor joined #darcs
20:57	kowey	darcs really does need to move towards being more modular
20:57	Heffalump	and it happens to be much easier to test, and much less intricate in what it actually does
20:57	kowey	we all accept this (david too)
20:57	dons	no experiments in the core. (while things like oh, libwww or sha256 v sha1, are in the core here, making it hard to clean up)
20:57	dons	well, i hope the bytestring stuff makes it in, since then i'd be inspired to clean up more things.
20:57	Heffalump	darcs started many years before xmonad
20:57	mornfall	kowey: Right, but he makes it hard to get through patches making things more modular.
20:57	dons	there's photos of all the modernisation that we thought was needed on the whiteboards
20:57	Heffalump	We'd all like to see the bytestring stuff in, I think.
20:57	Igloo	I think a rewrite is the best way forward, which is partly why camp exists
20:58	Heffalump	But your attitude that David is a roadblock is extremely counterproductive, IMO.
20:58	dons	i think more of the problem is not that he'd reject things, just that he's not around
20:58	dons	so we spin our wheels wondering if this work is useful or not
20:58	Heffalump	We should fit our processes around the fact that people have lives outside Haskell.
20:58	dons	Heffalump: i'm not sure he is a roadblock. just the position of a) holding veto power, and b) not being around, means we do all this work under fear of rejection
20:58	Igloo	dons: Did you send him an e-mail that he didn't reply to?
20:58	dons	which is the counterproductive thing
20:59	mornfall	Heffalump: Exactly, and that's why having a single bottleneck like we do is making the project uninteresting to contribute to.
20:59	kowey	well the issue according to david
20:59	dons	Igloo: no, so he might well have replied to stuff.
20:59	mornfall	Heffalump: Because we don't have 8 hours a day to sink in it...
20:59		RaceCondition joined #darcs
20:59	Heffalump	The cost of having multiple people pulling the project in different directions would also be high.
20:59	dons	uncertainty is what worries me.
20:59	kowey	is that darcs is still easy enough to break and there still are not enough people who understand darcs well enough for him to give up the veto power
21:00	kowey	(again, this is the conservatism)
21:00	kowey	I agree that david's using the veto on non-core issues (build systems, versions of GHC to support, etc) is not very productive
21:00	mornfall	kowey: The fear of breaking it makes things worse.
21:00	Heffalump	from some points of view, that is a core issue
21:00	dons	and there's a lot of uncertainty left over from the darcs 2.0 release, and other comments about not reading mailing lists and so on. so that's casts a pall over contributions for new people.
21:00	mornfall	kowey: If someones breaks it in a way that goes through the suite, we at least know that suite needs to be improved, and how.
21:01	mornfall	kowey: If we keep not doing things because they might break something and there's noone who knows whether they are fine or not, we won't help anything.
21:01	kowey	mornfall: subtle breakage, like the pending patch bug that bit darcs 1.0.9
21:01	kowey	and that we only found out about after darcs 2.0
21:01	mornfall	kowey: The core is crufty, opaque and bitrotting.
21:01	dons	the expected standards for contributions are also a bit ill-defined , imo. (purely a process thing)
21:01	dons	we could start requiring all code to be commented...
21:01	Heffalump	I don't find the code that opaque.
21:01	mornfall	kowey: And because we don't want to change it due to breakage-fear, that's not going to change.
21:01	Heffalump	It's somewhat crufty, but not that bad.
21:01	mornfall	Catch 22, congrats.
21:02	Heffalump	Actually I think it's perfectly possible to clean up the cruft, carefully.
21:02	mornfall	Heffalump: Then where all those difficult, elusive bugs come from?
21:02	Heffalump	the fact that darcs is solving a difficult problem
21:02	kowey	those aren't code issues
21:02	mornfall	Heffalump: Well, not really.
21:02	dons	Heffalump: the number of times duncan and i asked lispy "what the hell is this?" was funny. :) mostly from lack of comments, or lack of standard idioms. so i'd hope we could get some style gudies
21:02	dons	basic contribution standards
21:02	mornfall	Heffalump: Darcs is making it a difficult problem.
