IRC log for #darcs, 2010-01-11

All times shown according to UTC.

Time	Nick	Message
00:14	twb	mornfall: siva is back up after three failed boots :-/
00:24		gbeshers joined #darcs
01:25		zooko joined #darcs
01:26	zooko	Hm, is there something special I need to do to make darcs emit a new --version when I build it with cabal?
01:28		Because it still says 2.3.1 (+ 271 patches)
01:28
01:28		but that's what it said before I pulled 155 more patches and rebuilt
01:28		mornfall joined #darcs
01:29	zooko	bbiab
01:36	twb	zooko: i the new darcs in your path?
01:37		cabal install -fcurl-pipelining && ~/.cabal/bin/darcs --version
01:43	sshc	if I'm updating a version number in a file, how do I set its patch to depend on other patches?
01:44	twb	sshc: --ask-deps
01:45		sshc: but "darcs tag" is probably easier
01:45		I usually just bump the version and then immediately make a tag patch
01:46	sshc	twb: thanks. I'm learning darcs and I'm really liking it thus far
01:46		I'll look at the documentation to see what --ask-deps does
01:47		but how do I revert a patch without a removing it?
01:47		ie apply the inverse of the patch
01:47	twb	rollback creates an inverse patch
01:47	sshc	I see
01:47		thanks, I could'nt find that
01:51		"The obliterate command can wisely be disabled in a dedicated main repository by adding obliterate disable to the repository's defaults file. "
01:51		I don't see "defaults in _darks/
01:52	zooko	twb: yes, this is the new darcs.
01:52		How can I figure out whether this darcs executable has curl-pipelining enabled?
01:55	sshc	where is a defaults file?
01:55	twb	sshc: _darcs/prefs/defaults (per repo) and ~/.darcs/prefs/defaults (per user)
01:55		zooko: dunno
01:58	sshc	twb: that doesn't exist, but _darcs/prefs/defaultrepo does
01:59	twb	sshc: create it
01:59	sshc	twb: is there documentation for what can go in that file?
01:59	twb	Unfortunately I have been very very lazy in the last six months, and I haven't written a FILES section for the darcs manpage / darcs --help
02:00		The TeX manual documents it, but it's not very nice
03:08		Heffalump joined #darcs
03:08		Igloo joined #darcs
03:08		darcscommitbot joined #darcs
03:08		sshc joined #darcs
03:08		gwern joined #darcs
03:08		dons joined #darcs
04:56	sshc	is it possible to tag an old repository state?
04:56	twb	Yes, by making a branch, obliterating patches from that branch, creating the tag, then pushing the tag back into your main repo
05:21	zooko	sigh
05:21		I have a repo on a server -- R1. And I do darcs get R1 to get a copy locally, let's call it R2. Then I get one more patch into R2 from another repo, then try to push to R1, and I get conflicts.
05:22		So then I darcs get from R1 to a separate local copy, R3, then try to push from R2 to R3. Works fine.
05:22	twb	Ow.
05:22	idnar	ďoes R1 have unrecorded changes?
05:22	zooko	So then just for good measure, I try to push from R3 to R1.
05:22		idnar: ! Good idea.
05:22	twb	Oh, do you- damn, you beat me to it
05:22	zooko	Clearly I'm tired.
05:23		That's always the answer to this one.
05:23		Oh wait
05:23		except I just recreated R1 from scratch by mv R1 R1-broken && darcs get R1-broken R1
05:23		So, no.
05:23		Now double-checking.
05:23		sourceorg:~/darcs/pycryptopp/trunk-hashedformat$ darcs whatsnew -l
05:23		No changes!
05:23
05:24		Oh, and although the darcs executable on my side did say "darcs failed: Refusing to apply patches leading to conflicts." nonetheless the patch is now there in R1 on the remote server.
05:26	* zooko	wonders what he should do
05:26	zooko	I think "go to bed" is the right answer
05:26		This is the same repo that had http://bugs.darcs.net/issue1669 , although I've since obliterated the patch that triggered that bug.
05:28		Hm, maybe I'll try converting to darcs-2-format.
