Camelia, the Perl 6 bug

IRC log for #darcs, 2012-02-18

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

All times shown according to UTC.

Time Nick Message
00:14 mornfall joined #darcs
01:58 kolmodin_ joined #darcs
03:09 intripoon_ joined #darcs
04:17 wagle joined #darcs
04:20 wagle hi..  never used darcs before (git expert though)..  if a repo has version 6.6.3 of happstack-server current, how do I extract version 6.5.*?
04:35 sm hi wagle. perhaps: darcs show tags; cd ..; darcs get REPO --tag 6_5_TAG REPO-6.5
04:36 wagle 20120216173513-8e95d-64e232197e​8404601a9b8e6857e1da47fde5a418 is what i'm looking for, i think
04:37 sm that looks like a hash, which is another way to select a patch. Usually matching on a tag or the patch name is easier
04:38 wagle yeah, got the hash from the web browser..
04:39 sm darcs help patterns gives some examples of complex matching
08:15 gal_bolle joined #darcs
11:27 owst joined #darcs
11:28 owst morning all
11:28 Heffalump 'lo
11:30 owst So, we're sprinting in Southampton?
11:30 Heffalump I'd hope you'd know if we were :-)
11:30 owst ;-)
11:30 Heffalump though that was the impression I had from kowey
11:30 owst Yeah
11:30 owst So, I had better think about accomodation
11:31 owst Also bsrk needs "accommodation details with a supporting letter from
11:31 owst the occupant confirming that you are able to stay there" whatever that is
11:33 Heffalump eek
11:34 owst Yeah, for visa purposes, apparently.
11:34 owst I wonder if a booking for a travel lodge or whatever is enough :-s
11:39 stulli joined #darcs
12:13 owst Heffalump: do you have a clearer idea about your availability for the sprint?
12:28 owst This company look like good value for a nice-looking place: http://www.town-or-country​.co.uk/Clench-Street.html
13:07 iago joined #darcs
13:19 Heffalump owst: I'll check. What are the dates?
13:24 owst 30/03 - 1/04
13:54 awagner joined #darcs
14:18 awagner joined #darcs
14:47 awagner joined #darcs
15:52 kowey joined #darcs
15:53 kowey hey owst, happy to make 31 March - 1 Apr official for the sprint?
15:54 owst Hi kowey!
15:54 owst Not the Friday too?
15:54 kowey if Friday is possible too, that's great
15:54 owst Either way is fine, the room's booked
15:54 kowey so 30 March to 1 April it is then?
15:55 owst So far it's me, you, iago, bsrk (subject to visa stuff) and Heffalump potentially.
15:55 owst Yep :0
15:55 owst :)*
15:55 kowey cool!
15:55 kowey thanks for organising this
15:55 kowey now that I know a bit more about Brighton, we will hopefully be able to do one here someday :-)
15:55 owst Cool :-)
15:56 owst It's not too long-a trip for me if we do ;-)
15:56 kowey I suspect I'll even be able to travel without going through Brighton
15:56 owst Accommodation is the next thing to sort out.
15:56 kowey oh yes, thanks
15:56 kowey totally forgot about that
15:56 owst There's no Hostels I can find, but I did find a good-looking place earlier
15:56 owst For not *too* much
15:57 kowey so
15:57 kowey we're going to have 2 sprints actually
15:57 owst ooh
15:57 kowey unfortunately not in synch
15:57 owst :(
15:58 kowey owst: do you have a capacity limit?
15:58 owst Ah Cordoba too, cool
15:58 kowey just wondering how much to advertise
15:58 owst kowey: you mean the room?
15:58 kowey yeah
15:59 kowey I'm happy with a small sprint in principle
15:59 kowey but if we can expand, I'll think about it (not likely we'll grow that much, but there may be some people in the UK who are interested)
15:59 gh_ hi
16:00 owst kowey: I'm trying to think, but the room can probably hold ~30 :)
16:00 owst hi gh_
16:00 kowey wow! what about internet access and all that?
16:01 kowey http://wiki.darcs.net/Sprints/2012-03
16:01 owst It's a mini conference room
16:01 kowey also, is there a standard directions page I could link to?
16:01 owst yeah, will find one
16:02 owst Internet access *should* be fine (wireless, not sure about wired), but I will confirm
16:02 owst Anyone who can use eduroam will definitely be fine though
16:03 owst So, I'll just need to check for non-uni people
16:06 * kowey will hopefully better resist the blabla urge this time
16:07 kowey although one thing that *might* be interesting to try is to throw up a kanban board (have no idea how this sort of stuff works, but the pictures look neat)
16:07 kowey just a visualisation, you-are-here
16:07 owst The main thing is a "what next" discussion/decision
16:08 owst IMO
16:08 kowey sounds like a great food/drinks conversation to have
16:09 owst Yeah
16:09 kowey so! does http://wiki.darcs.net/Sprints/2012-03 look like it's in a tweetable state?
16:09 owst Directions need adding :)
16:10 owst (just finding them)
16:10 kowey If so, I'll hold off till tomorrow as I've spammed people already
16:10 kowey ah right
16:10 owst http://www.ecs.soton.ac.uk/about/getto.php
16:10 owst The room is on the ground floor of B59, parallel to the reception
16:11 kowey accomodation too
16:11 owst I wonder if I can get a Google maps streetview of the building entrance, so people know where they are!
16:11 Igloo I can't read that URL without thinking ghetto
16:11 owst Yeah
16:11 owst Heh, Igloo yeah, it is a bit ;-)
16:11 kowey ghetto is an example of a process called semantic bleaching, apparently
16:11 owst :-s
16:12 owst kowey: http://www.town-or-country​.co.uk/Clench-Street.html was what I found earlier
16:12 owst it was
16:12 owst ~ £250 for 4 people 2 nights
16:12 owst IIRC
16:13 Igloo owst: Do you have a postcode of the location?
16:13 owst (sorry for many newlines in that sentence)
16:13 owst Igloo: Of uni?
16:13 kowey is that a package deal of some sort?
16:13 Igloo I think so
16:13 Igloo Of B59, if I followed correctly
16:13 owst Ah, right gotcha :)
16:14 owst Yeah, one sec
16:14 iago owst, did you know about a cheap place to stay during such days?
16:15 owst That's what I'm trying to find at the moment, I'll need to ask around for any hostels - googling didn't tell me much
16:16 iago yep
16:16 iago the best way is walking :P
16:19 kowey oh I see, Clench St is a house
16:19 owst http://goo.gl/F0l2r is a streetview of the building in question
16:20 kowey and you book the whole house with its 4 rooms (fits 7 people)
16:20 owst kowey: yeah, it's the cheapest option (and nicest) so far...
16:21 Igloo Hah, NRE wants me to change at Reading West (presumably due to engineering works). Can't be often you see that.
16:22 kowey booking the house would be great if we actually did get quite a few people
16:22 owst Igloo: from Cambridge? :)
16:22 Igloo No, Oxford
16:22 owst Ah, right, that makes a bit more sense
16:22 owst I've never heard of Reading West ;-)
16:23 owst Mind you, I've only been through Reading a few times
16:24 Igloo It's the last tiny stop you'd go flying through seconds before you get to Reading  :-)
16:24 owst Ah, right, one of those
16:24 owst Reading is in uber-redesign state at the moment I think
16:25 Igloo Yup
16:25 iago owst, how much it costs?
16:25 kowey we own darcs.org by the way
16:27 owst kowey: looks like 30/31 are the only nights free on that specific house atm
16:27 owst which is lucky
16:27 owst iago: £220 apparently
16:27 kowey should we just grab it?
16:28 owst They seem to have a few properties, I'll check the location in terms of busses/walking places
16:28 owst Hmm, "busses" doesn't look right
16:29 iago ... not aware of the plural of "bus" ?
16:29 iago :P
16:29 kowey the Chicago accent pronounces it "bosses", so if you hear it without context...
16:29 kowey traffic is "jammed salad"
16:29 owst haha
16:29 kowey I vaguely remember they are sort of being the vanguard in this vowel shifty thing
16:29 Igloo By which they probably mean jelly salad, or something!
16:30 * kowey will randomly spool out half-remembered bits of undergrad courses
16:30 kowey "she walked 3 blacks down the street"
16:31 kowey and (sorry, last memory fragment spewage) despite television, regional accents are diverging
16:32 Igloo Oh, that surprises me
16:32 Igloo In the US, or the UK, or both?
16:32 kowey TV is one way is the explanation
16:32 kowey need people who talk back to get convergance
16:32 kowey think it was a US context that I heard this
16:33 kowey would have to whip out a sociolinguistics text or something for the UK
16:33 Igloo I wonder if that was also the case before they let people with non-London accents on TV
16:37 owst Argh, trying to translate pdf bus routes into google maps is a pain
16:39 kowey Igloo: may be of interest, http://www.gla.ac.uk/media/media_70544_en.pdf
16:41 Igloo kowey: ta
16:41 kowey oh sorry, not really addressing the question
16:41 kowey but talking about a feature (th fronting, f-like ths?) in Glasgow
16:41 kowey via Eastenders TV-Cockney
16:48 owst kowey: http://www.town-or-country.co.uk/Westgate.html same price, looks like a better bet in terms of "near the bus route"
16:49 kowey updated
16:50 shenshei joined #darcs
16:52 owst probably a ~10 min walk to the bus stop, which has buses going down until nearly midnight, so perfect
16:53 kowey I think once there's 2 of us we should probably just book
16:53 owst Yeah
16:54 owst http://goo.gl/jd1HJ map of house to bus stop, easy :)
16:55 owst http://www.unilinkbus.co.uk/pdfs/networkmap.pdf bus network map, stop is at the bottom, 'High Street' and is served by the U1 and U6 routes
16:55 kowey ah, brilliant
16:55 owst 'Highfield Campus' is the main interchange for the bus, and is conveniently at uni.
16:55 Igloo If you're going to the big uni blob, then you've already walked about 1/3 of the way there by that point
16:56 kowey if N metres may be simpler to just walk it
16:56 owst That network map is not to scale :)
16:57 * Igloo was looking at the google map
16:57 owst Ah, right :)
16:57 owst You sure? ;-) http://goo.gl/q0gm6 is the walking route to uni
16:58 Igloo Ah, that's a different university blob.
16:58 Igloo Is so171bj the postcodeI want?
16:59 owst Yeah, that's the main uni postcode apparently - it doesn't accurately locate our building though
16:59 Igloo But it's within 1min on foot from it?
17:00 * owst looks where it actually places it on the map :)
17:01 owst Yeah, so it's just across campus, if you're looking on google maps, the building site below 'University Sub Post Office' is where the building is now.
17:01 * owst realises the map data for uni is fairly out of date.
17:03 iago so, any thing about Darcs future direction was clarified during the last 2 months ?
17:03 raichoo joined #darcs
17:04 owst iago: not really. I guess the sprint is the time to do it.
17:04 iago some features/changes planned?
17:04 owst It's a pity mornfall can't make it.
17:05 iago and Heffalump is not sure to attend if I understood correctly
17:06 owst He didn't say, but maybe not, which sucks too
17:07 kowey iago: you're asking because you'd like to be able to prepare a bit more in advance for the sprint?
17:07 Igloo owst: Are you looking for accom there because it's near the station, or because that's where cheap accom is, OOI?
17:08 owst Igloo: near the bus and I can't find "better" accom yet. Where better is cheaper/convenient
17:08 iago kowey, that's one of the reasons, yup
17:08 owst Igloo: and near the station is handy too.
