Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2014-01-09

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

All times shown according to UTC.

Time Nick Message
00:10 sm hey, movement on ohloh darcs support: https://github.com/simonmichael/ohloh_scm/pull/1
00:11 sm just need to figure out all this crazy git talk
00:12 mornfall joined #darcs
01:04 mornfall joined #darcs
01:31 mornfall joined #darcs
02:13 MasseR_ joined #darcs
02:13 dleverton_ joined #darcs
02:35 stepkut joined #darcs
04:34 preflex_ joined #darcs
04:49 gh__ joined #darcs
04:52 rdesfo joined #darcs
04:56 rdesfo left #darcs
07:04 lelit joined #darcs
08:42 raichoo joined #darcs
08:42 mornfall joined #darcs
10:43 f-a joined #darcs
10:54 amgarching joined #darcs
11:11 nomeata joined #darcs
12:45 f-a joined #darcs
12:59 amgarching joined #darcs
13:07 galbolle joined #darcs
13:52 notdan sm: ha, that's a lengthy pull req
14:08 _ilbot joined #darcs
14:08 Topic for #darcs is now http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.8.4 is out http://darcs.net/Releases/2.8
15:06 gh_ joined #darcs
15:28 mizu_no_oto joined #darcs
16:10 vikraman joined #darcs
16:15 galbolle left #darcs
16:48 favonia joined #darcs
18:02 f-a joined #darcs
18:13 amgarching joined #darcs
18:56 lelit joined #darcs
19:19 pointfree-w530 joined #darcs
19:31 stepkut joined #darcs
19:36 dcoutts joined #darcs
19:36 dcoutts joined #darcs
20:33 dcoutts joined #darcs
20:33 dcoutts joined #darcs
21:08 gh_ I wonder if it would be better to come back to sending patch tracker notifications to darcs-users
21:13 gh_ it's more convenient in the cases when user-visible changes are debated around some code submission
21:13 gh_ eg export/import
21:24 sm I wish there was a way to leverage darcs hub to make patch submission/review/acceptance easier
21:25 mornfall there probably is
21:25 mornfall but it involves work
21:25 mornfall (yuck)
21:26 sm I feel there probably is, but not being an active darcs developer I can't say for sure what it is
21:29 mornfall sm: at the rate of ~10 patches a month you could wield considerable voting power by submitting 2 or 3 patches :-)
21:31 sm true! :) unless they were all rejected
21:31 sm as usual and perhaps like a few of us I'm on the fence about spending (more) time on darcs
21:32 sm what would be pros and cons of this baby step: instead of posting attachments on roundup issues, post links to patches in forks of darcs-screened on hub
21:33 mornfall sm: conversely, at the rate of 10 patches a month, it doesn't make sense to have any process at all
21:33 mornfall sm: just point at one guy and tell him to deal with them
21:33 sm there's always the chance with a different process, we'd get more patches
21:34 mornfall sm: the bug is that any *interesting* patches get bogged down; all you are permitted to work on is boring stuff
21:34 mornfall nobody likes that, you don't get patches... big surprise
21:35 mornfall see, my fingers already write "git pull" when trying to update the darcs repo :-(
21:36 sm AAAAAAH
21:36 sm IT IS THE END TIMES
21:36 mornfall there's no choice -- you can't seriously tell people to use darcs for new projects
21:37 mornfall it has critical bugs and noone to fix them
21:37 gh_ hmm there's so much to say…
21:37 sm you have the floor gh_
21:37 mornfall I wish I could, but my new project lives on github and there's basically nothing I can do about it.
21:38 * sm tells certain people to use darcs for new projects, and doesn't know of critical bugs they'll encounter
21:39 mornfall sm: merge is broken... they may be unlucky and their projects will never become big enough to notice
21:40 sm what do you mean by that ?
21:40 gh_ sm, yes we could accept more forms of code submission, why not. But I don't believe this will make contributors appear. I believe a 2.10 release would create awareness of the project, OTOH.
21:40 sm ("merge is broken" ?)