05:29	twb	Man, ghc takes a loooong time to compile
05:30		Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz and 4GB of RAM, and it's still taking gaes
05:30		*ages
05:30	zooko	darcs convert takes a long time to convert this repo
05:30		I kind of feel like converting to darcs-2-format is not the right thing to do.
05:31		I should probably understand what is going on better first.
05:31		Dammit. I'm going to sleep.
06:33		solidsnack joined #darcs
07:08		solidsnack joined #darcs
07:15		solidsnack joined #darcs
07:43		lelit joined #darcs
08:02		darcscommitbot joined #darcs
08:03		darcswikibot joined #darcs
09:32		kowey joined #darcs
09:32	kowey	morning
09:45		more from Evan M. http://community.livejournal.c[…]_tech/270665.html
09:46		he says in his comment that even bzr repos are easier for him to figure out (as a git user)
09:49	twb	in that post? Where?
09:49	kowey	in his reply to my comment
09:49	twb	Oh, in a comment
09:49	kowey	it may be good if we had a Darcs for Git Users guide of some sort answering common question from experienced git folks
09:50		I don't think the right answer is necessarily to do what Johan/Evan say, but it's good to think about what's really causing them pain and how we can provide relief
09:51		(over the long term as we get a better handle on things)
09:53	twb	I think commit's current behaviour is stupid
09:54		It was done that way because at the time, CVS and centralized VC were prevalent
09:55		Having said that, I don't find it significantly annoying
09:56	kowey	could you clarify?
09:56	twb	Which part?
09:56	kowey	are you referring to the deliberate choice not to use 'commit' to signal that we're doing something different?
09:56	twb	Currently commit just says "you either meant record or record+push. Do those."
09:57		Whereas in other dVCSs, commit means record
09:57		The reason is that "cvs commit" (and svn commit) is record+push, and droundy didn't want people to accidentally commit only locally and not realize they hadn't pushed their changes.
09:58		At least, I assume that's why
09:58	idnar	in bzr it might mean either
09:58	twb	I wasn't bothering to explain bzr's bipolar setup :P
09:58	idnar	and then git has its own brand of weirdness
09:59		where you select what changes to commit using a separate command
10:44	kowey	we've had a pretty big discussion on this before
10:44		and there was a consensus that we should keep the distinction
10:45		http://lists.osuosl.org/piperm[…]-July/020433.html and http://lists.osuosl.org/piperm[…]-July/020438.html
10:47	twb	kowey: like I said, I don't care enough to shout down others
10:48	kowey	nod (I actually personally would rather commit be an alias for record for reasons in http://lists.osuosl.org/piperm[…]-July/020434.html ) too
10:48		I'm just scared of opening that can of worms again :-)
10:49		although maybe this is the sort of thing I should maybe just push through by iron fist and apologise for later
10:49		lots of indecision + opinions, so just choose, stick to guns and hope for the best
10:50	twb	Like Fred the Great
12:23		gh_ joined #darcs
13:09	gh_	hi
13:09	lambdabot	gh_: You have 1 new message. '/msg lambdabot @messages' to read it.
13:10	gh_	kowey, thanks for the hostel room
13:12		sshc, "whatsnew" is a diff between the recorded state and the current state, "diff" shows the diff associated to one patch
13:22		solidsnack joined #darcs
14:02		gh_ joined #darcs
14:25		intripoon joined #darcs
14:39		gh_ joined #darcs
15:20		pozic joined #darcs
15:20	pozic	How can I see the difference between a remote repository and mine?
15:22		lispy joined #darcs
15:29		lispy joined #darcs
15:29	kowey	pozic: darcs pull --dry-run to see what they have that you don't
15:29		pozic: darcs send --dry-run for the other direction
15:30	lispy	Hello
15:30	kowey	if you want a diff, I think you have to actually fetch with darcs pull or darcs get
15:30	lispy	kowey: my benchmark run failed again last night -- I'm clueless as to what a get would ask.
15:30	kowey	lispy: I have some patches to darcs-benchmark that enable verbose mode, if it helps
15:30	lispy	kowey: that would help
15:30	kowey	I'll send them to you (pending review from mornfall when he's less busy)... one issue is that this changes the UI
15:31		OK, off they go
15:31	lispy	Well, even if they are not in the mainstream darcs-benchmark it saves me the time of implementing that stuff myself :)
15:31		I'm sad no one replied to my Darcs.IO question
15:32		That's a bad sign for module with no comments :(
15:32	kowey	no luck from darcs changes, huh?