17:09 owst Well, the U1 bus goes to the station and to that bus stop I mentioned earlier too, which helps
17:09 kowey iago, so personally I've been out of the loop for enough that I can't say much that's constructive
17:09 owst I think there needs to be consensus on direction
17:09 kowey iago, what would be a high priority for me is to think about how to tackle GSoC this summer
17:09 owst Hmm.
17:09 kowey I suspect the general view is that we need to work on interoperability
17:10 iago kowey, so, instead of improve darcs
17:10 owst I don't know, I'm not convinced it's *that* important
17:10 kowey oh ok
17:10 owst I think "make Darcs better" is more so
17:10 iago make darcs a way to access other CVS.... ?
17:10 iago :P
17:10 kowey well ok so we need to figure two things, it seems
17:10 owst Especially since without a "direction", working on bugs feels slightly aimless, since it may be "planned" to remove a bunch of the affected code.
17:11 kowey what's our general strategy? (previously it was: make darcs faster to buy time, then fix the code, then do darcs 3)
17:11 owst I.e. the "conflictor missing case" bug - it's been there for years, but nobody has felt inclined to invest the time in sorting it out
17:11 kowey and the other thing is - what are we going to put our attention on in particular?
17:12 kowey what would make darcs better for me is to start figuring out our conflicts story a bit more (from a UI standpoint)
17:12 owst Yeah.
17:12 owst big time.
17:12 gh_ think about using (part of) what mornfall did during last year's summer of code?
17:12 kowey yes!
17:13 kowey do we a have a list of the sort of big questions we've not been asking ourselves?
17:13 kowey one of them is how to go forward from prim3?
17:13 iago kowey, conflicts story is very ... delicate
17:13 owst But then, I also wonder if it's worth putting time into thinking about conflicts in a new way without having some sort of "proof" of correctness - i.e. camp
17:14 kowey I think we should probably avoid driving ourselves crazy with mental dependency thinking
17:14 kowey pick something we feel we can contribute to
17:14 owst kowey: no list I know of
17:14 owst would be useful though
17:14 kowey look at the bigger picture
17:14 kowey then chip away at the thing we feel we can contribute to
17:14 owst Yeah, we need to know what the big picture/direction is though
17:14 kowey if we get some clearer UI stories, it may help us understand what we need the theory to do
17:14 kowey or it may not
17:15 kowey just we don't know until we try
17:15 kowey and if we let ourselves get too hung up about the possibility of the work leading nowhere, we just waste time running from tree to tree wondering which one to bark up
17:15 owst Yeah true.
17:16 kowey I'm a believer of the idea that this mental stuff is messy; you can't draw nice graphs like good theory -> ui
17:16 iago kowey, I would appreciate very simple English
17:17 owst Yeah, we need a simple bullet pointed list ;-)
17:17 kowey that the sorts of things you can think about are so interrelated
17:17 kowey that you get these sort of feedback loops
17:17 kowey so thinking a bit about UI raises theory questions
17:17 kowey and vice versa
17:17 kowey so you start somewhere
17:17 iago yep
17:17 kowey and deepen
17:18 kowey iago, sorry about my trees and barking idiom
17:18 kowey do you know the expression “barking up the wrong tree?”
17:18 iago kowey, what I don't know is "bark up"
17:18 kowey it's a visual idiom, think of a dog chasing a squirrel
17:18 kowey and the dog for whatever reason mistakenly thinks the squirrel ran up some tree X
17:18 Igloo owst: I definitely think you need some sort of correctness proof, although with a different approach it may be much simpler
17:19 kowey so you see him with his paws on the trunk, barking like hell
17:19 kowey bark bark bark
17:19 owst Igloo: agree.
17:19 kowey all the while the squirrel is laughing at him from tree Y
17:19 iago kowey, I got it, thanks
17:19 owst I'm afraid I couldn't get far into the coq stuff
17:19 iago coq?
17:19 kowey iago: yeah, english and its post-verb particles (eg. call _up_ a friend)
17:20 owst iago: the camp stuff
17:20 owst specifically Igloo's theory document
17:20 iago I don't know if I'm lost, but please throw away any idea about using Coq for anything
17:20 iago no offense
17:21 owst Perhaps
17:21 iago Coq is of theoretical interest, but any non-trivial problem requires too much experience
17:22 owst The one thing that gets me is that it's ok having a coq proof that ensures the underlying idea is sound, but then surely you can just introduce a bunch of errors when implementing?
17:22 iago I think that to make sure we don't screw up
17:22 iago there is a point between do-nothing (Darcs) and Coq (Camp)
17:22 owst Yeah, quite possibly, but what is it :-)
17:23 iago good testing, some model checking
17:23 Igloo owst: Well, (a) those are just bugs you can fix when you find them, much the same as any other bug, and (b) the code is actually very simple; it's just the proof that's large
17:23 owst plus properties encoded in the types
17:24 kowey I think one question that's important here, iago (here meaning for the sprint)
17:24 owst Igloo: yeah, true I guess
17:24 kowey is what are *you* interested in working on
17:24 kowey not saying it's “every man for himself”, just this sort of bottom-up input helps shape the direction
17:24 kowey we can only go where people will be intrinsically motivated to go
17:25 owst kowey: definitely, but we do need to also agree on a general direction
17:25 Igloo I currently intend to finish my proof, even if darcs goes a completely different direction
17:25 owst Igloo: yeah, I think it's a good idea
17:25 kowey owst: agreed
17:25 kowey owst: I meta-agree
17:25 owst heh
17:25 iago kowey, yeah I know
17:26 kowey so you should probably remind us
17:26 kowey what do you fancy working on during the sprint?
17:26 iago I think that I already say my point
17:26 kowey we could still talk about the more general direction
17:26 iago I'm more motivated in "re-born" Darcs for v3
17:26 owst Yeah, me too.
17:27 kowey OK
17:27 gh_ yes
17:27 Igloo FWIW, I think the two major roadblocks with camp are that the proof isn't complete (although I'm fairly confident that the theory is correct based on how far I've got), and that I don't know how conflict marking should work
17:27 iago I think current conflict handling is too messy, and that has to be fixed
17:27 kowey I tend to care more about UI issues, personally
17:27 kowey and be happy to let other people think about the hard stuff :-)
17:27 owst We're never going to convince people that darcs is worth using again, otherwise.
17:27 iago kowey, Darcs can be proud of have a good UI
17:27 iago I think
17:27 owst Yeah, I think so too
17:27 kowey iago: so what I'm saying is that
17:27 kowey iago: I'm quite happy to follow the consensus there
17:28 kowey if the community pushes for darcs 3 thinking, I'm not going to have grounds to have a counter-opinion really
17:28 * Igloo thinks that darcs would benefit from a rewrite, which it could do by using fresh code for darcs3 stuff, and calling the darcs2 lib functions when dealing with old repos
17:28 kowey I'll just be "OK, let's see how I can facilitate us shifting to darcs 3"
17:28 owst Igloo: interesting, you mean conflict marking in terms of 'mark up the working tree when there are conflicts' or something else?
17:28 owst Igloo: I have a feeling that I agree
17:28 Igloo owst: Yes
17:29 Igloo owst: I have no idea how darcs conflict marking actually works
17:29 owst but that we need to be careful with the "version2 rewrite syndrome" or whatever the Brooks' term is
17:29 owst Igloo: me either, I guess magic ;-)
17:29 owst I'm not sure how a lot of Darcs works :-)
17:30 Igloo I understand the rest, although I'd have to look up some of the harier details
17:30 owst kowey: 'Hey, darcs 3 is easier to use than darcs2 and actually works!' </sarcasm>
17:30 iago nobody knows
17:30 iago :P
17:30 owst This is the problem.
17:30 kowey actually working is kind of important from a UI standpoint
17:31 owst I think I'm fairly confortable in my understanding of the basics, perhaps not things like the implementation of commutation of Conflictors
17:31 iago yep, seems everybody would like to focus on v3
17:31 iago and everybody agree that confictors are a thing "to change"
17:31 owst I think I'd like to hear input from Heffalump and mornfall too ;-)
17:32 iago "to rewrite", "to simplify", ...
17:32 iago but then there are two major and different views
17:32 iago Heffalump is going to say about rewrite conflictors
17:32 mulander hi all
17:32 iago he has an idea but no time to explain it
17:32 iago mornfall will say that he prefers to keep conflicts out of patch theory, as in the E/D idea
17:33 owst hi mulander
17:33 kowey although we do now have a http://wiki.darcs.net/Ideas/Graphictors stub
17:33 kowey heya mulander
17:33 owst Yeah, I started that, then forgot about it ;-)
17:33 mulander interesting discussion I've been following for the past 5-10 minutes
17:34 iago so, who is going to win ?
17:34 iago Heffalump idea could be great, but if he cannot explain it to humans it is valueless
17:34 darcsbb_ joined #darcs
17:34 Igloo He explained it on IRC fairly recently
17:35 owst iago: I think Igloo understands it
17:35 owst basically anyway
17:35 kowey the graphictors idea is searchable on the IRC logs
17:35 iago Igloo, a was present in some discussion, but I cannot consider that an explanation
17:35 Igloo It's basically just a different representation of the set of conflicts in a conflictor
17:35 kowey http://irclog.perlgeek.de/search.pl?cha​nnel=darcs&amp;nick=&amp;q=graphictors
17:35 iago an IRC discussion is not an explanation
17:35 iago in general
17:35 iago it's messy
17:35 iago and not well structured
17:36 owst iago: yep sure, but it's not that different to the current implementation
17:36 iago Igloo, yeah, that's something I knew
17:36 iago since the last Sprint
17:36 iago :P
17:36 owst that's a fair point
17:36 shachaf_ joined #darcs
17:37 gbeshers joined #darcs
17:37 byorgey joined #darcs
17:38 alpounet joined #darcs
17:41 mulander from the list of ideas as a user I mostly like the idea of: http://wiki.darcs.net/Ideas/GraphicalInterface
17:41 mulander and http://wiki.darcs.net/Ideas/FancyRecording
17:41 iago I have a very basic idea of what graphictors are
17:42 iago just use graphs to represent conflicts
17:42 iago and that is supposed to avoid problems of duplicates
17:42 iago and to be more efficient
17:42 owst Well, duplicates within a single conflictor
17:42 iago and when I read Heffalump talking about them
17:42 iago I read lot of "I think"
17:43 owst you still have duplicates globally, AIUI
17:43 owst iago: well, that's because it's not written down ;-)
17:43 iago btw, I think a major problem of conflictors it is not if they are efficient or not
17:43 iago is that nobody understands them
17:44 Igloo I understand them
17:44 iago and there is no answer to "can you implement it in a clear and comprehensible way"
17:44 owst iago: I understand them on a basic level
17:44 iago Igloo, ok, so please fix the famous bugs that exist
17:44 owst I *don't* understand the implementation in Darcs at present
17:44 Igloo iago: I believe I have done. I need to finish the proof to be sure.
17:44 iago Igloo, Darcs code
17:45 owst iago: which can be extracted from the proof
17:45 Igloo iago: I don't understand
17:46 iago Igloo, http://bugs.darcs.net/issue2047
17:46 iago so, you understand conflictors
17:46 iago and their implementation
17:46 iago so maybe you are able to explain why recommute fails
17:46 iago and fix it
17:46 iago it's a kind of challenge
17:47 iago I reported that bug about a year ago
17:47 iago after improving Darcs unit tests
17:47 iago owst, sorry, "which can be extracted from the proof" ?
17:47 owst iago: we know... and we know that development has been very stagnant for the last year
17:47 Igloo I'm not interested in fixing that sort of bug in the darcs2 implementation
17:48 owst iago: I meant from the proofs, you can extract code that can be fitted into darcs
17:48 iago Igloo, the problem is that Darcs users care about Darcs implementation
17:48 iago not about a simplified model in Coq
17:48 Igloo iago: I'm working on darcs3, not darcs2
17:48 iago and nobody understand Darcs 2 conflictors
17:49 iago that was what I was saying
17:49 owst iago: it's not a simplified model!
