Camelia, the Perl 6 bug

IRC log for #darcs, 2012-06-01

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

All times shown according to UTC.

Time Nick Message
00:05 owst @tell gh_ thanks for all the patches! One thing: it'd perhaps be better if you could keep the imports/reformatting in separate patches, to make it more obvious that things are "just being moved"
00:05 lambdabot Consider it noted.
00:18 owst That's enough reviewing for one night
00:18 owst night all!
01:02 darcscommitbot joined #darcs
01:18 intripoon_ joined #darcs
03:36 smoge joined #darcs
03:58 JaffaCake1 joined #darcs
05:05 smoge left #darcs
07:02 darcscommitbot joined #darcs
07:20 mekeor joined #darcs
08:11 drdo joined #darcs
09:08 owst joined #darcs
09:13 dcoutts joined #darcs
09:13 dcoutts joined #darcs
09:55 donri joined #darcs
10:49 donri joined #darcs
11:05 nomeata joined #darcs
11:08 kowey joined #darcs
11:44 schlaftier joined #darcs
13:02 darcscommitbot joined #darcs
14:21 intripoon joined #darcs
14:24 bsrkaditya joined #darcs
14:32 donri_ joined #darcs
14:40 schlaftier joined #darcs
15:13 kowey joined #darcs
15:58 kowey bsrkaditya: hi! I think I'm ready for our meeting if you are
15:58 lambdabot kowey: You have 1 new message. '/msg lambdabot @messages' to read it.
15:58 bsrkaditya Yes, I am ready
15:58 bsrkaditya This week, I have
15:58 bsrkaditya reimplemented match,
15:59 bsrkaditya reimplemented isPatchIndexInSync,
15:59 bsrkaditya and implemented annotate over a directory.
16:00 kowey sounds good
16:00 bsrkaditya Due to the first two points, changes over
16:00 bsrkaditya GNUMakeFile reduced to 0.43
16:00 kowey do I understand correctly that match was about the style things suggested by owst?
16:00 bsrkaditya (it was around 1sec before)
16:01 kowey oh, I see, so your reimplementation is more efficient?
16:01 bsrkaditya Yes, match is that function
16:01 kowey in what way?
16:01 bsrkaditya Exactly
16:01 bsrkaditya It identifies the relevant patches faster, by using infomap
16:01 bsrkaditya Annotate over GNUMakeFile takes 0.86 sec
16:02 bsrkaditya compared to 2 sec.
16:02 kowey so these operations appear to be taking half the time on a couple of good candidate files
16:02 bsrkaditya Exactly.
16:02 kowey out of interest, are these files in the darcs darcs repo?
16:03 bsrkaditya yes
16:03 kowey and if so, (silly question), are you comparing the same darcs repo each time?
16:03 bsrkaditya Not really.
16:03 bsrkaditya Every time a few patches are being added.
16:03 kowey it might be worthwhile to consider taking a copy of the repo
16:03 kowey freezing it into a tarball
16:03 kowey and always testing from a fresh untar
16:04 bsrkaditya Okay. I will do so.
16:04 bsrkaditya Now to annotate over a directory
16:04 kowey so that your timing tests don't get confused by any state changes (even if adding patches would intuitively make it a hard test)
16:04 kowey hmm, hang on,
16:04 kowey sorry, I'm a bit slower :-)
16:05 * kowey looks in your high-level docs to see what an infomap is
16:05 kowey hmm, ok, not sure I get it
16:05 kowey could you elaborate a bit about these changes?
16:06 bsrkaditya Okay
16:06 bsrkaditya I created a function getPatches
16:07 bsrkaditya which identifies the patches related to a particular file
16:07 kowey this was in the past, not just last week?
16:07 bsrkaditya Patch index stores a Map PatchId INt
16:07 bsrkaditya no I created it this week
16:07 kowey oh ok, and patchid is these darcs patchinfo hashes
16:08 bsrkaditya and it also stores a Map FileId (Set PatchId)
16:08 bsrkaditya I identify the patches of a particular file
16:08 raichoo joined #darcs
16:08 kowey the patch index does
16:08 bsrkaditya and use the Map PatchId Int
16:09 bsrkaditya to identify their positions in the repo
16:09 kowey positions?