21:02	mornfall	That's the issue.
21:03	kowey	ok, dons, can we sum up what you're saying so far?
21:03	dons	http://code.haskell.org/xmonad/STYLE
21:03	kowey	(1) we need more documentation about our processes
21:03	dons	a style guide that was required, and applied to the existing code, would help contributions
21:03	Heffalump	I'm not yet all that familiar with the code, but I haven't found it that hard to understand. Invariants certainly need better documentation.
21:03	kowey	e.g. coding sytle
21:03	Heffalump	But that's true of most codebases.
21:04	kowey	(2) we need David to either be around more for decision making or to leave more decisions up to the community
21:04	Heffalump	Does David tend to reject patches on style grounds?
21:04	dons	kowey: 2) the contribution process , in its current form, needs to be documented.
21:04	kowey	did I miss or mischaracterise anything?
21:04	kowey	dons: so "How to submit a patch to darcs, and what to expect when you do so?"
21:04	dons	yes.
21:04	dons	http://haskell.org/haskellwiki[…]elopment_tutorial
21:05	lambdabot	Title: Xmonad/xmonad development tutorial - HaskellWiki
21:05	dons	how to contribute to xmonad.
21:05	dons	coding style requirements, http://xmonad.org/xmonad-docs/[…]Developing.html#8
21:05	kowey	ok, that sounds like something concrete we can do
21:05	lambdabot	Title: XMonad.Doc.Developing, http://tinyurl.com/58mu8n
21:05	dons	we're getting there.
21:05	dons	the sprint group is definitely modernising.
21:05	kowey	mornfall, dons, I really really really appreciate this sort of discussion
21:05	dons	so if those patches make it in, a lot of the uncertainty of the current political structure will be removed
21:06	kowey	but I hope you can understand that change can be a slow process that requires a lot of patient work
21:06	dons	yes.
21:06	dons	more than any project i've ever been involved in :)
21:06	mornfall	Well, let me know when things change, I'm too busy in the meantime. : - \
21:06	kowey	and the reason it's important that we have these discussions is to have a clear picture what we want to change INTO
21:06	kowey	just know that it won't happen overnight
21:07	dons	i'm fine for the troika (well, what's a 2-headed troika called?) of kowey and lispy driving things. makes it so so much easier
21:07	mornfall	(I will still consider contributing to darcs, as opposed to spinning off, although at some point I have been sort of decided to do the latter...)
21:07	kowey	mornfall: well you should work on Camp so we have 2 projects and not 3 :-)
21:08	mornfall	dons: Двойка maybe? : - )
21:08	mornfall	(That would be dvoika, transliterated.)
21:09	dons	i think we're on the verge here of opening up darcs a lot.
21:09	mornfall	kowey: As I have said before, there are fundamental disagreements on the theory, or so I think at least.
21:09	dons	one or two more sprints, a few more announcements, a more accessible code base (i.e. haddockised on hackage!)
21:09	dons	would help a lot
21:09	kowey	the kowey agenda is one of consensus building and gradual reform
21:09	kowey	and also building momentum by fixing technical issues
21:09	dons	and this thing could be revived. so i'm keen to see this happen
21:09	kowey	like the lack of a contributor's guide and darcs transplant
21:09	Heffalump	dons: hackage/cabal is not the be-all and end-all of a project
21:10	mornfall	kowey: I think darcs transplant is more of a problem than a solution.
21:10	kowey	ok so I'm relieved we agree that there is progress at least
21:10	dons	Heffalump: not suggesting it is, but it helps a lot attracting new devs.
21:10	mornfall	kowey: Given that darcs will break horribly when transplanted and original repos get mixed...
21:10	dons	and we're trying to do that aren't we?
21:10	Heffalump	I'm not convinced about that.
21:10	Heffalump	autoconf is hardly scary for a developer
21:11	dons	?