15:32	lispy	I didn't actually check that yet :)
15:32	kowey	the problem is that our institutional knowledge is David
15:32		hmm, I wonder if he would be interested in coming to Zurich and seeing how darcs is doing after all this time
15:32	pozic	kowey: ok, it was darcs send --dry-run that I was interested in. Thanks.
15:33	lispy	It can't hurt to ask?
15:33	kowey	yeah, I'm not sure about the timing... we may be better poised to do this at a later sprint
15:33	lispy	This new framework is intended to unify the current apply_to_slurpy and
15:33		(hopefully) patch_check codes. It also will allow us to apply a patch
15:33		directly to a filesystem, without first slurping the directory. This
15:33		should allow (for example) faster gets.
15:33	pozic	Does anything actually get done at a sprint?
15:34	lispy	(sorry about the long paste)
15:34		pozic: yes!
15:34	pozic	I can imagine that going to take a look at the city is much more interesting ;)
15:34	lispy	pozic: lots of code and discussions that frame the next N months of development
15:34	kowey	pozic: mostly planning and team bonding, but see http://blog.darcs.net/search/label/sprints
15:35	pozic	kowey: I can sort of belief in team bonding.
15:35	kowey	over time I hope we get better at using sprints to train newbies
15:35	pozic	kowey: the rest can work via IRC, but I might be wrong.
15:35	kowey	we made our first attempt at the last one, and in time we'll get better (at training newbie darcs hackers, that is)
15:35		as for planning, it's surprising how much difference a real world discussion makes
15:36		but I think I'm a bigger believer in real life interaction than most folks here :-)
15:36	pozic	kowey: when is someone qualified to become a darcs newbie?
15:37	kowey	as soon as they get started, or express an interest in getting started! http://wiki.darcs.net/Development/GettingStarted
15:37		see also http://tinyurl.com/darcs-probablyeasy2
15:37	mornfall	lispy: You want to make it ReadableDirectory m.
15:37	lambdabot	mornfall: You have 2 new messages. '/msg lambdabot @messages' to read them.
15:37	mornfall	Come on.
15:37	lispy	mornfall: That's what I started writing this morning
15:38		mornfall: I found that I need to tweak the readHunk code to do what I want though
15:38	mornfall	Hm. twb said 1 month ... I guess I ignore lambdabot a lot.
15:38	lispy	TBH, lambdabot has issues with saving her state reliably :)
15:38	mornfall	lispy: Anyawy, Darcs.IO just abstracts filesystem operations into classes.
15:38	lispy	mornfall: Okay, but do you know why?
15:39	mornfall	lispy: Well, one thing is safety (better types) and another is that we want different filesystem "semantics" in different contexts.
15:40		ReadableDirectory m => ... says that this is not going to ever do any writes in m.
15:40		Filesystem writes, that is.
15:40	pozic	That something is Readable doesn't mean it isn't also writable, imho,
15:40		.
15:40	mornfall	As for semantics; there's tolerant IO which we use for working copy: where for pristine, you want write failures to treat as catastrophic, in working you want to treat them as, "too bad but life gotta move on".
15:41	pozic	If the property is not writable, it should just be named NotWritable.
15:41	mornfall	pozic: It's pretty clear from the class definition. I don't really care.
15:42	kowey	Readable is probably closer to what we really mean, as in "NotNecessarilyWriteable"
15:42	lispy	pozic: Well, WriteableDirectory assumes ReadableDirectory. So if you want both use WriteableDirectory
15:43	mornfall	lispy: Anyway, is it clearer now?
15:43		\|Mike\| joined #darcs
15:43	lispy	mornfall: Well, I had gathered the same things you're telling me from reading the code. But, it's still nice to hear someone else confirm it
15:43	\|Mike\|	wow :-)
15:43	kowey	perhaps we can see some haddocks out of this, lispy? :-)
15:44	mornfall	Back to bassoon.
15:45	lispy	kowey: perhaps
15:45	kowey	lispy: also, did you get my patches?