16:09 bsrkaditya The patches come in an order when you say readRepo repo
16:10 bsrkaditya Hence every patch has a position in the repo
16:10 kowey does this ordering have a particular meaning
16:10 kowey or should it be considered essentially arbitrary?
16:10 bsrkaditya arbitrary.
16:10 kowey does the ordering depend on the implementation of readRepo?
16:10 kowey if somebody changes readRepo, does that invalidate the ordering?
16:11 kowey and does it matter?
16:11 bsrkaditya No
16:11 bsrkaditya basically readRepo returns a patchset
16:11 bsrkaditya and as long as it returns the same patchset
16:11 bsrkaditya after a reimplemnetation
16:11 bsrkaditya It will not invalidate anything
16:12 kowey oh hang on
16:12 kowey this ordering, is it the order of the patch sequence in the repo?
16:12 bsrkaditya yes.
16:12 kowey ie. the ordering you see as a user when you type "darcs changes?"
16:12 kowey oh ok
16:12 kowey so it's not *really* arbitrary is it?
16:13 kowey the order corresponds to the order that the patch was applied or created in the repo
16:13 kowey which is not something that depends on the implementation of readRepo?
16:13 bsrkaditya Yes.
16:13 kowey i see, so the int is basically the same int you get when you use darcs changes --number
16:13 kowey ok
16:14 kowey so how does knowing the position of a patch in a repo help you?
16:14 bsrkaditya So, I use the positions to make a linear pass over the patches of the repo to filter the relevant patches
16:15 bsrkaditya Before, I used to compare every patch with the Set PatchId
16:15 kowey ok so if I understand correctly, two things have changed
16:15 kowey 1. is that you're comparing ints vs patchids? which may be cheaper?
16:15 bsrkaditya no
16:15 bsrkaditya I am not comparing ints
16:15 kowey still comparing patchids?
16:15 bsrkaditya No.
16:16 kowey hmm
16:16 bsrkaditya Let me explain
16:16 bsrkaditya this function, getElems :: [Int] -> [a] -> [a]
16:16 bsrkaditya selects the elements of the second list
16:17 kowey you're doing basically the equivalent of a repeated (patches !! i)?
16:17 bsrkaditya based on the positions given by [Int]
16:17 bsrkaditya Yes, but the implementation is linear
16:17 bsrkaditya time performance
16:18 kowey because you consume the indices as you traverse the list in the order that the list is in
16:18 bsrkaditya Yes.
16:18 kowey ok, so here you gain by doing a single linear pass of the patch history with constant cost for each comparison
16:18 bsrkaditya Yes
16:18 kowey whereas before you were also doing a single linear pass of the history
16:19 kowey but each operation consisted in checking to see if the current item is in the set of patches
16:19 bsrkaditya Yes.
16:19 kowey so roughly the same complexity
16:19 kowey unless your Set PatchId happens to be huge
16:19 bsrkaditya Not really.
16:19 bsrkaditya I suppose so. I did not check for small cases. :-)
16:20 kowey well, what's this Set PatchId?
16:20 kowey theoretically, I imagine, it could consist of the entire history?
16:20 bsrkaditya The patch ids that correspond to the patches that modifed a particular file
16:20 kowey so we're going from O(n log n) to O(n) ?
16:20 * kowey isn't good at this kind of reasoning
16:21 bsrkaditya n -> number of patches
16:21 bsrkaditya m -> number of patches outputed
16:21 bsrkaditya We moved from O(nlogm) to O(n)
16:21 kowey when you say number of patches outputed
16:21 kowey you mean the number of patches that correspond to a given file?
16:22 bsrkaditya Yes
16:22 kowey what happens when you're looking at multiple files?
16:22 kowey do you make repeated linear passes?
16:22 bsrkaditya I implemented multiple files only in the latest iteration
16:22 kowey ah, this is to make annotate work on directories?
16:22 bsrkaditya Yes.
16:22 kowey how did that go?
16:23 bsrkaditya It worked.
16:23 bsrkaditya But the performance directly depends on the number of files in consideration
16:23 kowey multiple linear passes?
16:23 bsrkaditya If you run annotate over src/Darcs/Repository
16:23 bsrkaditya No, I merge all the indices
16:24 bsrkaditya and then make a linear pass
16:24 kowey which would make sense…
16:24 bsrkaditya The time taken is 0.8 sec
16:24 bsrkaditya (If you run annotate over src/Darcs/Repository)
16:24 kowey so what impacts the performance then is just the fact that the more files you match on, the more patches you return?