21:11	mornfall	Heffalump: Is it?
21:11	kowey	ah, I still think of transplant as a disposable operation
21:11	kowey	but mornfall, let's leave that discussion out
21:11	Heffalump	mornfall: it's painful to change, not to use
21:11	kowey	the point is that there is a LOT of work to do
21:11	dons	autoconf is fine. autoconf + cabal is fine. documentation is better. online source docs are great!
21:11	mornfall	Heffalump: It's painful to use when it disagrees with you.
21:11	kowey	and only some of the work is controversial
21:11	Heffalump	and cabal is still a crappy build tool
21:11	dons	kowey: and people are keen to do it, as long as things get in.
21:11	kowey	there is so much stuff we can do that everybody can agree on
21:11	kowey	stuff we can do that is easy to get in
21:12	kowey	and what my aim is for us to do as much of that stuff as possible first
21:12	dons	Heffalump: did you see the cabal stuff doesn't require autoconf now? :)
21:12	kowey	one reason being that this allows david to be exposed to newcomers more
21:12	kowey	and also to build up trust in the community
21:12	dons	so, say i wanted to write some patches today, where do they go?
21:12	Heffalump	dons: so? It does that at the cost of being less portable (Windows excepted). Which is great for building distributions etc, but we should be aware of its limitations.
21:12	dons	the sprint repo is now read only?
21:12	kowey	still open
21:12	dons	or how do we keep working on that branch?
21:13	Heffalump	I would think briefly about the most appropriate branch to work on, then work on it.
21:13	Heffalump	If it's a follow-on to bytestring, do it on sprint.
21:13	dons	ok.
21:13	Heffalump	If not, and it won't conflict, then do it on a fresh unstable.
21:13	dons	and then push to sprint, which really means 'push to kowey's branch' ?
21:13	dons	which will then flow through to the head ?
21:13	kowey	that's one way to look at it
21:13	Heffalump	I think you have to take charge of getting your patches accepted.
21:14	Heffalump	Rather than dumping it all on kowey indefinitely.
21:14	dons	i just don't know where they go though.
21:14	dons	i'd like to build on the weekend work.
21:14	dons	ah well. either unstable or sprint.
21:14	kowey	dons: easy answer: please keeping pushing to the sprint branch
21:14	Heffalump	Has someone sent an email to David about that work yet?
21:14	kowey	no
21:14	dons	lispy summarised portland's stuff on the list.
21:14	Heffalump	It was a rhetorical question :-)
21:14	dons	but it's kowey's job, iirc?
21:14	Heffalump	we need to get his buy in for it, and explain the 6.6 strategy.
21:15	dons	right. so revisit those discussions.
21:15	kowey	i was waiting for him to polish off his current backlog, for what it's worth
21:15	Heffalump	I think kowey might have kindly volunteered to help out. It's not his "job" per se.
21:15	kowey	but in the meantime we should be drafting
21:16	dons	whats the 6.6 strategy to be?
21:16	dons	we /must/ continue supporting 6.6? or do we hope for the sprint code not to be released until support for 6.6 is dropped?
21:16	dons	if the former, then i'll install 6.6 and make the bytestring stuff work with it.
21:16	dons	if the latter, we continue as is.
21:16	kowey	the easiest thing we can do, imho is just to support 6.6
21:16	Heffalump	well, easiest from the point of view of acceptance
21:17	dons	i hope not too many people use 6.6-compiled darcs
21:17	Heffalump	harder from the point of view of extra work
21:17	kowey	because this allows us to separate the question of accepting these patches from that of dropping 6.6
21:17	kowey	both things we want to achieve
21:17	Heffalump	how easy would just importing bytestring into the sources conditionally be?
21:17	dons	Heffalump: i think we can avoid that.
21:17	dons	Heffalump: since 6.6 comes with bytestring 0.9
21:17	dons	and we use very little of the api
21:17	Heffalump	what about withForeignPtr?
21:17	dons	and you happen to have a bytestring author here, who can make it work.