21:41 mornfall sm: that non-trivial merges crash darcs
21:41 gh_ sm, cons of the fork approach: 1) not possible to create a fork from commandline darcs (so you should probably think about writing code to implement that if you believe in it :-) ). 2) not sure how we are supposed to discuss code submissions. maybe just take it on darcs-devel?
21:41 mornfall sm: and once your branch has enough non-trivial history, you can no longer merge without rewriting history ... extremely carefully
21:42 gh_ sm, you could write an RFC in darcs.net/Ideas
21:42 sm mornfall: you mean the old exponential run time with highly conflicting patches issue ? If so, I guess I have to believe you.. I never see it
21:43 mornfall sm: basically all the repositories I "own" and that matter for anything have collected some amount of brokenness already...
21:43 sm I avoid conflicts in my repos
21:43 mornfall sm: not really, this is a correctness problem; duplicate hunks do not merge correctly
21:43 mornfall sm: yes, avoiding conflicts is fine when there are 2 people on a project, maybe 3
21:44 sm mornfall: don't some git projects follow a similar policy ? ie always rebase your changes before pushing to trunk ?
21:44 mornfall I mean, you have to tiptoe around darcs the same way you have to tiptoe around git to avoid breaking its internal state.
21:44 gh_ mornfall, from time immemorial, humans have managed to work their way around broken software :-)
21:45 sm gh_: re 2.10 release.. I'
21:45 gh_ not that I'm happy that anything be broken in darcs, but I wonder if we can limit damage at the UI level
21:45 mornfall gh_: sure, but here's a piece of broken software nobody uses and is written in a language that very few people understand is hardly an argument to replace a different broken system that has thousands upon thousands of users
21:46 sm oops. I feel growing the darcs project is quite a long shot/long-term project by now. I feel we could at least make it easier for the existing community to scratch their own itches, first
21:46 gh_ mornfall, agreed, that's why I don't see myself telling anyone to switch from X to darcs, but as a "starter" VCS I still think it's worth using
21:46 mornfall gh_: so when I come to a bunch of people that I am supposed to work with, all of them have github accounts, who am I to tell them to go try darcs which I know is badly broken?
21:47 sm I don't see how darcs is more broken than git. They each have a sweet spot and practices that are good to follow
21:48 sm I think you are overstating the brokenness
21:48 mornfall sm: well, git will screw you over if you make a mistake; darcs will screw you over even if you make none (within the stated interfaces of the respective system)
21:49 sm that's not my experience
21:49 mornfall sm: the user manual doesn't say, never create conflicts and never let darcs resolve a conflict by creating duplicate-type patch or it'll explode in your face a year later
21:50 sm fine, a documentation problem - no need to throw out darcs for that
21:50 sm git has had a few of those
21:50 gh_ sm, +1 for scratching our itches, I'd suggest you describe the process on the wiki so that we can really discuss it
21:50 sm gh_: re that, raising RFC right here and just possible on the list if necessary is plenty of process IMHO
21:50 mornfall well, I guess nothing has really changed, we were discussing the process in 2010 too
21:51 sm the wiki is way out of scope for me at this point
21:51 amgarching joined #darcs
21:52 gh_ sm, fair enough
21:52 mornfall I'd summarise the process as: find someone who's willing and able to code and then let them do it
21:54 gh_ yeah maybe we could try stop using the bug/patch tracker altogether (keep it only for arceology purposes).. actually you proposed that some time ago sm , didn't you?
21:54 sm probably, yes
21:55 sm any other points of view on darcs future in the channel ? I feel there's some people likely to keep or start working on darcs if they feel others doing so. If there isn't the interest, then...
21:56 mornfall I mean, check out http://bugs.darcs.net/patch1100 for all the process is worth.
21:57 gh_ the whole idea of eliminating overhead is growing on me, especially since the screened / reviewed branches enable differed code review
21:57 mornfall That's a personal itch, trivial bug, comes with a fix and it'll soon have its first anniversary.
21:57 sm mornfall: that's why I bring this up every so often. Our process sucks
21:57 sm it has sucked for a LONG TIME
21:57 mornfall and nothing ever helped
21:58 sm still, today's a new day! never too late!