16:24 owst Does annotate on a directory actually do anything sensible?
16:24 owst For me, it just shows the file listing, with "unknown" where the patch would be
16:25 kowey at least in the old version, I recall vaguely it had some sort of use
16:25 bsrkaditya owst: I have that doubt myself. :-)
16:25 * kowey still hasn't fully caught up with the brave new you-can-use-annotate-now world
16:25 owst Last patch per-directory?
16:25 owst *file
16:25 owst (in the dir)
16:25 owst No point wasting time on a feature that dubious at best ;-)
16:26 kowey well, I guess multiple files is going to be needed, eg. for changes
16:26 owst Alos,
16:26 owst Argh, keyboard fail
16:26 owst Also: --(no-)summary -u don't seem to do anything to annotate on files
16:27 kowey is this a more-tests-needed situation?
16:27 owst (So while you're touching annotate, it might be a good chance to clean it up)
16:27 kowey or a what-is-this-supposed-to-do situation?
16:27 owst The second one ;-)
16:27 kowey :-) ah, darcs :-)
16:28 kowey bsrkaditya: sorry, maybe we should get back to things
16:28 * owst stops interrupting
16:28 bsrkaditya So annotate over a directory could take as much as 6sec(with patch index)
16:28 kowey ok, so you have annotate working on directories, does it also work (owst: please do jump in; makes me less anxious that we're boring the channel)
16:28 kowey ...on multiple files typed into the cli
16:29 bsrkaditya I do not think so
16:29 kowey is it ever *slower* than the original pre-patch-index version?
16:29 owst % darcs annotate src/*
16:29 owst darcs failed:  annotate accepts at most one argument
16:29 owst That's what happens :-p
16:29 kowey ah-ha!
16:29 kowey problem solved by ui constraints
16:29 bsrkaditya You should try darcs annotate src/
16:29 kowey well, it may matter when we do darcs changes
16:29 bsrkaditya no *
16:30 kowey not ever slower?
16:30 bsrkaditya It is atleast as fast as the no-patch-index-version
16:30 kowey intuitively, it'd make sense
16:30 Heffalump is darcsden down?
16:30 bsrkaditya Yes.
16:30 kowey it may be good just in case to implement a --no-patch-index
16:30 kowey Heffalump: I'd noticed and gotten in touch with Alex
16:30 Heffalump ta
16:31 kowey bsrkaditya: also, what about darcs changes?
16:31 owst Why aren't we in charge of darcsden, again?
16:31 owst :-)
16:31 bsrkaditya kowey: Can you elaborate?
16:31 owst bsrkaditya: oh, I see, annotate on a directory *does* do something
16:31 Heffalump owst: well, we didn't write it, for one thing :-)
16:31 owst (I'd only tried on a small local repo)
16:31 kowey is that for later, or does it come for free thanks to your recent work?
16:32 owst Heffalump: I meant in terms of having it on darcsvm
16:32 Heffalump don't think that'd cope
16:32 owst :-(
16:32 bsrkaditya kowey: it is now faster
16:32 kowey I created a hub.darcs.net
16:32 bsrkaditya for free
16:32 Heffalump den.darcs.net, surely
16:32 kowey which I think sm will point to his darcsden instance on joyful
16:32 Heffalump :-)
16:32 owst heh
16:32 kowey yeah, wonder why I decided against not, that seems really good now
16:33 * kowey shrugs and makes a mental note to create that alias later
16:33 Heffalump we don't need to ape github too slavishly
16:33 owst I noticed that there's no way to set the motd file for darcsden
16:33 kowey and what happens if you darcs changes foo bar baz?
16:33 owst darcsden repos*
16:34 bsrkaditya No, I did not implement ui support for multiple files in changes
16:34 kowey but files and directories will go faster
16:34 bsrkaditya Yes.
16:34 kowey does it break if you give it mulitple files?
16:34 bsrkaditya No
16:35 bsrkaditya sorry, what do you mean?
16:35 kowey what happens?
16:35 kowey does giving darcs changes with multiple files somehow cause something bad to happen in the patch-index-enabled darcs in its present state?