21:18	Heffalump	with how much effort?
21:18	dons	that's fine. we could just CPP our own.
21:18	dons	its a one liner.
21:18	kowey	ok, I'm not married to the idea of absolutely making sure it works on 6.6, but we should at the very least see if it is feasible to support 6.6 without much effort
21:18	dons	i think it is not too much work.
21:18	Heffalump	In that case I think you should do it.
21:18	dons	by doing all the refactoring, we reduced the code surface area
21:18	dons	making it easier to support 6.6
21:18	dons	but until we decided to toss things out, it was intractable
21:18	dons	too much old code to support
21:18	Heffalump	great, then
21:18	dons	so yay, remove more code and 6.6 will continue to be easier.
21:19	dons	get rid of some of the alternative crypto and http thingies, and we'll have even easier code
21:19	kowey	where's our Fitness Guru? gwern would be happy to hear this
21:19	Heffalump	which http thingies do you want to throw away?
21:19	dons	smaller code helps accessibility and reduces the chance for people duplicating existing functions, too
21:19	kowey	dons: do you know about http://wiki.darcs.net/index.html/DarcsLibraries ?
21:19	lambdabot	Title: DarcsLibraries - DarcsWiki
21:20	dons	there's a lot of similar functions quite alike, sprinked around for various purposes
21:20	kowey	this was one of the first pages created after the GHC move
21:20	dons	Heffalump: perhaps we could just keep libcurl support?
21:20	dons	just choose one method.
21:20	Heffalump	and how do we ensure it's available everywhere?
21:20	mornfall	Well, I'd actually vote for the bytestring mini-http package. But libcurl is apparently much more real-world friendly, so that probably makes more sense.
21:21	dons	and mini-http doesn't exist.
21:21	mornfall	network-minihttp
21:21	dons	removing old experiments from the source seems like a profitable , simple task, anyway
21:21	dons	mornfall: where's that live?
21:22	Heffalump	what old experiments?
21:22	mornfall	At least I have a tar of that under my .cabal/packages...
21:22	Heffalump	You both seem to be ignoring the fact that the current HTTP choices are an aid to portability. How will you replace that if you just pick one?
21:22	kowey	dons: http://bugs.darcs.net/issue1178
21:22	lambdabot	Title: Issue 1178: task: write up tutorial on how to contribute to darcs - Darcs bug tr ...
21:22	dons	Heffalump: we seem to have what, 3 network interfaces? 2 sha implementations?
21:23	dons	how many other conditionally compiled things?
21:23	Heffalump	yes, because of the performance/portability tradeoff
21:23	dons	but which ones are used in practice? since most people use the distro packages, which just compile with the defaults?
21:23	dons	are we supporting code used by only a handful of people?
21:23	Heffalump	what about on Windows?
21:23	mornfall	Heffalump: CURL is used on windows.
21:23	mornfall	Heffalump: And CURL is used also everywhere else.
21:23	Heffalump	with pain, yes
21:24	mornfall	Heffalump: Maybe. But still, noone seems to be really using the alternatives.
21:24	dons	the darcs.cabal file is good for seeing all the conditionally compiled variants of darcs
21:24	dons	who's using haskeline?
21:24	mornfall	Heffalump: So it probably makes sense to have one with less pain than 3 that noone uses and one painful.
21:24	Heffalump	how do you know that noone uses them?
21:24	kowey	haskeline addresses some requests for readline type features, I thought
21:25	dons	we can support N variants of darcs, but it is a drain on developers, and makes the code more complex. the project devs know which variants are best, so we should just pick one.
21:25	mornfall	dons: Fully agreed.
21:25	Heffalump	You seem to know which one is best. I say there is no one best choice.
21:26	mornfall	Heffalump: Because all HTTP options are inferior to cURL?
21:26	dons	color/type-witnesses/curses/terminfo/haskeline/external-zlib/external-bytestring/http/libwww/curl-pipelining/curl
21:26	dons	Heffalump: so are there N variants? or is there only 2? one for unix (curl) and one for windows (?)