21:58 gh_ mornfall, yes but only saying it on IRC was not exactly part of the "process"
21:59 gh_ then you are one of the committers, and there is the "hello, anyone?" rule that says you can auto-commit after nobody reacts for one week
21:59 gh_ you never used that
22:00 mornfall of course, because I am no longer really a commiter
22:00 mornfall pretend I'm a stranger for a moment and contemplate how anxious I would be to submit more fixes
22:00 gh_ mornfall, you no longer have commit access? how strange
22:01 mornfall I probably do, I wouldn't know.
22:01 mornfall I didn't try for years.
22:01 gh_ you could at least try…
22:01 gh_ yeah, I remember cases of stranger submitting patches and no aswer for months…
22:04 mornfall what's up with patch1093?
22:05 sm contributing patches needs to be fun
22:05 sm any ideas for alternate process gh_ ?
22:06 sm I didn't get what you meant by "cons of the fork approach: 1) not possible to create a fork from commandline darcs" yet
22:06 mornfall sm: he means you can't submit a patch without clicking anymore
22:06 mornfall sm: if you need to create hub forks for patches
22:06 sm oh I see.. you can create a hub repo from cli but it's not a proper fork of darcs-screened
22:07 mornfall also, that's extremely slow because you need to push thousands of patches
22:07 gh_ mornfall, I'd like to accept it. I recall the "blocker" was that we wanted to see whether darcs should switch to fslib before. but that's probably to be a long discussion so I'm going to have a look at the bundle (thanks for the reminder!)
22:08 gh_ ah, ok, I remember it's not only a darcs change but a hashed-storage change, so I need to upload a new version of h-s on hackage first.
22:08 gh_ sm, yes that's what I meant
22:09 sm good to know, and an easy enhancement
22:09 mornfall Well, if you happen to grow some culture, I may be in possession of some free coding time over the next 6 months. I can be convinced to brush off fslib I think.
22:10 sm handy if you fork often (I was imagining one long-lived fork per user mostly)
22:11 mornfall Anyway, I said before that if you want to stick with 0.5.x hashed-storage, you should just copy the source back into darcs proper. That library is dead as a doornail.
22:12 gh_ sm, ideas for alternate process? maybe "discuss code sent to darcs-devel under any form". and disable bug and patch tracker.
22:13 gh_ mornfall, I was more thinking about having an official h-s repository on hub.darcs.net
22:15 gh_ "offical" as , it's just one repo like another, but somewhere on the wiki we say the latest version is at *that* repo
22:16 mornfall Well, if you insist on keeping it as a separate library, I can't stop you, but do remove me from the cabal file...
22:17 gh_ mornfall, as you want. the point is not to pretend that there is real active development on that library, but just to have it somewhere with the changes needed for look-for-moves.
22:21 gh_ mornfall, merging it into darcs is more work that keeping it separate, moreso if you plan to work on fslib and we want to use it.
22:21 mornfall gh_: the hashed-storage patch is pretty broken, API-wise
22:25 mornfall gh_: I don't see how it is more work to copy the code into darcs
22:28 gh_ mornfall, oh, you mean just copy the code in one separate folder in the darcs darcs repository?
22:29 mornfall sure
22:31 gh_ we'll lose the history, but I guess we can still keep a h-s repo somewhere to look into it if we need.
22:32 mornfall you won't need it
22:33 gh_ but we like to have it, we are VCS people :-)
22:35 sm almost every day working on $dayjob, I wish they had preserved the original pre-darcs history :)
22:35 mornfall you could fastconvert it, filter out files that exist in darcs darcs repo, fastconvert it back and retain it...
22:35 sm yay fastconvert
22:36 mornfall that'll lose some history but most will stick around
22:37 mornfall (but then yes, it involves some extra work to retain that history)
22:42 sm time to eat - later gh_, mornfall
22:44 sm thanks for the chat
22:45 gh_ sm, see you!
23:13 gh_ joined #darcs

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