15:45	* kowey	doesn't trust these newfangled computer things
15:46	lispy	kowey: not yet, but the gateway I use to get my email tends to delay things by about 4 hours
15:46		so if I don't get it by tomorrow then I can worry
15:47	kowey	I once had a setup where my mail would go into a staging area and only be sent if it had not been edited in the last hour
15:47		... only it turned out that I kept using my needs-immediate-delivery override, so I gave up
15:48	lispy	Yeah, I don't think I would like that system despite it originating from good intention
15:51		solidsnack joined #darcs
15:56	pozic	Does every committer has atleast one of CS, Math, or Physics degree?
15:56	lispy	pozic: several do, but there have been so many it's hard to keep track :)
15:57		kowey: the patches arrived
15:59	kowey	if you're keen on finding out, pozic, http://wiki.darcs.net/DarcsTeam
15:59		it might take some intensive googling or emailing though
16:00	pozic	kowey: yes, some of them definitely do.
16:01	mornfall	What kind of degree is required? : - P
16:02	pozic	mornfall: none, since you can fake everything over the Internet.
16:08	* mornfall	fakes his existence.
16:22	darcswikibot	11 Jan 16:16 - DWN 2010-01-11 (Guillaume Hoffmann)
16:22		11 Jan 16:17 - TAG 2010-01-11 (autotag) ()
16:26	kowey	50th edition of DWN
16:47		gh_ joined #darcs
16:48	gh_	yup
16:54		pozic left #darcs
16:59		gal_bolle joined #darcs
17:00		gbeshers joined #darcs
17:02		gbeshers joined #darcs
17:15	mornfall	kowey: As for our darcs-benchamrk patches, I'm not opposed, although to me it seems a little superfluous to have subcommands there.
17:17		kowey: Well, it probably deserves to get a "prepare" subcommand at least then, to produce a tarball from a repository. ; - )
17:18	kowey	oh yeah, I didn't really want the subcommands... it was just the easiest way I could think of to get cmdargs to work
17:19		darcs-benchmark prepare could also be handy for things like making sure that darcs-1 repos get named --hashed
17:19		and are in the right format, etc
17:20		lelit joined #darcs
17:20	kowey	note that I messed up your help text... it now needs to say 'darcs-benchmark run' in place of 'darcs-benchmark' (oops)
17:30		kolmodin joined #darcs
17:49	mornfall	kowey: Pull darcs-benchmark.
17:53	kowey	ooh, lispy: Pull darcs-benchmark :-)
17:53		more transcript goodness (although I'm not sure if we should be unhappy about cmdargs' use of unsafePerformIO)
17:53	lispy	kowey: hmm?
17:53	kowey	I guess we have no right to talk since we do the same in Darcs for some global vars
17:54	lispy	We should be unhappy about the use of it, if it prevents referential transparency in any way
17:54	kowey	oh, darcs-benchmark may now be a little more informative with --verbose, but if I understand correctly, you'd need to be on laptop anyway
17:54	lispy	If it doesn't violate RT, then it's actually a safe usage
17:55		(that's all that the 'unsafe' means)
17:55	kowey	I think it only uses it so that you can saythings like verbose <- isLoud
17:55		without having to pass around any config state
17:55		a bit like how we do it
17:56		anyway, it's just for darcs-benchmark
18:01	mornfall	It doesn't leak IO into non-IO code, since the uses are restricted to IO code, IIUIC.
18:32		arjanb joined #darcs
20:18		tux_rocker joined #darcs
22:03		florent joined #darcs
22:45	darcscommitbot	06 Jan 17:55 - Make release.sh handle beta releases correctly (Reinier Lamers)
22:57	* mornfall	earns review cookies.
22:57	lispy	woot
22:58	* idnar	cooks reviews
23:04	mornfall	lispy: Do you have utf8-enabled darcs on windows?
23:05		I'm wondering if I screwed something up or it breaks for everyone.
23:05		With the utf8 patches, my darcs crashes in darcs init on windows vista.
23:05		(See buildbot...)
23:13	lispy	mornfall: I haven't built darcs on windows for a long time; currently I wouldn't know how due to the lib unicode snafu
23:23	mornfall	Ok.
23:32		adu joined #darcs