21:26	dons	in practice, what binary of darcs do people use?
21:26	dons	just the default?
21:26	Heffalump	I don't know.
21:26	kowey	strike type-witnesses and curl-pipelining from that list
21:26	mornfall	dons: There's at least HTTP, cURL and libwww.
21:26	Heffalump	For Windows, there is no good default now.
21:27	Heffalump	Mainly because it's hard to build curl, for example.
21:27	dons	fwiw, galois wrote their curl binding becaues it was good for windows.
21:27	mornfall	Heffalump: But having 3 bad choices is about the same as having one bad choice.
21:27	Heffalump	It's a trade off
21:27	Heffalump	they are bad in different ways.
21:27	mornfall	Heffalump: It's not.
21:27	Heffalump	easy installation versus performance
21:28	dons	there's an interesting tradeoff: the majority (by far) use the defaults , since they use their distro packages
21:28	Heffalump	What do Windows users use?
21:28	mornfall	Heffalump: Binaries, if they can.
21:28	dons	but we support funky variants for a handful of people who compile by hand (and who then also tweak settings)
21:28	mornfall	If not, usually nothing.
21:28	Heffalump	There's no reliable Windows build yet, because of the pain on building on Windows.
21:28	mornfall	Only a handful will install GHC, (and msys in the past) to compile their own darcs.
21:28	mornfall	You say curl is difficult, but so is darcs itself.
21:29	mornfall	Heffalump: Which could have been solved by now, but it's everything entrenched on the anti-cabal sentiments, so we don't really make that much progress as we could.
21:29	dons	no more tax on development!
21:29	mornfall	And the fact that we are so afraid of cutting down on choices.
21:30	* dons	isn't afraid. toss the garbage. toss anything that adds tricky dependencies. just keep one of each thing. make it the default.
21:30	mornfall	All those options just show that we deal with bikeshedding by including all the choices. That doesn't scale and we suffer from that.
21:31	Heffalump	Toss portability. Toss out all the users that don't do things the dons way.
21:31	dons	e.g. haskeline support shoudl go, since haskeline adds about 10 new dependencies
21:31	dons	Heffalump: no. choose one portable solution (probably curl)
21:31	mornfall	dons: Right, but unfortunately, you still apparently can't overturn the rest.
21:31	dons	and support that well.
21:31	dons	rather than doing half-arsed efforts 3 or 4 ways
21:31	Heffalump	no, support it well, then remove the others as properly redundant.
21:31	dons	sure.
21:31	Heffalump	If you think we should pick one, you should make that one obviously the best, then drop the others.
21:31	dons	either way, a decision is made to just pick one.
21:32	Heffalump	Not just claim that they don't matter.
21:32	mornfall	dons: Or the other option is to require haskeline, if cabal-install can get it right on all platforms. But we just need to pick one. (I'm fine with either.)
21:32	* dons	isn't sure Heffalump isn't in fervent agreement
21:32	Heffalump	You sound like you would be happy to drop the others first and then clean up the mess later. I wouldn't.
21:32	mornfall	Heffalump: Well, that's how development is done.
21:32	dons	Heffalump: its an atomic process :) as long as before mess >>= after clean
21:33	Heffalump	mornfall: rubbish.
21:33	dons	we just need to agree that afterwards the result should be clean.
21:33	dons	no one is suggesting releasing without support
21:33	Heffalump	Good software engineering involves treading carefully and not taking unnecessary risks.
21:33	mornfall	Heffalump: You need to break some eggs to make an omelette.
21:33	dons	and releases are the atomic transactions
21:33	mornfall	Heffalump: Oh, untrue.
21:33	mornfall	So very untrue.
21:34	dons	so everyone agrees that it isn't a useful use of our time to support N variants for M features?
21:34	dons	that where possible, one should be chosen and made the default?
21:34	mornfall	Of course, you have to tread carefully when you are dealing with a stable line. But if you have done your homework, you can make radical changes. If you haven't, you have to do radical changes, then do your homework and then start to feel safe again.