16:35 bsrkaditya It gives the patches corresponding to the first file.
16:36 kowey meaning you can get the wrong results?
16:36 kowey (for now)
16:36 bsrkaditya Yes. :-)
16:36 kowey :-)
16:36 kowey ok, and one more thing on my mind
16:36 kowey what about the non-range matchers you were hoping to work on?
16:36 bsrkaditya Not yet done.
16:37 bsrkaditya I will do it over the weekend.
16:37 kowey ok, do you still feel it's priority in your grand-bsrkaditya-plan?
16:37 bsrkaditya Yes. :-)
16:38 kowey cool, right, that brings us to the next part of the meeting, I think
16:38 kowey what are you thinking of working on next week? (and the week after perhaps, as I won't be here)
16:39 bsrkaditya I want to make the patch-index versions of annotate and changes fully functional
16:39 kowey feature parity
16:39 bsrkaditya But there is something I want to discuss about annotate over a directory
16:39 bsrkaditya there is a file
16:40 bsrkaditya tests/issue1248.sh
16:40 bsrkaditya which gives different results(as of now) between the two version. (when called upon tests/)
16:40 bsrkaditya patch index annotate gives: unknown | tests/issue1248.sh
16:41 bsrkaditya whereas ordinary annotate gives: script for issue1248 | tests/issue1248.sh
16:41 bsrkaditya This is because
16:41 bsrkaditya /tests/issue1248.sh
16:41 bsrkaditya is two files over the repo history
16:41 kowey ah, lovely lovely dogfood
16:41 bsrkaditya ./tests/issue1248.sh -> 2#./tests/issue1248.sh from 20090815232335-6fd16-7a67607033​3eecd2bbd0a98acd268f42280d9de4 to -
16:41 bsrkaditya ./tests/issue1248.sh -> 1#./bugs/issue1248.sh from 20090512205429-81bb2-5b722e0d26​34f18bba920d41d1320b4c35bf195c to 20081115205509-72aca-69ac772583​bbbd8f6b7848f7ce37fb6621a30d25
16:42 bsrkaditya The patch:
16:42 bsrkaditya Tue May 12 03:12:36 IST 2009  Ganesh Sittampalam <ganesh@earth.li>
16:42 bsrkaditya * script for issue1248
16:42 bsrkaditya is a patch of file: 1#./bugs/issue1248.sh
16:42 kowey OK
16:43 bsrkaditya but I gave patch index annotate the patches of 2#./tests/issue1248.sh
16:43 bsrkaditya because that is the latest file with that path
16:43 kowey so my general knee-jerk reaction would be that patch-index-darcs should behave identically to patch-index-less darcs BUT
16:43 kowey or rather UNLESS
16:43 kowey patch-index-less darcs is doing the Wrong Thing
16:44 bsrkaditya is it the wrong thing?
16:44 kowey well, I don't know!
16:44 kowey I wonder if this is "just" a UI issue
16:44 kowey or if there's something very deep/fundamental we need to be thinking about here
16:45 kowey I guess one tool we could use for this
16:45 kowey is to imagine we have a future darcs
16:45 kowey that has GUIDs for files
16:45 amgarchIn9 joined #darcs
16:45 kowey and which just maintains a sort of local mapping between GUIDs and filenames
16:45 kowey in that hypothetical future Darcs
16:45 bsrkaditya We already have that in patch index.
16:45 kowey what would be the right sort of behaviour?
16:46 kowey (well I guess I was thinking maybe a bit more of the case where you for example have two unrelated repos with 'Makefile' in them and you want to merge the two)
16:46 * kowey wonders if he's actually being relevant
16:46 bsrkaditya If you want a different behavior, you will have to reimplement annotate
16:46 kowey or just doing his usual make-a-bogus-connection thing again
16:47 kowey it sounds like at the very least
16:47 kowey it would be a good idea to understand deeply why your version of annotate behaves differently from the current one
16:47 kowey now, I think you do
16:47 kowey because you see what patchids the filenames map to etc
16:47 kowey but what I mean is something a bit more abstract
16:48 bsrkaditya fileids not pathids
16:48 kowey like, what are the implicit philosophical stances of the two versions?
16:48 kowey and what are the implications for a user?