17:49 iago yes, it is
17:49 Igloo "understand" and "interested in working on" aren't the same thing
17:49 iago just compare hunk commutation
17:50 iago for instance
17:50 owst Igloo: good point!
17:50 mulander honestly - adding something like camp-view I saw on one of the screencasts would be amazing for the current darcs.
17:50 mndrix joined #darcs
17:50 mndrix joined #darcs
17:50 mndrix joined #darcs
17:50 mulander that feature alone could help avoiding several problems/better understanding how things work under the hood.
17:50 iago Igloo, understand *your* implementation of conflictors as seen by *you* it is not the same that you understand *Darcs* conflictors
17:51 Igloo iago: I understand both
17:51 mulander I saw that feature here - http://www.youtube.com/watch?v=iOGmwA5yBn0
17:51 iago Igloo, ok, I believe you
17:51 Igloo although camp conflictors are fresher in my mind
17:52 mulander 2:51 in that video - camp-view is being run
17:53 owst iago: but clearly, darcs conflictors are broken!
17:53 mulander why there is no darcs-view for the current implementation?
17:53 iago owst, sure
17:53 iago :P
17:53 owst mulander: because nobody has written one!
17:54 iago kowey, owst: anyway, this conversation always end.... bad. My point is just that there are two different opinions about how to manage the conflictors stuff, and no consensus
17:56 iago and IMO that stops every conversation about Darcs3
17:56 iago and it seems that nobody want to vote for one of those opinions
17:56 iago well, I did
17:56 iago but no other did
17:57 iago at least, clearly, or I missed it
17:57 mulander I am in no position to vote about darcs future :/
17:57 owst iago: I'm not clear on what the options are :)
17:58 iago owst, I am
17:58 owst What are they?
17:58 iago theoretically infinite, but in practice 2
17:58 owst Graphicators and...?
17:58 iago a) Heffalump idea, spend time working on it and see if it really solves the problems
17:59 iago b) mornfall idea, something like E/D, the key idea is that conflicts are handled outside patch theory
17:59 Igloo iago: what was your vote?
17:59 iago b)
18:00 Igloo Isn't the vote of mornfall, Heffalump and myself also clear?
18:00 iago of course
18:00 iago lol
18:00 * kowey does not feel comfortable voting on things he doesn't understand
18:00 iago I was referring to owst  and kowey for example
18:00 mulander what stands in the way of parallel research?
18:01 Igloo iago: Are you prepared to spend time on implementing mornfall's idea?
18:01 owst b) is appealing since it is simpler (?) to understand, but I'm not sure...
18:01 kowey that seems to be the current default
18:01 mulander if Heffalump want's to try a) and you and mornfall want to try b)?
18:01 kowey I don't see why it's a problem
18:01 kowey well, yes it's a problem in that the lack of consensus is perceived as blocking progress on either front
18:01 iago mulander, Heffalump has no time to write it down, so imagine to code it
18:01 owst mulander: nothing much I guess, but time is seriously limited
18:01 owst kowey: that may well just be my feeling though
18:01 kowey the thing that still matters at the end of the day
18:02 owst others may be happy :)
18:02 iago Igloo, I am prepared to spend time on anything with is simple to understand
18:02 kowey is what are *you* prepared to contribute to
18:02 kowey and what are you prepared to spend energy advocating
18:02 iago and if the "author" takes the time to write it down
18:02 owst yeah, talking about doing is a lot easier than doing!
18:02 kowey you can't dictate successfully
18:02 kowey and you can't make people vote on things
18:02 Igloo If you agree with me on a rewrite being necessary, then whichever approach you take you basically need to start from scratch, so we need an implementation of at least one of them
18:02 kowey and expect them to be binding unless you actually manage to get buy-in
18:03 owst Igloo: I certainly agree on that front.
18:03 Igloo The Haskell part of camp is one for something similar to approach a, incidentally
18:03 kowey ok so if you want to say “I vote B” then you need to be prepared to contribute to a prototype
18:03 kowey in some way
18:03 iago kowey, the point of "voting" is that noone has time to write some PoC or something, and both approaches are "promising" but no clearly good
18:03 kowey (likewise with A)
18:03 kowey so the situation is “we don't know”
18:04 kowey ok, so how about C?
18:04 owst iago: true, but it's not necessarily possible to decide, because we don't know!
18:04 Igloo Although camp also has some things, like UI differences, that I think would make sense no matter which approach is taken
18:04 kowey why not start saying "we need prototypes"
18:04 iago kowey, that is what I proposed to build a prototype of E/D for the sprint :D
18:04 owst Do you understand all the implications of chosing B?
18:04 kowey because "without prototypes, we'll get nowhereº
18:04 iago s/what/why
18:04 kowey and once we establish that as a goal
18:04 kowey *instead* of picking a theory
18:04 kowey but creating a number of prototypes
18:04 kowey we can start talking about subgoals like say creating a common framework
18:04 kowey that the prototypes can share
18:04 iago owst, yeah the vote is not about "what is better", is about "what we should try first"
18:04 kowey so that we don't duplicate effort needlessly
18:04 owst I guess the thing to avoid is the prototype becoming the de-facto implementation
18:05 owst which is what I guess sort-of happened originally.
18:05 kowey darcs 1, you mean?
18:05 kowey (and 2)
18:05 kowey fine, I think iago working on a prototype of E/D is a fantastic idea for the sprint
18:05 owst maybe v2
18:05 iago owst, me? if I understand all the consequences? of course not
18:05 kowey maybe I'll even help if there's some way my feeble brain can contribute
18:06 kowey it's much easier to build a consensus when people can see things
18:06 kowey you can't demand one up front
18:06 owst How does the v3prims fit into the camp picture, Igloo?
18:06 owst v3prims work*
18:06 kowey sorry, don't mean to go off into lecture/rant mode
18:06 Igloo owst: What is the v3prims work?
18:06 owst mornfall's GSoC stuff from last year
18:06 owst kowey: you're not saying anything ranty ;-)
18:07 Igloo owst: I have no idea what that is. But if the prims satisfy the camp axioms then they can be used in the camp theory
18:07 kowey http://web.mornfall.net/blog/soc_reloaded​:_darcs_primitive_patches_version_3.html
18:07 raichoo joined #darcs
18:07 kowey the basic idea is to generalise the notion of a prim
18:07 Igloo The axioms being things like "if AB <-> B'A' then B'A' <-> AB"
18:07 kowey so that you have generic objects and things that hold objects
18:07 owst Igloo: I'd hope those axioms are satified ;-)
18:08 kowey which makes it easier to start thinking of arbitrary patch types
18:08 kowey I may be getting this hideously wrong, and invite mornfall to throw rotten fruit at me for being a bad mentor
18:08 Igloo kowey: Is the short version that it is a more efficient implementation of commute?
18:08 owst Yeah, it does help that
18:08 kowey and one nice thing about it is that you get commute definitions quite easily
18:08 kowey just define it generically
18:09 kowey no more of that, "oh no, I have to think of a way for the new X to commute with all the old Ys"
18:09 kowey (now I'm really not sure if I'm just making things up)
18:10 Igloo owst: Perhaps a better answer would be: If you can do it in darcs, you can probably do it in camp in much the same way
18:10 kowey easier extensibility is how I think of it, rather than efficiency
18:10 owst Igloo: cool, that's a nice way of looking at it
18:10 kowey oh, I forgot
18:10 owst maybe I should try looking at the camp code, rather than the doc
18:10 kowey there's also the fact that you store patch contents on disk
18:10 kowey and not memory
18:11 kowey so yes you get faster commute
18:11 owst kowey: I think that's a major performance-improver :)
18:11 owst And things like add-add conflicts go away
18:11 owst AFAIU
18:11 kowey http://web.mornfall.net/blog/​soc_reloaded:_progress_1.html
18:13 iago kowey, I can work on that, if mornfall does it as well
18:13 iago or someone else
18:13 iago but mornfall should be available at some extend
18:14 kowey well! you'll have to have that discussion with him
18:14 kowey personally, I would be able to relate to the feeling of not being able to keep up with another major commitment
18:15 kowey we still need to find a framework that lets us make friends with that Day Job Problem
18:15 kowey but! I do seem to remember mornfall saying
18:15 kowey that progress is motivating
18:16 kowey impact is motivating
18:16 owst Yeah, of course it is!
18:16 kowey so iago: if you can find a way to create the conditions for intrinsic motivation
18:16 stulli So camp is still an active project? The last time i visited the website it seemed to be quite abandoned.
18:17 kowey iago: eg by developing something that's just enough-usable to entice people
18:17 kowey you may be able to create a virtuous cycle where people start jumping on board, etc
18:17 Igloo stulli: Yes, it's still active. It did take quite a long break while I got married.
18:17 owst Igloo: you mean life gets in the way of coding?! ;-)
18:17 kowey (aside from the Day Job Problem, there is the Having a Life Problem)
18:17 Igloo Sad but true
18:18 * owst is married to the first few months of a PhD at the moment
18:18 iago kowey, I get your point
18:18 iago but note it is not something about *me*
18:18 iago Darcs is stuck
18:19 iago I'm just the one that is saying "we should move on"
18:19 owst indeed.
18:19 kowey well, yes it is about *you* (and *you* and *you* and *me*) :-)
18:19 owst I am too, by way of "let's decide on a plan" ;-)
18:19 owst anyway, brb, going to collect a pizza
18:20 kowey basically somebody needs to start doing leader-y things in an open source sense
18:20 kowey which means actively creating the conditions necessary for the project to succeed
18:20 kowey talking is part of the solution
18:20 kowey saying "we should move on" is part of the solution
18:20 kowey see, we've made progress thanks to you
18:20 kowey but there's more leader-y things we could all do
18:21 kowey there are loads of leader-y things I could have done (but did not)
18:21 stulli Igloo: Ah, i see. So the plan still is to merge camp + darcs somewhen in the future? Or will camp stay independent?
18:21 kowey it's very much a lead-by-example thing as far as I can tell
18:21 kowey start doing stuff
18:21 kowey and people will follow
18:21 kowey want patch review? start reviewing patches
18:21 iago kowey, as I said I'm very interesting in a prototype
18:21 Igloo stulli: It's not clear what darcs wil do in the future
18:21 kowey yep, so that would be wonderful iago-leadership
18:21 kowey getting it going
18:21 kowey not necessarily building a working prototype mind you
18:21 kowey just creating the conditions that lead to a prototype happening
18:22 kowey building a skeleton
18:22 kowey fleshing out some sketches
18:22 kowey making sure somebody is paying attention to discussions about prototypes on the mailing list
18:22 iago kowey, lead-by-example sometimes lead to do-everything
18:22 kowey iago: I hope you realise I'm partly talking about this because it's stuff I never figured out how to do
18:22 kowey iago: what I'm doing is grasping, grasping at something; I know there is a much more effective leadership style out there
18:23 kowey iago: it's one of those easier-to-recognise-than-do situations
18:23 kowey so I'm not suggesting a plan of action so much as relating my current mental framework for thinking about these things
18:23 iago kowey, well, I'll talk with mornfall when he appears
18:24 kowey so mistakenly or otherwise /me believes that open source leadership consists in creating-the-conditions-for-$desirable-thing
18:24 kowey up to you to decide if that framework makes sense or is just a bunch of empty talk
18:24 iago but I recall him saying that the conflictors-related positions seem irreconcilable
18:25 mulander well I can tell you from experience what made auroraux completely fail as an open source project
18:25 iago even to produce a "fork"
18:25 mulander and that was mostly - no clear milestones/goals
18:25 iago I know a fork is not going to happen because we are too few
18:26 iago but if I'm going to spend X big quantity of a prototype and nobody ends by caring about it
18:26 mulander generally ~15-20 people devoted time to the project but no one had clear goals to pursue even though a lot of time was available when considered in bulk
18:26 kowey vindication for the iago/owst we-to-consensus-up-a-direction
18:26 iago it's likely that I'll quit
18:26 kowey so mulander, one of things 2008-me thought he tried to do
18:26 kowey is to provide “narrative” as a service
18:27 kowey ie. keep consistently talking up one thing: we're going to work on darcs performance, we're going to work on darcs performance, then we're going to X, etc
18:27 iago mulander, heh, that reminds me some project...