21:34	Heffalump	No, you'd just like it to be untrue, because it would save you having to actually think about people that don't do things your way.
21:34	dons	with long term benefits of a smaller, simpler code base, simpler build system, simpler dependencies and simpler deployment?
21:34	* kowey	goes home
21:35	Heffalump	dons: I agree that it comes at a cost.
21:35	dons	yep. we'd be removing support for things.
21:35	mornfall	Heffalump: Cost that for one I won't pay. It's easier to use git than to fight with darcs development model just to make darcs good enough to use.
21:36	dons	yes, there's bigger movements afoot that all this is for.
21:36	Heffalump	dons: the simpler deployment is only in situations that fit in with your world
21:36	dons	like xmonad's fight against awesome and dwm, we have to compete on several fronts, while paying a tax for not using gcc
21:37	dons	projects cannibalise each others features. the one that is hardest to install loses, without a very compelling use case.
21:37	Heffalump	if someone would cannibalise patch theory, that'd be great.
21:37	dons	users don't care about the theory :/
21:38	mornfall	Yeah, that's why they just use git.
21:38	dons	worse is better, and all that.
21:38	Heffalump	I mean the UI properties that the theory provides.
21:38	mornfall	Heffalump: There aren't many.
21:38	dons	and git offers branches and rebase
21:38	mornfall	Right. And the interactive commit is better than darcs', these days, I hear.
21:39	Heffalump	Well, they're important to me. If you don't need them, why aren't you using git?
21:39	dons	deep cherrypicking doesn't seem compelling, sadly
21:39	mornfall	And they have a GUI browser.
21:39	mornfall	Etc.
21:39	dons	Heffalump: "If you don't need them, why aren't you using git?" mmm. are we actually interested in competing with git?
21:39	dons	or is it game over?
21:39	dons	what's the compelling case we're making to the world for darcs over git?
21:39	mornfall	dons: If attitude doesn't change, I'd call it a game over.
21:39	dons	now that git cannibalised most of our good stuff?
21:39	Heffalump	Why is it a competition?
21:40	mornfall	Heffalump: Because there's only a finite pool of developers.
21:40	Heffalump	Aren't we just trying to improve the world?
21:40	dons	i'd be happy to see darcs released and frozen, perfectly formed.
21:40	mornfall	If you lose enough of them, it's game over.
21:40	dons	i think kowey's doing a great job helping keep the big vision out there, fwiw.
21:40	mornfall	dons: No such thing as a perfect piece of software.
21:40	dons	but we can do a lot more to state the case for darcs.
21:41	dons	the only simple vcs.
21:41	mornfall	I wish the core was simple, too.
21:41	dons	"don't get git rebase? here, my grandma gets darcs"
21:41	mornfall	The conflictor theory makes me feel uneasy.
21:41	lispy	mornfall: yeah
21:42	lispy	we can't argue robustness even
21:42	lispy	and we're using Haskell
21:42	mornfall	Right. That's sad.
21:42	Heffalump	until Igloo proves it, anyway.
21:42	mornfall	Heffalump: That won't make it any less complex and error-prone.
21:42	lispy	Heffalump: right
21:42	lispy	Heffalump: we need more research
21:42	Heffalump	someone could write it in Coq/Agda, perhaps.
21:43	lispy	Heffalump: While I think that would be helpful, I think we need more chalkboard proofs
21:43	Heffalump	there's stuff to do. And yes, to some extent there's a race for mindshare.
21:43	Heffalump	But we can't throw everything else out to compete in that race.
21:44	Heffalump	lispy: that's what Igloo is doing now
21:44	lispy	dons: with window managers you don't get pressure from other people to use their WM in order to collaborate with them
21:44	Igloo	mornfall: I don't think camp's theory is complex
21:45	mornfall	Igloo: Maybe you don't, because you are very familiar with it.