16:48 kowey if called upon to write the user manual, how would you possibly explain this sort of situation?
16:48 kowey like so if I'm a regular old darcs user, right?
16:49 kowey I don't care about patchids
16:49 kowey nor fileids
16:49 kowey that's a bunch of technical gobbledygook
16:49 bsrkaditya the current annotate works by inverting the patches, and applying them. Hence if you want to give the last patch, it will by default give the last patch of the first file
16:49 kowey so what I'm trying to get at is for you to try to work on getting an understanding for yourself of what *you* think should happen
16:50 kowey (which is not the same as just-do-what-you-want, to be clear)
16:50 bsrkaditya But, on the other hand, the user will prefer to get the last patch of the current file.
16:50 kowey ok, does this mean that pre-new-annotate darcs does something the opposite?
16:50 owst % darcs replace '`fmap`' '<$>' src/TypeReconstruction.hs
16:50 owst darcs failed:  '`fmap`' is not a valid token!
16:50 bsrkaditya Yes.
16:50 owst :-(
16:50 kowey wow
16:51 kowey hmm, perhaps it'd be worth consulting with mornfall if he's up to it
16:51 kowey see what he thinks the right thing is and why
16:51 kowey this is the sort of thing you're going to want some sort of documentation on whatever happens
16:52 kowey I guess my recommendation would be to make a provisional choice, and run with it
16:52 kowey just be prepared to defend said choice
16:52 kowey another tool you could use is to launch a discussion on darcs-devel and see what people think
16:53 bsrkaditya I will go by the current implementation/
16:53 kowey meaning the current implementation that works differently from current annotate?
16:53 kowey was it clear what I meant about pre-new-annotate?
16:53 kowey new-annotate was recently implemented by mornfall
16:53 kowey it's great! actually usable compared to old annotate
16:53 bsrkaditya I mean make the patch-index-annotate give the same output as annotaet
16:54 kowey which one?
16:54 kowey there are 3 annotates here
16:54 kowey the ancient one, the current one, and patch-index one
16:54 bsrkaditya mornfall's annotate. (the annotate in the repo)
16:54 kowey does the current patch-index annotate behave the same way as the ancient one?
16:54 bsrkaditya I have no idea
16:55 mornfall Recently like few years ago? :-P
16:55 kowey well hang on, what does your response to "ok, does this mean that pre-new-annotate darcs does something the opposite?" mean?
16:55 schlaftier joined #darcs
16:55 * kowey 's time-sense has gone totally haywire
16:56 mornfall Either way, old annotate goes from time 0 forwards.
16:56 mornfall New annotate goes from time "now" backwards.
16:56 mornfall They presumably both (new for sure) stop when they have full annotation of the file.
16:56 mornfall There was a bug in the new annotate that made it miss a line and go all the way back.
16:56 mornfall Not sure it was ever fixed.
16:57 mornfall For some files only though.
16:57 kowey did you catch the context for this discussion? (what's the right behaviour when the same name gets used for two different files at two different points in history?)
16:58 kowey my rough not-thinking-hard-about-it guess is that as a user, I certainly mean the file in my working dir
16:58 mornfall kowey: That's orthogonal.
16:58 kowey that foo means the most recent foo
16:58 mornfall Purely UI issue.
16:58 mornfall You need to select a file identity.
16:58 mornfall Once you do, annotate is impervious to filenames.
16:59 bsrkaditya We mean annotate over a directory.
16:59 kowey we do?
16:59 kowey ok
16:59 mornfall Annotate over a directory is obsucer.
16:59 mornfall obscure
17:00 mornfall But you always show the version you are annotating ("now" by default).
17:00 mornfall And info pertaining to the lines (files) that exist in that version.
17:00 bsrkaditya If you run annotate tests/
17:01 mornfall So filenames are of the same relevance in directory as line content is to normal annotate.
17:01 mornfall It's totally borked. :P
17:02 kowey bsrkaditya: you aren't really expected to deal with the total borkage of annotate over directories
17:02 kowey but do document whatever choices you make
17:02 mornfall "unknown" is just another way to say "BUG"
17:02 kowey making clear what the issue is and what solution/decisions you've chosen
17:02 bsrkaditya Okay/
17:02 mornfall It's probably a trivial bug.