18:27 mulander iago: I explicitly mentioned auroraux
18:27 kowey (not trying to defend or anything just tying this in to the notion of goals)
18:28 kowey (just the sort of idea of giving people a single thread to rally around)
18:29 mulander kowey: well I think it worked considering the recent performance improvements in darcs itself right?
18:30 kowey I think it helped for a while
18:30 kowey I think that partly due to a lot of bungling on my part
18:30 kowey and wasting energy on the wrong things
18:30 kowey I was not able to fight developer-attrition
18:30 kowey and that the momentum petered out
18:30 kowey and we somewhat lost our sense of narrative
18:31 kowey it's gotten a bit fuzzy lately
18:31 kowey I think personally we need a mission statement
18:31 kowey it's the sort of thing that mornfall would likely think is just more silly nonsense :-)
18:31 kowey but I need one-sentence thing that we can always come back to
18:31 kowey that we can hang every thing we do around
18:31 mulander well my observations from a couple of weeks spent here
18:32 mulander is that darcs seems to have a maintenance problem.
18:32 mulander as in from what I hear in this channel a lot of the code is not trivial to understand
18:32 mulander and that hinders changes - still that's an outsider pov
18:33 kowey yes, this has been one of mornfall's singular complaints
18:33 kowey actually, this was Stage 2 in the narrative, when I had much more fighting spirit
18:33 mulander and it seems like a sensible approach
18:33 mulander there is no benefit from getting more users
18:33 kowey it was at the juncture of "ok, we've made those short term performance goals”
18:34 mulander if we can't get more people to help with maintenance since more users = more bugs/requests to maintain/develop
18:34 kowey “now we need to start thinking harder about maintainability”
18:34 kowey but then we got bogged down in a lot of needless fighting
18:34 kowey and that just about killed me
18:34 mulander oh don't let those things get to you :)
18:35 kowey and so this summer we had one of those “ok, this hasn't worked? how do we do it differently”
18:35 kowey oh yeah, hard to get entangled
18:35 kowey the funny thing was for me that I wasn't even taking any of the sides
18:35 kowey but furiously trying to pump my fire hose
18:35 kowey eventually took its toll
18:35 kowey got a sense of futility
18:36 mulander well as someone wanting to work 'on' darcs as a new dev I would mostly look for:
18:36 mulander a) specs for the formats used (storage)
18:36 mulander b) flow/transformations performed on the data
18:36 mulander c) overview of the modular parts of the code
18:36 mulander d) how do they connect, what's their purpose
18:36 owst Yeah, it is a problem that things are generally specified in-code, with a few exceptions
18:36 mulander e) detailed documentation for specific parts
18:36 kowey (I should mention that the fighting was productive in some sense... it got us to seriously rethink our infrastructure)
18:37 kowey (so creativity of conflict)
18:37 kowey Great, that's good list
18:37 mulander generally I see 'software' as specific ideas having their own implementation
18:37 iago > kowey> it's the sort of thing that mornfall would likely think is just more silly nonsense :-)
18:37 lambdabot <no location info>: parse error on input `of'
18:37 owst mulander: was it you who was talking about an 'ada darcs'?
18:37 mulander there is only one 'git' but it may have several implementations - same with programming languages (python, jython, pypy etc)
18:38 iago well, mornfall thinks that document code is silly nonsense
18:38 mulander owst: yes, you are correct I am the one :)
18:38 owst Ah, ok cool, I couldn't remember who I spoke to
18:38 iago so I would not seriously consider his opinion in some areas
18:38 iago :P
18:38 iago I agree with you
18:38 owst I don't think mornfall thinks that documenting code is silly nonsense.
18:38 iago owst, I could quote him
18:39 owst Unnecessarily spending time commenting every little piece of code with specs/comments etc, maybe
18:39 owst but not in general
18:39 owst iago: ok...
18:40 mulander generally what I'm missing the most while trying to get into darcs is a,b,c from my list
18:40 mulander as d is quite quick to deduce from code
18:40 kowey mulander: don't give up on advocating for this
18:40 mulander and e doesn't matter without a,b and c
18:40 mulander brb
18:40 kowey here's a meagre effort at working in this direction: http://wiki.darcs.net/Internals
18:40 kowey one of the issues with darcs development is that the holes are everywhere
18:40 iago mulander, because people consider that document is silly nonsense
18:41 kowey documentation is non-existent
18:41 kowey code is hard to understand
18:41 iago mulander, and you should search through IRC logs and blog posts
18:41 kowey patch theory is broken
18:41 stulli mulander: That's a great list. As someone who just dives into darcs code i approve.
18:41 kowey performance sucks (less now)
18:41 owst kowey: w00t!! :)
18:41 kowey and so what we struggle with is a sense of oh-we-should-do-X-we-should-do-Y-we-should-do-Z
18:42 owst Yeah, kowey, that's what I'm thinking
18:42 kowey which is why I'm much more sympathetic these days to the grumpy-hacker school of thought advocated by mornfall
18:42 iago kowey, the problem of Darcs is that noone wants to start working in a direction that will eventually fix those problems
18:42 kowey which is that the more you talk about doing things
18:42 kowey the less you do them
18:42 owst iago: don't they?
18:42 iago owst, I don't think so
18:42 owst iago: that sounds like "no one wants to fix darcs" which seems stupid
18:43 iago ok, so we are tired to say "document is non-existent"
18:43 iago and I know that document all the existing code is boring
18:43 iago but if you agree with that, you should write down your ideas for new development
18:43 iago and have specs
18:43 kowey and see, that's what we try to encourage people to do
18:44 iago so, where is such documentation/specs for the recent work ?
18:44 owst What recent work?
18:44 kowey that's what all this wiki obsession is
18:44 kowey that's what all this where-do-I-file-X thinking is
18:44 iago prim v3 for instance
18:44 kowey it's actually about how do we create an environment that makes it easy for people to contribute documentation
18:44 owst We've discussed the lack-of-documentation for primv3 many times
18:44 iago yeah
18:44 mulander back
18:44 kowey how do we make it straightforward for people to find out where to hang their new documentation on the tree
18:45 kowey that's why we now have http://wiki.darcs.net/Ideas
18:45 iago kowey, in the case of mornfall that's not the problem I think
18:45 kowey that's why we now have http://wiki.darcs.net/Internals
18:45 kowey iago: well it's persuasion
18:45 kowey iago: and it's hard
18:45 kowey I'm not a good persuader
18:45 iago let me do a search
18:45 iago and I'm get some interesting quotes
18:46 kowey iago: I tend to think that you meet people where they are
18:46 kowey not sure if that makes any sense
18:46 iago 19:36mornfallComplaining about (lack of) documentation is the most favourite pastime in opensource, ever.
18:46 kowey but that you try your damndest to put yourself in the other person's shoes
18:46 iago mornfallI document what I want and how I want.
18:47 iago :)
18:47 kowey and then see how to meet your $desirableThing (eg. moar documentation) from that vantage point
18:47 iago that's the right direction to fix the documentation problem, right?
18:47 kowey how do we get $desirableThing if at the end of the day from your perspective, talk is wasted time
18:47 kowey from your perspective, if people want to know about the library, the source is right there to be read, thank-you-very much
18:47 kowey if at the end of the day, darcs hacking is supposed to be a fun hobby, and not some kind of job
18:48 kowey meet your interlocuteur there, where they are
18:48 kowey argh, sorry, being all rambly again
18:48 owst iago: mornfall has a point there.
18:49 owst And also, he's not advocating not writing documentation, just that it's always complained about.
18:49 iago kowey, but it is something hard to change
18:49 iago owst, well, the "I document what I want and how I want."
18:49 kowey so one concrete way here is to respect the “talk is cheap” world view
18:49 kowey (which I fail at because I talk a lot)
18:49 owst oh, right, I didn't realise that was a quote.
18:50 iago makes it a bit clear he does not care about other people claiming that his work is not comprehensible enough
18:50 kowey and not try to persuade a hacker to write documentation
18:50 kowey but actually start submitting patches
18:50 kowey not necessarily write the documentation yourself
18:50 owst iago: lots of complaints, not many patches...
18:50 kowey but to recognise what somebody who thinks that talk-is-cheap wants
18:50 kowey concrete meaningful action
18:50 iago owst, come on, don't say me that
18:50 kowey put up or shut up (again this is me trying to shift perspectives)
18:50 iago I heard that long time ago
18:50 owst Well, what can we do?
18:50 iago from kowey
18:51 iago and I see that contributing changes anything
18:51 mulander well here is my point of view as someone new
18:51 mulander I started using darcs a bit more often
18:51 mulander reported ~11 bugs
18:51 iago I showed that Darcs unit tests were VERY bad
18:51 mulander most of them are somewhat described - 'yep, it's a problem but we are not sure why darcs works that way'
18:52 iago and nobody care about that
18:52 iago I contribute with patches, that didn't made that other people started to write tests when contributing
18:52 mulander my question is - what prevents us to know because such responses indicate that the code itself is not enough to learn darcs itself.
18:52 * kowey listens carefully to mulander
18:52 mulander and if I hear that from a regular maintainer
18:52 kowey one aspect of darcs is that the original author moved on
18:53 mulander then as someone new to the code base - I am not encouraged to try and find out.
18:53 kowey and so the Second Generation is still (!) trying to learn the code
18:53 kowey lots of progress and cleaning up
18:53 iago mulander, right
18:53 iago my English skills are pretty bad, I don't express myself very good
18:53 mulander please don't treat this as a complaint - I am just presenting my pov
18:53 kowey darcs is project that has had to transition from being an essentially one-man-plus-fans-show
18:53 kowey to a team effort
18:53 owst iago: sure, I try to account for that
18:54 kowey only the one man is no longer part of that team
18:54 kowey (btw, I hope mulander that I'm not making excuses for darcs here)
18:54 iago but I already said that the Darcs general way of working stop other people to contribute
18:54 kowey because I very much need that new-blood perspective
18:54 mulander let's consider a simple example
18:54 mulander issue2123
18:54 iago no doc -> people see what there is and "quit"
18:54 mulander darcs doesn't inform users that it begins pulling patches on a lazy repo
18:54 kowey http://bugs.darcs.net/issue2123
18:55 mulander no one commented it on the issue itself
18:55 mulander but we discussed it on the channel
18:55 mulander came to the conclusion that 'downloading' msg prompt would be appropriate
18:55 owst mulander: yeah, good point, I know that I did that
18:55 owst I also complained about the same 'lack of feature' myself
18:56 mulander I understand that IO in haskell is not a trivial use case to plug in at random parts of the software
18:56 owst I have a feeling we mostly ignore the bugtracker, since there are a lot of bugs
18:56 mulander but we should be able to pinpoint a place where this should be correctl yhandled?
18:56 owst maybe, maybe I'm wrong.
18:56 kowey once upon a time, we had an issue manager
18:56 owst mulander: yeah, in theory.