21:45	Igloo	And if it's proven correct then it's not error prone
21:45	lispy	dons: so I think the point about "which is easier to install" is different for vcs. It more about, which projects use it
21:45	dons	yes. again, something we've not addressed in a long time.
21:45	mornfall	Igloo: If it's complex, it is hard to implement right. No matter how many proofs you have for it.
21:45	Heffalump	Igloo: if it's proven correct and that proof corresponds to the source.
21:46	Heffalump	hence Coq/Agda.
21:46	lispy	if the darcs project itself uses darcs but seems unhealthy, then why would anyone pick darcs?
21:46	mornfall	lispy: Historical reasons, habit.
21:46	mornfall	I use darcs out of habit, really.
21:46	mornfall	I don't have any arguments left, and there's pressure to drop it, for a good measure.
21:47	lispy	Heffalump: it's also not clear that we could end up with an implementation that is also high performance if we extract it from Coq/Agda
21:47	Heffalump	if you really don't see the benefits of the darcs UI, or believe in patch theory at some level, you probably should do that
21:47	Heffalump	lispy: hmm, true
21:47	mornfall	It's a very hard argument to fight that you need the "nice" properties when it makes everyone's everyday life harder.
21:48	dons	Heffalump: so what do you think the goal for darcs now is? just to provide an implementation for the patch theory true believers? or ?
21:48	Heffalump	I'd like to see it take over the world. But only on its own terms.
21:48	Igloo	mornfall: The conflictor implementation is not complex
21:49	mornfall	For me, the goal is to provide a practical, robust and elegant VCS. Git fails on the elegant and partially on robust (due to flaky merge).
21:49	mornfall	Igloo: So why does darcs get it wrong?
21:49	lispy	People seem more tolerant to incorrect merges than they do to corrupted repositories
21:49	Igloo	mornfall: I'm talking about camp's conflictors here, darcs's are more complicated
21:49	mornfall	lispy: That's why I say partially.
21:50	Heffalump	An important thing is that git actually deliberately has a different philosophical standpoint: git thinks it's better to decide on the merge algorithm late, darcs thinks it's better to decide on it early.
21:50	Igloo	darcs currently gets it wrong becuase it doesn't have a complete implementation of the darcs2 theory, and the darcs1 theory is incorrect
21:50	mornfall	Igloo: But you aren't implementing darcs2 theory either.
21:50	Igloo	The thing that is hard is getting a correct theory, not implementing it
21:51	Igloo	No, because darcs2 theory is more complicated than is necessary
21:51	Igloo	And counterintuitive, IMO
21:51	mornfall	Igloo: But the part of the theory that deals with changes that don't conflit is extremely simple, right?
21:51	Igloo	Yes (simple, but large, as it's quadratic in the number of patch types)
21:52	mornfall	Igloo: Yours is counterintuitive on the identical change conflict again, one that was one of the worst bits of darcs1...
21:52	dons	Heffalump: and git doesn't care about theory. theory is like 5% of their effort.
21:52	Igloo	mornfall: Are you saying darcs handles non-conflicting changes incorrectly?
21:52	mornfall	dons: That's probably why they get so many users. : - )
21:52	Heffalump	mornfall: you're confusing intuition about the theory with intuition about the effects of the theory, I think
21:53	mornfall	Heffalump: Possibly. But that's why I don't think I want to practically use that version.
21:53	mornfall	Igloo: No, I'm not.
21:53	Heffalump	AFAIK camp's theory is agnostic to whether identical changes are conflicts or not.
21:54	Igloo	It's not agnostic, but I believe that it can be extended to make them not conflict
21:54	Igloo	But making them not conflict makes the theory more complex, which is why I'm not doing that at the moment
21:54	mornfall	Well, I don't think the "megapatch" approach is really compatible with non-conflicting prims. (And that's also where I disagree with the darcs2 stance on identical changes...)
21:55	Igloo	I think the megapatch approach is very important for an intuitive UI
21:55	Heffalump	mornfall's idea of an intuitive UI is nothing like everyone else's idea, IIRC :-)
21:56	Heffalump	btw, the philosophy of supporting multiple ways of doing things is also an advantage if you want to replace something, because you can introduce the new way first then obsolete the old way once it beds down.
