Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2015-03-06

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

All times shown according to UTC.

Time Nick Message
04:47 amgarching joined #darcs
06:25 mizu_no_oto joined #darcs
08:07 carter_ joined #darcs
08:42 IbnFirnas joined #darcs
13:26 gh_ joined #darcs
14:06 mizu_no_oto joined #darcs
15:17 mizu_no_oto joined #darcs
15:54 mizu_no_oto joined #darcs
17:14 Riastradh joined #darcs
18:44 mizu_no_oto joined #darcs
18:51 xauth joined #darcs
18:53 Heffalump sm: a few patches for you in http://hub.darcs.net/ganesh/darcsden - the 2.10 updates depend on a couple of changes in branch-2.10 that aren't merged in yet, but should be soon
18:59 alexei joined #darcs
19:29 mizu_no_oto joined #darcs
19:49 sm Heffalump: thanks! merged
19:53 alexei joined #darcs
19:59 Heffalump btw did you see my comment from a day and half ago or so (on here) about darcs-1 repos?
20:01 sm Heffalump: I did, no need for apologies at all. Hopefully Henning will respond
20:06 Heffalump I would actually quite like darcs-1 support for one my own repos if nothing else :-)
20:07 Heffalump anyway, I'll get the code written because it'll be useful for local dev, and we can see how things pan out for hub - I guess I could make it a configuration option
20:07 sm nod. At least under the current economics of darcs hub, I think the burden of motivation should be on the proposer. Why do you want darcs-1 for it ?
20:08 sm for that repo, I mean
20:12 sm since I always forget, pardon the following ascii art note to self:
20:12 sm patch semantics:   |------- darcs-1 --------| darcs-2 |
20:12 sm repository format: | old-fashioned | hashed | darcs-2 |
20:13 sm darcs hub currently only accepts darcs-2/darcs-2 repos, is that right ?
20:15 sm so much compilation failure on hackage.. I think I should move to stackage for darcs hub builds
20:31 sm not as easy as it sounds, either
20:32 sm I persevere
20:39 gh_ joined #darcs
20:48 Heffalump oh, yes, I had to constrain a bunch of packages to get it to build, I can send you the constraint line if that helps..
20:50 Heffalump why do I want darcs-1? Mainly because I think making a breaking repo upgrade starts at a fairly significant negative value, and I don't see much in the darcs-1 -> darcs-2 transition that overcomes that. I don't like exponential merges or the darcs-1 merge code, but they'll never hit that repo, and I don't like the duplicate semantics in darcs-2.
20:50 Heffalump (and I don't like the darcs-2 merge code much either, but it's marginally better than the darcs-1 code)
20:50 sm thanks, I'm updating a bunch of bounds and figuring out a more stable stackagish build process
20:50 Heffalump -constraint 'http-client-conduit<0.2.0.1' --constraint 'yaml<0.8.10' --constraint 'lens<4.7' --constraint 'pandoc>1.12.3.3'
20:51 sm I had several different constraints here. Building darcsden is a real pain
20:51 Heffalump on average I think about 2/3rds of the build failures I get when building things come from packages written by the person who runs stackage :-)
20:51 Heffalump hvr has been fixing cabal files on hackage recently which has made a huge difference
20:54 sm Heffalump: I'm finding it hard to parse a good reason for supporting darcs-1 from what you just wrote :)
20:54 sm can you run that by me again ?
20:55 Heffalump points in favour of migrating: (1) no more exponential merges (but value reduced because they're unlikely in that repo), (2) I don't like the darcs-1 merge code at all
20:56 Heffalump points against migrating: (1) I'm opposed to breaking repo format changes in principle (2) I don't like the darcs-2 merge code that much either (3) I don't like duplicate semantics in darcs-2
20:58 sm of these, only the exponential merge issue seems relevant to me as darcs hub operator
21:00 sm I don't want that bug running on my server, requiring the darcs-2 semantics fixes it (AFAIK)
21:00 Heffalump well, I guess the point for you as operator is that there are users who would like darcs-1 - I'm explaining why I (as a user) would like it
21:01 Heffalump it doesn't, there are still exponential merge cases in darcs-2, just quite a bit rarer
21:02 sm yes, it's good to hear why you want this. So far I don't find the reasons very compelling though
21:03 Heffalump I guess a slightly higher level point is that if we are going to force migrations on users, they might be more inclined to just migrate to git
21:03 sm I don't think there's much point trying to stem that tide
21:04 sm people really interested in darcs won't leave because of lack of darcs-1 support on darcs hub (I think!)
21:04 Heffalump anyway, if you really don't want to run it I'll see if I can host a darcsden instance targeted at darcs-1 repos myself, I don't imagine it'll need too much in the way of hardware resources
21:04 dolio If I may interject...
21:04 sm of course, please do
21:05 dolio I doubt most people would have those objections to upgrading to darcs-2, because they don't develop darcs.
21:06 Heffalump and therefore don't know the trade-offs well, you mean?
21:06 dolio Any time I've hit a 'your repo isn't darcs-2,' I just upgrade, at least.
21:06 dolio Right.
21:07 dolio Doing the upgrade is easy enough, and I suspect most people don't study enough to know the difference.
21:07 dolio It's possible I'm off base, though.
21:07 Heffalump I don't consider it easy, because I have to find all my own local copies of the repo, let alone other people's
21:08 Heffalump which is why I'm opposed to breaking the format of an existing repo on principle
21:09 sm I have yet to see examples of where a breaking upgrade is causing a real problem in darcs-land
21:09 dolio Does doing a convert on repos independently result in compatible darcs-2 repos? I'm not sure I've ever tried that.
21:10 Heffalump no, at least not in general
21:10 Heffalump in many cases it would in practice, but it's hard to characterise those precisely
21:10 dolio I see.
21:10 Riastradh You could take each darcs-1 repo, and make attempts to write to it fail until the owner of the repo decides to upgrade it to darcs-2.
21:10 Heffalump broadly, if there are no conflicts in the history at all, I think it'll be fine; if there are then it might end up being sensitive to the order of patches
21:11 Heffalump Riastradh: each darcs-1 repo where?
21:11 Riastradh darcshub?
21:11 Heffalump there aren't any there now, this conversation is about me wanting to add support.
21:11 Riastradh Oh.
21:11 Riastradh I was under the misapprehension that sm wanted to rid darcshub of darcs-1 repos.
21:12 Heffalump no, he's trying to keep the gate closed in the first place :-)
21:23 gh__ joined #darcs
21:32 sm incidentally hub.darcs.net has 1500 repos today, and oh! I've lost my number one spot
21:33 sm to http://hub.darcs.net/thielema
21:33 Heffalump wow
21:33 sm about time
21:33 Heffalump he must have a lot of darcs-2 repos as well then!
21:33 sm is that henning ?
21:33 Heffalump I think so
21:38 * sm seeks lunch o/
21:39 sm oh one random thought, around the 2.10 release would be a great time to get on FLOSS Weekly podcast, which currently has a short wait list and is actively looking for projects
21:46 lispy joined #darcs
22:52 mizu_no_oto joined #darcs

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