18:57 mulander providing as little info as 'hey this could be done in File.hs X because Y decides on pulling patches'
18:57 mulander then someone new as me would have a place to grab onto and try to make it work.
18:57 owst mulander: yeah, I expect there is such a point
18:57 owst I can't tell you where it is though :-(
18:57 kowey right so separating the doing from the diagnosing
18:58 kowey and just making some progress, any progress to get closer to a diagnosis
18:58 mulander it was done correctly for issue 2126
18:58 kowey case studies!
18:58 mulander http://bugs.darcs.net/issue2126
18:58 kowey from real life (and recent to)
18:58 kowey *too
18:58 mulander here we diagnosed that there is a way to handle the colors properly on windows
18:58 mulander listed the api and an example that does it correctly
18:59 mulander pinpointed where the implementation should happen
18:59 kowey mulander: would pointing issue2123 at http://wiki.darcs.net/Internals/CacheSystem have been the right thing to do?
18:59 mulander the only reason why I didn't work on it is that ganesh marked this as somewhat non trivial as a starter task
18:59 owst kowey: that was what I was thinking
18:59 kowey that's a GSoC 2009 effort
18:59 kowey with (notice!) an attempt at building developer documentation to go with the work
18:59 kowey not as in comments, but as in overview-how-it-works
19:00 mulander kowey: yes and there is a bug I reported regarding the cache - that breaks it on windows.
19:00 kowey right, so concrete technical detail
19:00 kowey even if it's not a formal diagnosis
19:00 mulander this should be worked on if that list accumulates cache tasks
19:00 kowey but hey, maybe you could look here
19:01 kowey right now one "constraint" wrt bugtracker (for me)
19:01 kowey is that I simply get overwhelmed/discouraged by trying to make myself process bugs
19:01 kowey so I just systematically archive now
19:01 kowey ostrich style
19:01 owst kowey: yeah, there are lots and lots of bugs...
19:01 kowey I used to be very gung-ho and GTD on the tracker
19:02 owst GTD?
19:02 stulli gung-ho?
19:02 mulander kowey: generally bug trackers can either be a time sink or overview/info gathering space
19:02 kowey at least one change I think we've made recently is the need-foo split
19:02 kowey Getting Things Done
19:02 kowey bugtracker with a theory
19:02 kowey but the theory was wrong
19:02 kowey it didn't account for the non-GTD practioners wtf'ing their way through the tracker
19:03 kowey needs-action? uh, yeah it needs action; somebody better act on this bug!
19:03 kowey so the more recent 2012 split into more concrete things
19:03 kowey need-testcase
19:03 kowey etc
19:03 iago need-comments ? :P
19:03 iago ;)
19:03 mulander kowey: here is how I get into new projects
19:04 mulander 1. start testing and going through old bugs - doing test cases etc
19:04 mulander 2. report new bugs / talk with devs
19:04 kowey stulli: very enthusiastic and energetic
19:04 owst iago: sure, I think that's a valid state. I don't know about marking every patch on the tracker with it though ;-)
19:04 mulander 3. start looking over the code
19:04 iago owst, you people tend to pick what I say and put in the extreme
19:04 mulander 4. start implementing small tasks
19:04 mulander 5. go into general maint/code mode
19:04 owst iago: it was a joke
19:05 mulander for open source projects - I see the tracker as something that can keep context of the problem in one place
19:05 mulander each bit of information can help someone solve it now/a bit later
19:05 owst subtlety is completely lost over IRC :)
19:05 owst mulander: yeah, good point.
19:05 kowey the collective note-taking space
19:06 kowey what can I say on ticket X that will make the next guy's life a tiny bit easier?
19:06 mornfall mulander: As long as it's not misleading.
19:07 iago mulander, well I just pick a small portion of the code and start studying it
19:07 owst *a mornfall appears*
19:07 iago s/pick/picked
19:07 owst The Named/Prim/PatchInfoAnd distinction is a bit unclear
19:07 mornfall I have been around watching.
19:07 mornfall For a while.
19:07 mornfall Names been mentioned.
19:07 owst Ah, fair enough.
19:08 kowey :-)
19:08 iago yep
19:08 owst This is true.
19:08 iago basically by me, lol
19:08 owst and me, and kowey ;-)
19:08 owst that happens when you are a core contributor!
19:08 iago yeah well, I think I mention mornfall too many times
19:08 iago in comparison with kowey or you
19:08 iago that's why I said that
19:09 iago not a "how many times" competition
19:09 kowey mulander: OK so what I think we're talking about is how to create more favourable conditions for people to start working on the project from nothing?
19:10 mulander yeah, generally getting into the project
19:10 kowey and your example of the 1-5 workflow being how a new developer might approach it
19:10 kowey which we can turn around to use as material for figuring out how to ease people down that path?
19:10 mulander yes and those a-e elements I gave recently shows what's needed to help in that
19:11 owst Hmm, useful discussion tonight.
19:11 owst A pity that I have to go
19:11 kowey (for people getting lost, http://irclog.perlgeek.de/​darcs/2012-02-18#i_5169378 is what mulander is referring to)
19:12 kowey and http://wiki.darcs.net/Internals is our approximation of that
19:12 kowey and oh! how it falls short
19:12 kowey but the point is get better, not be good
19:13 mulander kowey: generally the end goal of the documentation should allow treating the whole thing like http, smtp etc
19:13 mulander I don't need to look at the source of sendmail or libcurl
19:13 mulander to implement something talking over http/smtp
19:14 mulander I can also spot bugs more easily
19:14 kowey create that can be described purely in English
19:14 kowey even if somewhat formalised English
19:14 mulander since if the implementation is the spec - the bugs are not obvious
19:14 kowey with abstract notation
19:14 kowey so in a word, we're missing specs
19:15 kowey (one way of sharpening the ahhh-our-documentation-sux thinking)
19:15 kowey kind of tempting for me to wonder if that's actually the one thing I should be throwing my energy at
19:16 kowey although now people are very motivated by the idea of darcs 3
19:16 kowey or if that's just another oh-we-should-work-on-X ADHD-style development symptom
19:16 mulander I think goals are nice.
19:16 iago kowey, I appreciate these conversations
19:17 iago but it looks like they are repeated over the time
19:17 iago without no result
19:17 mulander there are small tasks that people don't care enough to take care of
19:17 mulander but know how to fix
19:17 mulander detail them on the bugtracker and list on the wiki
19:17 * owst disappears
19:17 kowey we call that ProbablyEasy
19:17 mulander so someone with little to no knowledge can tackle, fix and test
19:18 kowey http://tinyurl.com/darcs-probablyeasy2
19:18 kowey what we still don't do well with probablyeasy
19:18 kowey is getting people up to speed quickly
19:18 kowey from zero to patch in 5 seconds
19:18 iago mornfall, so, are you interested in prototyping E/D ?
19:18 kowey another thing we don't do well with probablyeasy is how to move them down the funnel
19:18 kowey into probablyslightlyharder (not-an-actual-tag)
19:19 kowey so we have this gigantic gap
19:19 kowey from probablyeasy to OMFG
19:19 mulander well this one is marked as has-patch http://bugs.darcs.net/issue1371
19:20 mulander so what's happening with it?
19:20 kowey http://bugs.darcs.net/patch254
19:20 kowey (unfortunately it's still not easy to find the patch from the issue
19:20 kowey it's link buried down in the bottom
19:20 kowey iago: that's what happens when you talk about things with gaps in between
19:21 kowey iago: people forget things and repeat themselves
19:21 kowey iago: but notice we're making progress!
19:22 kowey notice the attempt at placing the emphasis on prototypes rather than picking a theory (perhaps I don't have buy-in from you on this yet)
19:22 mulander kowey: who reviews if an issues is still probably easy?
19:22 kowey notice mulander's contributions to the how-to-get-people-in problem?
19:22 mulander take this example: http://bugs.darcs.net/issue1725
19:22 kowey nobody has the job
19:22 kowey but what tends to happen
19:22 kowey is that a developer who has a feel for the codebase looks at the issue
19:23 mulander topic: + ProbablyEasy, UI 2010-01-09 17:54:37
19:23 kowey and senses it's probablyeasy just tacks it on
19:23 kowey (often me)
19:23 kowey (and often wrong)
19:23 mulander and a comment from mornfall on 2010-11-13.21:16:48
19:23 mulander which generally marks it as not really that easy
19:23 kowey ProbablyEasy sometimes explodes into ooh-actually-hard
19:24 kowey like there, I imagined that you could just do some sort of filtering on the level of Darcs.Commands.Whatsnew
19:24 kowey and I generally assume that anything I could imagine doing in Darcs.Commands.*
19:24 kowey is probably easy
19:25 mulander cool
19:25 mulander but it seems that the last comment says that it should be discussed
19:25 mulander how to treat recurse actions
19:25 mulander maybe a default flag etc
19:25 mulander and how to handle it in this case
19:25 mulander where is that discussion taking place?
19:25 kowey so there's a need for UI discussion
19:25 kowey do I understand correctly?
19:26 kowey in *theory* what people should do then is start launching discussions on darcs-users
19:26 kowey linking them back to the bts so that we can pick up the trail later
19:26 mulander well generally - a) problem not trivial - the naive approach is off due to performance constraints
19:26 mulander b) there is an ui discussion involved since it might require a standard flag across many commands and probably a similar implementation approach
19:27 mulander We don't differentiate the unknowns very well in existing
19:27 mulander code, from the knowns. So it could be still tricky, even if we agree on
19:27 mulander an interface.
19:27 mulander so c) how to implement discussion
19:27 mulander or how to 'solve'
19:28 kowey (the one difficulty I tend to experience most wrt bugs btw is the read-and-understand phase)
19:28 kowey getting over the stage of what-is-this?
19:28 kowey and so one antipattern is that I start piling up a list of things to-read-and-understand
19:31 mulander I handle issues in trackers in 6 phases. 1. Gather info. 2. Diagnose. 3. Find solution. 4. Implement. 5. Test. 6. Close
19:31 kowey this is how kowey-ca-2009? thought about bugtracking: http://wiki.darcs.net/BugTracker
19:31 kowey needs updating
19:32 kowey so I think what tends to happen is that I get stuck at stage 1
19:32 kowey and I think I (personally) could use help in learning how to do 1 more easily
19:33 kowey which I suspect may come in the form of learning a pattern-for-reading
19:33 kowey A. Look for X, write it down
19:33 kowey B. Look for Y, write it down
19:33 mulander so you are getting stuck in the gathering data phase?
19:33 kowey I'm not sure if I've identified the right phase
19:33 kowey I get stuck in the what-is-this-wall-of-text phase
19:34 kowey (eg, I'm still trying to read-and-understand that bug)
19:34 mulander here is what I do with the wall of text problem (based on my regular day job)
19:34 mulander 1. Did the client tell us what went wrong
19:34 mulander 2. What did he expect
19:34 mulander 3. What exactly did he do
19:35 mulander 4. Did he provide the data we regularly ask during diagnostics (logs, version numbers etc)
19:35 mulander 5. Where did the issue happen (machine, specific software running on it etc - something that might influence us).
19:35 mulander that's the 'did they tell me everything' phase.
19:35 iago mulander, join Darcs please :D
19:36 kowey you drill down a checklist
19:36 mulander kowey: yes
19:36 mulander if I have all 5
19:36 mulander I then go through phase 1 checklist 2 phase
19:36 mulander 1. Do I understand what they wanted to achieve?