17:02 kowey could be worth writing a test case too to illustrate the problem
17:02 mornfall (Most are. :-P)
17:03 kowey bsrkaditya: so perhaps back to the meeting stuff?
17:03 bsrkaditya Yes.
17:03 kowey what are you hoping to work on these next two weeks?
17:04 owst The annotate-going-all-the-way-back issue wasn't fixed, it involved playing with the mean-looking LCS implementation
17:04 bsrkaditya feature parity between patch-index-changes,patch-index-annotate and changes,annotate.
17:05 kowey what's missing?
17:06 bsrkaditya many things. :-)
17:06 bsrkaditya you pointed out changes on multiple files.
17:06 kowey do you know what's missing exactly?
17:06 bsrkaditya no. :-)
17:06 kowey or is it a matter of figuring out what's missing?
17:07 bsrkaditya I have to figure it out.
17:07 kowey ok, will write this down and also some of the things we noticed
17:08 kowey ok
17:09 kowey hmm, so next item on my agenda?
17:09 kowey (or do you have something you'd like to expand on?)
17:09 bsrkaditya nothing comes to me off the top of my head.
17:10 bsrkaditya Oh yeah, I pulled from screened
17:10 bsrkaditya a while back
17:11 bsrkaditya I just checked, I am uptodate wrt screened.
17:11 kowey lot's of refactoring/cleanup work lately, if I understand correctly
17:11 kowey really nice to see people getting into it
17:11 bsrkaditya yes.
17:11 kowey so what about this pulling from screened?
17:12 bsrkaditya patchindex repo should not get outdated?
17:13 kowey oh you were asking if patchindex should try to keep up?
17:13 kowey it sounds like it would save you a lot of pain later on if you did so
17:14 bsrkaditya I "assumed" it has to keep up. :-)
17:14 kowey but it may involve you rebasing your work from time to time
17:14 kowey is that something you'd be ok with doing?
17:14 bsrkaditya It is fine.
17:14 kowey in general, you're going to want to keep an eye on the question “how is this supposed to be merged back to darcs?”
17:14 kowey the important thing here being how to deliver your patches in a way/schedule that makes it possible for people to review and apply them
17:15 kowey not saying you need to have a plan right away
17:15 kowey just be sure that's on your mind somewhere
17:16 kowey so… next on the agenda?
17:16 bsrkaditya okay/
17:17 kowey thanks for that first blog post
17:17 kowey it's a decent start, and a good illustration of done being better than perfect; i'd maybe think a bit more about writing for general darcs users
17:17 kowey now, i sometimes think that sometimes it's easier to work with hard constraints than soft constraints
17:18 kowey so my question to you
17:18 kowey is if you'd be able to commit to having a blog post up before 1600 UTC every Friday?
17:18 kowey or if that would be difficult for you to keep up with?
17:18 bsrkaditya I think I prefer to do the blog post on weekends.
17:18 kowey ok
17:19 kowey So say 1600 UTC on Sunday?
17:19 bsrkaditya Allright.
17:19 kowey OK, let's see how this goes (and now you have #darcs as a witness to this commitment)
17:19 kowey I have a guess that actually this makes it easier, not harder
17:19 kowey because now you don't have to think about it
17:19 kowey you just do it :-)
17:20 bsrkaditya yes. :-)
17:20 kowey uh right, two more things on my agenda
17:20 kowey do you have anything that you need help with?
17:21 bsrkaditya I will ask when time comes. :-)
17:21 kowey ok, you mentioned earlier that you could use some help understanding type witnesses
17:21 kowey is that still the case?
17:21 bsrkaditya I think I will leave that. :-)
17:22 kowey ok, well hopefully the others here can say useful things if that time comes again
17:22 kowey final thing
17:22 kowey so, how was this week for you? :-), :-|, or :-(?
17:23 bsrkaditya The middle one.
17:23 kowey care to elaborate?
17:23 bsrkaditya I have a few leftovers, but I can do them over the weekend.
17:23 mornfall What have I done?
17:23 kowey hmm, ok
17:24 owst The question is, what *haven't* you done :-p
17:24 kowey will be interesting to see if how you allocate your :-)/:-|/:-( evolves over time
17:24 kowey bsrkaditya: ok, I think I'm going to have to wander off now
17:24 bsrkaditya bye
17:24 kowey thanks for this meeting! see you in a couple of weeks!