21:56	mornfall	Could be, apparently there are complaints about my UI design. Although there are quite some people who like it. (So it's, I guess, love/hate sort of design.)
21:57	Heffalump	I think to do the stuff you were talking about a while back, you'd need to start a project yourself. It just wouldn't fit into darcs at all.
21:58	mornfall	Yeah, I know.
22:00	mornfall	But since I have too many projects on my hands, I probably won't. Getting to conservatory is more important now. And I think that's where 24 hours set in to limit practicality of dealing with version control development.
22:00	Igloo	conservatory?
22:01		kowey joined #darcs
22:02	mornfall	Learning to play bassoon.
22:04		nomeata joined #darcs
22:06	Heffalump	lispy: how do I tweak my darcs settings?
22:09	Heffalump	there is context in the patches, AFAICT, is it just the email body that needs it?
22:15	Heffalump	hmm, do we have any patches to fix the LANGUAGE pragmas not working on 6.6 issue yet?
22:22	lispy	Heffalump: not darcs context, I mean diff context
22:22	lispy	Heffalump: I'll try to look at my settings
22:22		cytzol joined #darcs
22:23	kowey	Sprint #2 to be held in conjunction with Hac5, March 2009, I hope
22:23	kowey	and the new thinking is two sprints a year
22:23	lispy	Heffalump: turns out I'm not sure what to tweak. Eric's advice is better, upgrade your darcs
22:24	kowey	one with Hac, so we get to mingle with Haskellers and one by ourselves so we get quality darcs time :-)
22:24	kowey	send --unified
22:24	lispy	kowey: I think dons suggested to me that those of us in Portland should have more requent hack-events not necessarily darcs related
22:24	kowey	fun!
22:24	kowey	hmm, maybe something for LondonHug to consider
22:24	lispy	What is Hac?
22:25	kowey	Haskell Hackathon
22:25	lispy	ah
22:25	lispy	Heffalump: I was going to tackle the LANGUAGE problem, I just can't get to it today
22:25	Heffalump	ok
22:25	lispy	maybe tomorrow I can start
22:25	lispy	it's only about an hour of work I think
22:25	Heffalump	is --unified new?
22:26	lispy	I don't think so
22:26	lispy	I've had --unified for a while I think
22:26	Heffalump	ganeshdefoe:~/darcs-slurpy$ darcs -v
22:26	Heffalump	2.0.2 (+ 62 patches)
22:26	kowey	it's old and now you no longer have a choice
22:26	lispy	Ijust didn't realize --unified gave the context
22:26	Heffalump	ganeshdefoe:~/darcs-slurpy$ darcs send --unified --edit
22:26	Heffalump	darcs failed: unrecognized option `--unified'
22:27		int-e joined #darcs
22:28	Heffalump	I'm just building latest stable, anyway.
22:33	Heffalump	ok, so now I have the recent one, and it still doesn't like --unified. Should I just assume it'll do it automatically?
22:33	lispy	Heffalump: at some point it was removed, so yeah
22:34		int-e left #darcs
22:34	Heffalump	but then why was my old one not doing it? It didn't accept the option either.
22:35	mornfall	Heffalump: I had the same issue.
22:39	Heffalump	ok, I see some context in a test patch I just sent myself
22:39	lispy	Heffalump: the context is addicting
22:42	Heffalump	gah, darcs sends even when I abort the editor :-(
22:54		beschmi left #darcs
22:56	Igloo	lispy: Can you remind me, do you think darcs should be buildable without msys?
23:16		Jerri_ joined #darcs
23:31		felipe joined #darcs
23:39		tux_rocker joined #darcs
23:53		mmorrow joined #darcs
23:54		adu joined #darcs

← Previous day | Channels | #darcs index | Today | Next day → | Search | Google Search | Plain-Text | plain, newest first | summary