19:36 mulander (if not - ask)
19:37 mulander 2. Can I reproduce the problem (without understanding it at this point)
19:37 mulander 3. Is the provided data related to the problem (clients often mix wrong data with the issues leading us to blind spots)
19:38 mulander 4. Is there something that I might need in this particular case (more data, questions)
19:38 mulander 5. Is this really a bug, feature - can I categorize this problem
19:39 mulander that's the end of phase 1
19:39 mulander at this point I move on to diagnostics
19:39 mulander where I go through another checklist
19:39 mulander same for each phase
19:40 mulander the checklists I have at work are a bit more extensive
19:40 mulander but strictly related to our specific product.
19:40 kowey I find it reassuring that you go through a personal but formal process
19:40 mulander ie. the checklist I have for myself at work lists all the attachments I expect from the client.
19:40 kowey had always suspect that rigidity is a good thing for this sort of stuff
19:41 mulander kowey: I am workign on core banking software - some clients often report problems with one sentence of description and no data.
19:41 kowey eep!
19:41 mulander and that immediately counts toward our time to solve it with penalty fees if the issue is not solved in time.
19:42 mulander you either go through a checklist or go over the time and pay the penalty
19:42 kowey and so you have a reproducible process
19:42 mulander yeah
19:42 kowey that you've refined
19:42 kowey http://wiki.darcs.net/BugTracker gets it wrong in quite a few places, I sense
19:42 mulander it also helps with procrastination
19:43 kowey it's quite obsessed with the how-do-you-file-this-damned-thing
19:43 kowey without thinking hard enough about how-do-you-read-this-damned-thing
19:43 kowey well, not so much filing, but what actions do you need to take
19:44 kowey what roles do you need to play? (eg. customer service, triage, maintenance -- was basically written for somebody who should see themselves as issue manager)
19:44 mulander at work I don't see the big picture of the issue tracker (I wish I could)
19:44 mulander I just get issues assigned to me with a priority
19:45 iago "reproducible process" is they idea of any quality framework
19:45 mulander exactly
19:46 iago the key idea*
19:46 mulander and from experience - the more complex your documentation for the process
19:46 mulander the less people will adhere to it.
19:46 iago Heffalump, listen :P
19:46 iago well, read
19:46 kowey I've fallen on that one so many times
19:46 kowey and still I don't learn!
19:46 mulander it's good to have it thoroughly documented for people maintaining the process
19:46 mulander but you need something simple (like checklists, flow charts) for the end points
19:46 kowey but you need to get it tl;dr friendly
19:47 mulander as in the people that actually have to go through it
19:47 mulander I can scan through a checklist quickly
19:47 mulander I can't do the same with a 50 page document :)
19:47 iago mulander, there are processes that can be described as checklists
19:48 iago others need more text
19:48 mulander iago: sure
19:48 iago but yeah, checklists are preferable when possible
19:48 kowey (part of my reading problem is in the responses from developers to bugs, actually)
19:48 mulander iago: I am generalizing on issue tracking.
19:48 kowey but I suspect that there is a issue-manager checklist-isation which is also possible there
19:49 kowey http://bugs.darcs.net/msg12848 is an example
19:49 kowey I bet I could teach myself how to read this
19:49 mulander ok let's try
19:49 kowey as a GTD practioner, my thinking about it is "what's the next action?"
19:50 kowey but that's not helpful, so plan B
19:50 mulander oh we are in a different phase here
19:50 mulander as in from 'my book'
19:50 mulander look at that message.
19:50 kowey right, I think you're describing something that needs to come before
19:50 mulander it's clear that the problem at this point was at phase 3.
19:50 kowey which I'd never thought carefully enough at
19:50 kowey ah!
19:50 mulander hence: 1. Data gathering is done.
19:50 mulander 2. Diagnostics is generally done.
19:50 mulander 3. We are searching for a solution
19:51 mulander and that often may move us back to diagnostics for different problems emerging from this one
19:51 mulander here the developer tells us that:
19:51 mulander the code that takes action is shared among several features
19:51 mulander (record is mentioned)
19:51 mulander so it may lead to inconsistent behavior between record and whatsnew.
19:52 kowey “The -l (--look-for-adds) option exists for record as well, not just whatsnew”
19:52 mulander it's not easy and will require changes in the code including adding an extra diffing option
19:52 mulander that's no longer a bug fix
19:52 mulander that's a modification.
19:52 kowey so that's one thing I can identify -- where is the problem? doesn't just live in whatsnew-land
19:53 mulander it may affect performance greatly if done incorrectly
19:53 mulander that's what I see in this message
19:53 mulander kowey: oh yeah and I differentiate between what's required to make things work.
19:54 mulander fixing existing code is just a bug fix and might not require a lot of code being moved - less regression testing.
19:54 mulander modifying a lot of code to make something work is a full regression test ahead.
19:54 mulander implementing something new is a new feature and also requires a full regression.
19:55 mulander generally how intrusive the solution is.
19:55 kowey you measure pain by the amount of testing you'd need (or more to the point, intrusiveness)
19:55 mulander yes
19:56 mulander it has more implications at work but it's not related to our discussion here
19:57 kowey so in the above, I don't see the performance implications (although trying to retain perspective of the learning-to-read aspect of this discussion and not the actual message there)
19:57 kowey I can get a glimmer of the idea that mornfall's first couple of sentences up there brings me to step 3 of the checklist
19:57 mulander "Filtering the resulting diff is not going to work since
19:57 mulander it could easily take hours to compute it."
19:57 kowey OK - we've misidentified the where
19:58 kowey oh!
19:58 mulander if one approach might lead to performance problems - then other approaches might still hit it.
19:58 kowey I see, I'd focused on the one message
19:58 kowey and not the subsequent ones
19:58 kowey hah
19:58 mulander so it's still worth to keep in mind and do a performance test of the feature after the issue is fixed.
19:58 kowey so one sort of checklist-type question is what is the nature of the implications?
19:58 mulander *especially* if the performance problem happens in the 'naive' approach
19:58 kowey is it UI? performance?
19:59 kowey what are the user-facing consequences
19:59 mulander since most complex problems tend to deviate into the naive paths during development
19:59 mulander kowey: user facing consequences here is mostly performance - as that was first observed by the user
19:59 mulander and ui / logic of the software
19:59 mulander imagine the guy said he versiond his /home
19:59 mulander my home could be 20-30GB
20:00 mulander full of subdirs
20:00 mulander and I have a _darcs on toplevel
20:00 mulander and just did whatsnew
20:00 kowey ouch
20:00 kowey and that would be actually applying intelligence to your reading
20:00 kowey hmm :-)
20:00 mulander the reported told us that: "I consider this behaviour to be a bug. My main example is that I use darcs to
20:00 mulander manage the config files in my home directory"
20:01 mulander *reporter
20:01 kowey and I was focused on this as being a "ah, this doesn't do what I want"
20:01 mulander he just didn't expand on the example.
20:01 kowey as in "oh, I'm getting a bunch of irrelevant crap in my output"
20:01 kowey "that I'm not really interested in”
20:01 mulander well you don't know that
20:01 mulander you didn't ask him :)
20:01 mulander what he really wanted
20:02 kowey so your skillful data-gathering, do I really understand what the user wanted to achieve
20:02 mulander if you start a sentence with "So it looks like"
20:02 kowey (1)
20:02 mulander stop, rethink and change it into a question.
20:02 kowey instead of just assuming, ah, he's not happy with the output
20:02 mulander he also pointed us to http://bugs.darcs.net/issue1645
20:03 kowey which should have been a clue
20:04 mulander also changing this into a feature is bad PR
20:04 kowey (part of the GTD-disease is a mentality of "how can I get rid of this as quickly as I can [where get of means “put in the right mental bin/associate with an action"])
20:04 mulander especially since this can be clearly treated as a bug
20:04 kowey (and I suspect that's not necessarily a failing of GTD in and of itself, but poor application)
20:04 mulander I can easily imagine a situation when a whatsnew here would take ages.
20:05 mulander and the reporter wans't happy about this: "Oh. It's funny how the way a program is implemented can change bugs into
20:05 mulander features."
20:05 kowey Yep
20:05 mulander " I would like to bump this bug. Also, I strongly object to this
20:05 mulander being listed as a 'feature'. This is a *bug*. The behaviour is clearly
20:05 mulander and simply incorrect."
20:06 kowey eep
20:06 kowey I must have stopped reading by then :-(
20:06 kowey (as in given up)
20:06 mulander that was his 3rd response
20:06 mulander it's ok to change bugs into features but only after:
20:06 kowey ah he did bump (which we try to encourage)
20:07 mulander 1. explaining to the user why you consider this a feature
20:07 mulander 2. getting an 'ok' for the change from him.
20:07 mulander that's my general rule for it of course
20:07 mulander it's your bugtracker and you can do whatever you want with it :)
20:07 kowey do not make light of a user's issue
20:07 kowey treat it with the same seriousness (if not more) as they do
20:07 mulander well we have a bit of a different model here which I understand
20:07 mulander as this is an open source project
20:08 mulander but generally I don't consider bugs in my projects as something offending me personally
20:08 mulander and yes, sometimes a bug requires implementing a feature to get rid of :)
20:08 kowey I *thkn* darcs is lucky there
20:09 kowey in that bugs (or claims of bugs) do not upset people's egos
20:09 mulander still if the guy was so persistent on treating it as a bug
20:09 mulander I wouldn't change it without really good reasoning behind it
20:09 mulander and an ok from him.
20:09 kowey we should have noticed
20:09 kowey right
20:09 kowey that's a good principle
20:10 kowey the first bug -> feature was me, and that came from not actually understanding what the user was trying to achieve
20:10 mulander and he changed it back
20:10 kowey (ie. not wait 1 year for darcs whatsnew -l to complete in /home)
20:10 kowey right
20:10 kowey then we discussed
20:10 mulander and then he changed it back to bug -> feature
20:10 kowey and he changed it back to the feature
20:10 kowey but with a hint of frustration/annoyance in this text
20:11 kowey funny how...
20:11 mulander yeah
20:11 mulander he is generally now asking for a workaround in that place
20:11 kowey and an alert issue manager would have responded appropriately
20:11 mulander brb 5m
20:11 kowey or *better yet* pre-empted it in the first place
20:18 mulander back, sorry had to grab some toasts :)
20:18 kowey thanks for all the advice by the way
20:19 kowey I hope you don't have the frustrating experience of having it seem to fall on deaf ears
20:19 mulander don't mention it - hope it comes out the way it should - english is not my native language.
20:19 kowey a lot of self-retraining here
20:19 kowey (and part of it is learning how to be effective in a hobby capacity)
20:19 mulander oh so far I consider your way of handling the bugs as quite sound.
20:20 mulander I report a lot of bugs to open source projects
20:20 mulander seen much worse workflows.
20:20 kowey there's worse than "argh, can't read, archive it!"?
20:20 kowey :-)
20:20 mulander yes
20:20 mulander 'no action'
20:21 mulander 'don't like it - fix it'
20:23 mulander still I see a big problem with your issue tracker in one place.
20:23 mulander it works well when the issue is hot
20:23 mulander but after it dies off a bit - it drops off the radar
20:24 mulander and since it's mailing list based - it's rarely looked at later on in the future
20:24 kowey we don't follow up
20:24 kowey hmm, I use it as a web based tool actually, dunno if it's the same for others
20:24 kowey but oh!
20:25 kowey hmm, no was starting to think this may be a feature (email cc'ing) can have harmful side effects thing
20:25 kowey the underlying theory at the time
20:25 mulander btw are you using jenkins/sonar or just the build bot?
20:26 kowey just the build bot
20:26 kowey what does something like jenkins/sonar add?