17:25 kowey (and as usual, good luck!)
17:27 owst Ah, darcs unrevert, you save me again
17:27 owst :-)
17:27 kowey better to build forgiveness into the system than confirmations
17:28 kowey the Principle of Noncatastrophic Oopsies
17:29 bsrkaditya left #darcs
17:39 kowey alexsuraci reports that darcsden should be up again
17:39 kowey VPS issue
17:40 teratorn vps issues :(
17:53 kowey owst, I think the twisted msg is related to buildbot
17:53 kowey mornfall, perhaps something you could help with?
17:54 mornfall Oh, buildbot.
17:54 kowey sorry, I'd noticed a while back
17:54 mornfall Did I mention we switched for hydra for divine?
17:54 mornfall to hydra*
17:54 kowey but was being badly behaved in the ostritch-head-in-sand way
17:55 kowey got a link for that?
17:55 mornfall kowey: http://anna.fi.muni.cz:3000/
17:55 kowey hmm, using Duck Duck Go seems to have its disadvantages
17:55 kowey like, not finding stuff
17:55 kowey how are you guys finding the experience?
17:56 mornfall I especially like http://anna.fi.muni.cz:3000/build/630 and http://anna.fi.muni.cz:3000/build/635
17:56 mornfall One big question mark is Windows.
17:57 kowey sm: I've now created den.darcs.net too, points the same as hub.darcs.net
17:57 kowey I forgot why I mentally shot down den.darcs.net
17:57 kowey it wasn't to imitate github, was something else
17:57 mornfall Den Darcs
17:57 kowey but it seems really good now, weird guy, this past-eric
17:57 mornfall like Den Haag. :P
17:57 kowey ooh, packages
17:58 kowey is that because hydra does this sort of stuff practically for free?
17:58 kowey like tick the 'build me an rpm please box'?
17:58 mornfall kowey: Yes.
17:58 kowey ooh
17:58 mornfall Well, you need a specfile obviously.
17:58 mornfall For RPM.
17:58 mornfall It's using checkinstall for deb but I need to fix that anyway.
17:58 mornfall What is not clear is if, and how, it would be possible to run Windows VM-based builds.
17:58 kowey should we disable the buildbot hooks and just put it on our list to switch to hydra one day?
17:59 mornfall kowey: Don't know.
17:59 kowey any chance buildbot.darcs.net could come back up?
17:59 darcsbb joined #darcs
17:59 kowey oh hello there, darcsbb
17:59 kowey thanks, mornfall :-)
17:59 mornfall Seems it just died.
18:00 mornfall I have started up the buildmaster, if something breaks, call.
18:00 kowey will do
18:00 kowey see, simple problems, simple solutions
18:00 kowey head-in-sand is NOT a good tactic for dealing with life
18:00 kowey *sigh* :-)
18:00 mornfall (Or in case the slaves don't show up in a reasonable time. :-)
18:01 mornfall Well, things tend to rot.
18:01 mornfall Especially when not tended to.
18:01 kowey just thinking about my reaction to the darcs.net disk getting full
18:01 kowey (which led to the mail disruption)
18:02 kowey “oh, full? hmm, well, let's just delete some stuff”
18:02 kowey as opposed to something more constructive like, “hmm, maybe i should mention this to somebody”
18:06 mornfall There's a crisis going on.
18:06 kowey hmm?
18:08 mornfall A not enough time and motivation kind of crisis.
18:08 kowey what, where?
18:08 mornfall Nvm.
18:08 mornfall Bbl.
18:08 kowey see ya
18:21 donri_ joined #darcs
18:35 Guest82433 joined #darcs
18:39 schlaftier joined #darcs
19:02 darcscommitbot joined #darcs
19:48 alexsuraci joined #darcs
19:54 donri joined #darcs
20:00 thorkilnaur mornfall, Did you get a chance to look at http://lists.osuosl.org/pipermail/d​arcs-users/2012-April/026454.html?
20:04 mornfall No, sorry, I have never seen that mail.
20:04 mornfall I might have missed it, too.
20:28 schlaftier joined #darcs
20:33 nomeata joined #darcs
20:45 donri_ joined #darcs
22:02 donri joined #darcs
22:32 gh_ joined #darcs
23:07 donri joined #darcs

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