20:26 mulander jenkins is generally a nice central build bot with a lot of plugins
20:26 mulander and can provide binaries, run tests
20:26 mulander with plugins it can do much more (integrate with issue tracking etc)
20:27 kowey (it's also an example of us being pulled in several different directions, the problem of which gap to tackle first)
20:27 mulander sonar is a nice dashboard - can pull in info from a lot of places (ie. old issues from the tracker)
20:27 kowey (customer service, dev documentation, user documentation, infrastructure, theory, the list goes on)
20:27 mulander well I am currently interested in darcs from 3 perspectives
20:27 kowey I've wished for a dashboard since GSoC mentor summit last year
20:27 mulander 1. as a user
20:27 mulander 2. as a potential contributor (if I am able to help)
20:28 mulander 3. as someone implementing the same thing in a different language
20:28 mulander a lot of the stuff overlaps
20:28 mulander ie. in order to implement darcs I need to know 1 and 2 well.
20:29 mulander especially since the documentation itself is a bit sparse
20:29 mulander and the whole thing is a moving target
20:29 kowey darcs developers already talking about darcs 3
20:30 kowey btw, implementation in another language would be an interesting thing to watch (did you know David Roundy was working on a Go implementation of a patch-theory based version control system built on Git?)
20:30 mulander no I didn't know about it
20:31 kowey https://github.com/droundy/iolaus
20:31 mulander I knew that the first prototype for darcs was C++ based.
20:31 kowey after leaving Darcs
20:31 kowey David realised that actually maybe Darcs and Git aren't so far apart as you might think
20:31 kowey and that if you have a multi-headed Git repo
20:32 kowey you can do some Darcs-like things quite nicely
20:32 kowey I think it's a horizon type approach... before the horizon, you get to do all your cool darcs cherry picking
20:32 kowey after the horizon, it's a plain old Git repo
20:32 kowey (this is me potentially misunderstanding what he was trying to do)
20:33 mulander kowey: well my approach is much simpler - I want to be binary compatible (repo format, features) with regular darcs
20:33 kowey hence, specification
20:33 mulander mostly to learn ada/haskell better.
20:33 kowey mulander: I may use you as a focusing tool
20:33 mulander and as a fun side project
20:33 mulander kowey: sure - feel free to do it :0
20:33 kowey to cut through my mental noise of oh-I-should-do-X-Y-Z
20:34 kowey and so right now the two thoughts running through my head are
20:34 kowey (A) we need to focus on specification - even if we're doing New New Darcs
20:34 kowey and we don't actually know what we want
20:34 kowey maybe we need to start encouraging devs to write down ideas in more spec-like way (even if a bit informally)
20:35 kowey and start trying to spec out some of the stable-looking stuff like on disk representation for patches
20:35 kowey I think people like lispy have been asking for this forever
20:35 kowey and (B) there is hope for bug tracking, you *can* learn how to read
20:36 kowey a good checklist can help you look at an incoming report in a much more structured (and thereby less wtf-inducing) way
20:36 mulander generally I would go the route:
20:36 mulander 1. new things-  go through a spec phase (informal - mostly to have a nice discussion and track record for implementation)
20:37 mulander 2. go through the issue tracker - produce documentation during phase 2 (diagnostics) 3 (finding solution) 4 (implementation) and 5 (testing)
20:37 mulander point 2 covers documenting the existing code base.
20:37 kowey use the issue tracker to figure out what needs documenting the most
20:37 kowey if subsystem X keeps popping up
20:37 mulander it's mostly - 'hey you just spent 2 hours to understand how it works - can you please write it down?'
20:38 mulander even something like a 'dirty notes'
20:39 mulander you tell someone he can note somewhere (pastebin, google docs, a wiki page - whatever) and you will take care of the formatting and putting it into the structure of the wiki
20:39 kowey we try to practice Conspicuous Use of Archives
20:39 kowey well, I try to bang on that drum
20:39 mulander that way someone can make notes while he diagnoses / works on an issue
20:39 kowey the theory being that if you're consistent enough about it
20:39 mulander and we can move it into place after he is done
20:39 kowey people absorb the culture that putting things into linkable resources is a Good Thing
20:40 kowey (Conspicuous Use of Archives being deliberately making the effort to find prior documentation, even if just giving an answer would be easier)
20:40 kowey because that very action makes the existence of these archives more salient (sorry, one of my failings is a tendency to want to have a theory behind everything)
20:41 kowey right, so as you're saying, don't let the documentation effort be about the bureaucracy
20:41 kowey I'll take care of the formatting etc for you
20:41 kowey could you just throw something up?
20:41 mulander yep
20:42 kowey you're making the effort, I want it to be as easy as possible for you
20:42 mulander and every bit matters
20:42 mulander we don't care if it's just one sentance
20:42 mulander it can be something like 'File.hs' is never used during '--file' command so don't be mislead by the name.
20:43 mulander btw kowey what's the status on 2139?
20:43 mulander http://bugs.darcs.net/issue2139
20:43 kowey ah!
20:43 kowey see, more text :-)
20:44 mulander I tried to reproduce it
20:44 mulander and I have a hunch that it only happens on mac os x?
20:44 kowey hmm, let me try to step through this formally
20:44 mulander ok
20:45 kowey 1. what was I trying to accomplish? move some directories around, usually into cwd
20:46 kowey (sorry in the middle of pulling up the checklist actually)
20:46 mulander kowey: https://gist.github.com/1860976
20:47 kowey thanks
20:47 kowey ok, so here 1. kowey said that he did some moves and they crashed with seemingly unrelated messages
20:48 kowey 2. here he expected his moves to succeed, mv foo/bar . should give ./bar, not error
20:48 kowey 3. (ah-hah)
20:48 kowey he appears to have done some moves involving the pattern dir/file . (or some directory)
20:49 mulander 3 should also have ' what was the actual result'
20:49 kowey and this lead to two messages, either (The filename  is not valid under Windows.), although from familiarity with kowey he's on a Mac
20:49 kowey or (Use --case-ok to allow files differing only in case.)
20:50 kowey 4. no, not much in the way of details, except that he did specify it was post 2.5
20:50 mulander so what are we missing?
20:50 mulander 4.1 exact darcs version
20:50 mulander 4.2 operating system (not all devs know you)
20:51 kowey gah, this page needs to be simpler :-( http://wiki.darcs.net/BugTracker/Reporting
20:51 mulander 4.3 ls -al of the directory
20:51 kowey more tl;dr friendly
20:51 kowey yes, 4.3 would have helped
20:51 mulander kowey: as for the reporting page
20:52 mulander it should list:
20:52 mulander Hey. What did you do?
20:52 mulander What was the expected result of this action?
20:52 mulander What happened instead?
20:52 kowey the google code bug tracker seems to get this right with a template
20:52 mulander Can you tell us more? (Operating system, darcs --exact-version, bla bla...)
20:52 mulander thanks.
20:52 mulander kowey: exactly.
20:53 kowey huh, I bet I could do that
20:53 kowey pre-fill the textarea
20:53 mulander ok, let's get back to the checklist
20:53 kowey yup
20:53 mulander ok, so let's imagine you gave answers to 4.1 - 4.3
20:53 kowey ok, so a lot of details were missing (I would not have thought of 4.3 btw)
20:53 kowey if I had given 4.3
20:53 kowey we would have discovered the crucial missing detail
20:54 kowey which is that the file being moved (dir/f) is itself a directory
20:54 kowey (dir/subdir)
20:54 mulander ok, that is new info
20:54 kowey (which I sort of re-discovered trying to write the test case that I ended up throwing on the tracker)
20:54 mulander you are typing this out into the issue as we talk right?
20:55 kowey uh, yes!
20:55 kowey :-)
20:55 * kowey looks guilty
20:55 Heffalump oh, sorry
20:55 Heffalump I understand what's wrong in that bug but haven't written a fix yet
20:55 Heffalump I should have mentioned that :-)
20:55 mulander yeah :)
20:55 mulander that would really help :D
20:56 mulander and I am really curious what's wrong with it.
20:56 kowey I have a vague idea from looking at the code
20:56 kowey which is that we recently introduced a check to see whether or not the user had done 'mv foo bar' and then 'oops! darcs mv foo bar'
20:57 kowey new old feature
20:57 mulander kowey: you might want to extend your checklist with ls -al or ls -al -R as the default diagnostic data you ask for.
20:57 mulander and possibly tree if installed.
20:57 kowey that darcs should allow this and do the right thing, even if foo does not exist
20:57 kowey and that this work introduced some general refactoring (I may recall wrong)
20:58 kowey the upshot of which the code that checks on pre-requisites between the name of the src vs the name of the target
20:58 kowey only fires when the src is a file
20:58 kowey and not a dir
20:58 kowey sorry, I'm making a big hash of this; understood it better when I was on the train but it was late
20:59 mulander cool - but in general it seems that the issue is sorted
20:59 mulander but some poor soul might be wrestling against it because he has no way to tell that we just discussed it here.
20:59 kowey so what happened here is that you have one darcs developer (me) who did a semi-diagnosis and didn't tell anybody
20:59 kowey but did try to upload his test case before he forgot!
20:59 Heffalump kowey: I don't think it's that; I think it's an over-simplification that Alexey made when doign some code cleanups
21:00 kowey and another darcs developer who did do a diagnosis
21:00 kowey ah! ok
21:00 mulander yay for duplicated work :D
21:00 kowey I just jumped to conclusions from the darcs annotate
21:00 Heffalump well, I only figured it out a couple of days ago, I just didn't get round to writing it out
21:00 kowey I just saw Florent's name on some new code that seemed related, and was related to new fetaure
21:00 kowey and explicitly mentioned files in src but not dirs
21:01 mulander so this issue should be assigned to Heffalump or annotated with how to fix it and left for someone to pick up?
21:01 kowey so I said, "ah!" and thought no further of it
21:01 mulander kowey: well this still might be worth taking a look into.
21:01 Heffalump assign to me is fine, I did plan to comment on the issue to make that note. Just hadn't got round to it.
21:01 mulander kowey: maybe you found a possible bug?
21:01 Heffalump It'd be really good to have a decent set of test cases though.
21:02 kowey mulander: one of things I don't really know how to deal with yet
21:02 Heffalump mv commands are actually really hard to specify properly because it's quite subtle (e.g. does a trailing / make a difference to behaviour - probably not with a simple mv but it can do for a recursive operation)
21:02 kowey is that when things go wrong, users tend not to freeze the repo in the gone-wrong state
21:02 kowey (and I'm really bad about this myself, despite knowing I ought to!)
21:03 kowey hmm... which is one reason you say to try to ask for these things up front
21:03 mulander yup
21:03 mulander the software could also prompt.
21:03 kowey maybe we need a darcs bug command
21:03 mulander 'hey the repo has gone amok - can I freeze the bastard down so you can upload it too us or is it too private?'
21:04 mulander 'ok, can I have just some metadata please?'
21:04 mulander etc
21:04 mulander 'hey - go to this url and paste this there: some diagnostics'
21:04 kowey although here darcs has no idea something went wrong, just the user saw something unexpected
21:04 kowey but in the case where darcs *does* know something went wrong
21:04 mulander yes, exactly
21:05 mulander you will not detect all bugs
21:05 mulander but grabbing as much info as you can programatically will help.
21:05 mulander just be sure not to grab private data.
21:05 kowey heh, we just say See http://wiki.darcs.net/BugTracker/Reporting
21:05 kowey the more doing the Right thing happens automatically, the better
21:05 mulander "If you have never reported a Darcs bug before, your message will be held in a quarantine area until a human confirms that it's not spam."
21:05 kowey unfortunately
21:06 kowey (that human is me)
21:06 kowey and I don't know what to do about that
21:06 kowey (aside from releasing the brakes and hoping for the best)
21:06 kowey we get quite a bit of spam from bugs@
21:06 mulander set an auto-reply - thanks for submitting this issue we will take a look on it shortly it will appear on the issue tracker during 1-2 days bla bla'
21:07 kowey I've been superstitious about not doing that for fear of just getting more spam
21:07 kowey but maybe that's silly
21:07 mulander you will still get spam to it.
21:07 kowey and responding appropriately to users is way more important
21:07 mulander loads of it.
21:07 mulander and even more in time
21:07 kowey than trying to reduce your spam percentage by N%
21:07 mulander as people paste darcs output somewhere or something scrapes the url.
21:07 mulander generally that message would deter me from reporting.
21:08 mulander fortunately I didn't see that page before reporting an issue.
21:08 kowey make the right behaviours as easy as humanly possible
21:08 mulander "That means we need a set of instructions that will, at least for you, ALWAYS cause the bug to appear."
21:08 mulander this is also bad
21:08 ByronJoh1son joined #darcs
21:08 mulander as in the 'ALWAYS' part.
21:08 mulander it's not always possible
21:09 mulander and it's better to have a partial bug report than none.
21:09 ocharles joined #darcs
21:09 mulander that messages tells me - 'either you provide full details or scram with your problem away from us'
21:09 kowey don't all-or-nothing people
21:09 kowey meet them where they are
21:10 kowey (habit of trying to translate things into a big pile of principles, which I will inevitably forget)
21:10 mulander think of it the same as survey companies.
21:10 mulander you did online surveys right?
21:11 mulander do you know why they are split into parts
21:11 mulander and that even partially filled forms are stored?
21:11 mulander it helps them optimize.
21:11 mulander where do people stop? how can we get them to fill all of it?
21:11 mulander can we get a different pov from the partial results compared to the full results?
21:11 mulander did someone stop after hitting a question that was to personal?
21:12 mulander (Ever wondered why age/sex/email nowadays is mostly located at the last page of the survey)
21:14 mulander kowey: so are we updating issue2139 with any info?
21:16 kowey mulander, ah I went away for a while to write up http://bugs.darcs.net/msg15116
21:17 kowey mulander: oh wow, never thought of that
21:17 kowey I just thought it was splitting it into manageable chunks
21:17 kowey but never occurred that it would be a way of increasing the chances that partial results are useful
21:18 kowey taking advantage of people stopping at "natural" give up points
21:18 mulander yep
21:18 kowey incremental is the name of the game
21:18 kowey OK, so the reporting page needs to be simpler and less discouraging
21:19 kowey bugs@ needs to be configured to auto-respond to moderated messages
21:19 * kowey needs to configure the BTS to give a template (if possible)
21:20 kowey filed in Omnifocus under "Hobbies : darcs grease"
21:21 mulander :)
21:23 mulander kowey: http://bugs.darcs.net/issue2093
21:23 mulander brainstorm?
21:23 mulander could he have --lazy get the repo from that share?
21:24 mulander also why would rec search for files that are no longer in the repo?
21:25 mulander and if it's related to --lazy then this issue could be merged with the other --lazy not informing about it problem.
21:25 kowey hmm, being slow to read-and-understand
21:25 mulander also if --lazy requires pulling patches for regular forward workflow (record) then I see that as another problem.
21:26 mulander kowey: ok, what happened.
21:26 mulander the guy had a repo in his virtual machine
21:26 mulander small repo 1 mb
21:26 mulander he did darcs rec
21:26 mulander and it was hanging
21:26 mulander doing nothing
21:26 mulander he couldn't kill it.
21:26 mulander he mentioned that he pulled the repo from a shared mount on sshfs
21:27 kowey that's a Clue
21:27 mulander later on, running it with strace
21:27 kowey (in the sense of ooh-that's-unusual)
21:27 mulander revealed "Aha, strace showed me the problem.  Darcs was trying to access a patch
21:27 mulander from my local copy of the repository during the record, and found it
21:27 mulander missing.  Then it went to get the patch from the sshfs-mounted
21:27 mulander repository that I had pulled it from.  It turns out that the reason for
21:27 mulander hanging then had nothing to do with darcs.  All processes would hang
21:27 mulander (once the sshfs connection had been on for a few hours) when they tried
21:27 mulander to go near that directory.  A bit of googling suggests I am falling foul
21:27 mulander of this Debian/Ubuntu sshfs bug:"
21:28 mulander cool so he hit an sshfs bug.
21:28 mulander but the problem is that darcs didn't tell him what it tried to do
21:28 kowey lack of feedback
21:28 mulander and that record required a missing file?
21:28 mulander why would record require that file/patch?
21:29 kowey so mulander this is one of those situations where without sitting down and thinking about it
21:29 kowey I'm not sure if the second behaviour should be considered buggy
21:29 kowey from the surface it seems obviously wrong
21:29 kowey that record should not need to grab anything ever, but it could be a consequence of...
21:29 mulander well we don't know if he did a --lazy get.
21:29 kowey (and then my thinking gets fuzzy from here on)
21:30 kowey so that's a step 4 question
21:30 kowey by the way, could I just double-check something; did you fetch that repo using darcs get --lazy?
21:30 mulander yep
21:30 mulander and more
21:31 mulander 1. did you remove that file from the path where darcs was searching?
21:31 mulander because we don't know why it's missing
21:31 mulander he could have removed it.
21:31 mulander or he could have a lazy repo.
21:31 mulander also we don't know if the change he made was related to that path in some way
21:31 mulander and this issue lacks a resolution
21:31 mulander look
21:31 kowey two things that darcs could be looking for
21:32 kowey patches
21:32 kowey pristine files
21:32 mulander it looks like resolved on 2011-08-01.
21:32 mulander but at this point
21:32 mulander it should be either merged to lack of info from darcs on such operations
21:32 mulander or some other 'conclusion'
21:32 kowey made into a UI bug
21:32 kowey don't accept the easy "done now" answer
21:32 mulander also we have some info here
21:32 mulander that could help with our --lazy problems
21:33 mulander and pulling patches
21:33 mulander look:
21:33 mulander However, --debug-verbose does say that it is copying
21:33 mulander a patch (from my sshfs mountpoint), so that would have been almost as
21:33 mulander useful as the strace in pointing to the parent repository as a problem.
21:33 mulander so --debug-verbose seems to show patches being copied.
21:33 mulander I wonder if it's done on lazy repos during the actions I 'hit' by accident.
21:34 mulander if so, then maybe changing the level of that debug line to default would fix the whole issue.
21:34 mulander (making assumptions without knowing how the debug output works)
21:35 kowey promoting it report that information on normal verbosity?
21:35 mulander yes
21:35 kowey if so it sounds like a too-much-too-little fine line issue (I need a name for this class of problems by the way)
21:36 kowey by too-much-too-little I mean that in some contexts X level of whatever is insufficient
21:36 mulander I would call it 'software-feedback'
21:36 kowey but that in other contexts, it is too much
21:36 kowey and that it's terrifically easy to provoke a too-much by trying to fix a too-little
21:36 mulander well it's still UI.
21:36 kowey and vice-versa
21:36 mulander think about it this way.
21:36 mulander maybe it should be promoted only for --lazy repos?
21:37 mulander or when that code is hit from something different than darcs get/pull
21:37 kowey I guess what we're looking for
21:37 mulander but it's a pointer for somoene
21:37 kowey is to tell users that we're fetching things they wouldn't otherwise expect to be fetching
21:37 mulander on where to look for the action that does the actual pulling
21:37 mulander yes.
21:37 kowey so if you're doing a darcs get or darcs pull
21:37 mulander and that little bit of info we got in that issue.
21:37 kowey you know you need to fetch some patches
21:37 kowey no output
21:37 mulander tells us that --debug-verbose has the place pinpointed
21:38 mulander so we have the problem 'diagnosed'
21:38 mulander and need to find a solution for it.
21:38 mulander also we should if if this issue is a merger
21:38 mulander or a bug itself.
21:38 mulander it would be a bug if darcs record needs to lazy fetch patches without a good reason.
21:38 kowey hmm, so now I'm off thinking about the UI
21:38 mulander but that might be worth noting as a new issue.
21:39 kowey I tend to create new tickets for new issues that arise from an issue
21:39 mulander and a merger with http://bugs.darcs.net/issue2123 if it only needs to tell that it pulled from lazy
21:39 kowey rather than renaming because renaming seems to create more confusion in the long run
21:40 mulander I noted a possible relation to issue2093 in issue2123
21:40 * kowey starts responding to things in the old-impulsive way
21:41 mulander brb 5m
21:46 raichoo joined #darcs
21:49 mulander back
21:51 mulander kowey: so now we should really decide what to do with issue2093
21:55 mulander kowey: also I saw you changed issue2127 to wont-fix and also suggested a solution. This seems to contradict each other.
21:55 * kowey creates a new issue out of issue2093
21:57 mulander also does roundup provide a way to break down the statistics of your bug tracker?
21:57 mulander closed/open/by status etc?
21:59 mulander issue1835 <- this can be closed.
22:00 kowey mulander: I don't remember
22:00 kowey mulander: I'd bet there are
22:00 kowey in the worst case, it's scriptable with python
22:00 kowey http://bugs.darcs.net/issue2143 from issue2039
22:00 mulander found the grouping
22:01 mulander so it's ok to review by status
22:01 kowey err 2093... note that I couldn't figure out how to mark its priority/status
22:01 kowey so I just left it in the unknown state
22:01 kowey bad sign :-)
22:02 mulander I think it neds reproduction
22:02 mulander as we don't know if that really happens.
22:03 mulander because we don't know if the repo was lazy
22:03 mulander and if not we don't know why it wanted to check the patch from the source repo
22:03 kowey and now my attempt at responding to NEil
22:03 mulander maybe it was something related to 'local' caches?
22:03 kowey again using my old instincts and not trying t obe systematic about it
22:06 mulander why did you change issue2070 from resolved to unknown?
22:07 mulander btw are you the only one allowed to change statuses?
22:07 kowey everybody is
22:07 mulander what are the rules on this :)
22:07 mulander I wouldn't want to interfere to much with your workflow
22:08 kowey it is documented (I hope), but I will have to find it
22:08 kowey basically my attitude to it is this
22:08 kowey the workflow should be baked into the system in such a way
22:08 kowey that users don't have to worry about doing it wrong
22:08 kowey (very far from this!)
22:09 kowey the *unknown thing is actually just being hyper-conservative
22:09 kowey oh no! they said something!
22:09 kowey is it still resolved?
22:09 kowey this is an example of too-much-too-little for me
22:09 kowey actually most of the time this is way too conservative and you get these surprises
22:10 kowey but the times where it's the right thing to do are more severe
22:10 kowey oh, not sure if I remember my stats correctly, but analogy being class I vs class II errors
22:11 mulander generally you want to avoid shooting up an email to the reporter if the issue was not fixed yet?
22:12 kowey sorry?
22:12 kowey the default of popping states back to unknown
22:12 kowey which seems silly now
22:12 mulander ah
22:12 kowey is to avoid bugs being marked resolved when they weren't
22:13 mulander I misunderstood you
22:13 kowey ie. auto-reopen issues if the user says a new thing
22:13 kowey but often the new thing is just "thanks!"
22:13 kowey so most of the time it's the wrong thing to do
22:13 * kowey is off to bed now
22:14 kowey I might not work on darcs again for a while (maybe next week?)
22:14 kowey mulander: thanks again! I learned a lot from this
22:14 mulander kowey: thanks for a nice chat :)
22:14 kowey I hope I will act on some of the good intentions, but I can't promise!
22:14 kowey good night
22:14 mulander it's a hobby project :)
22:14 mulander no pressure on anything
22:14 mulander good night
22:26 raichoo joined #darcs
22:31 shachaf joined #darcs

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