Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2015-04-22

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

All times shown according to UTC.

Time Nick Message
00:00 dolio dino-: On 7.10, I have a giant rpath.
00:15 dino- ok
00:16 dolio I don't know what version I built with that didn't have the rpath.
00:25 dolio Even when I tell it explicitly to not use dynamic linking or shared libraries, it has a huge rpath.
00:29 dino- How are you telling it to do that? I keep breaking my build with things like --ghc-options="-static" for instance, which fails when it gets to the linker
00:33 dino- oh hm, maybe I'm missing -optl-static
00:35 dolio 7.6.3 produces no rpath.
00:37 dolio Oh, in .cabal/config, I have two options.
00:37 dolio shared: False    executable-dynamic: False
00:40 mizu_no_oto joined #darcs
00:45 dolio 7.8.4 also has a huge rpath.
00:45 dino- I have those too in my (very old) .cabal config
00:45 dolio So, something changed from 7.6.3 to 7.8.
00:52 dino- dolio: What version of cabal are you using?
00:52 dino- I'm reading something now from January that talks about cabal 1.22 taking over rpath things where 1.20 did not used to do that
00:52 dino- https://github.com/haskell/cabal/issues/2330
00:52 dolio Cabal 1.22.2.0 with cabal-install 1.22.2.0
00:53 dino- I'm using 1.22.0.0, and I'm pretty sure I installed it not that long ago.
00:53 dolio There's a minor version greater than mine. 1.22.3.0
00:53 dolio I'd be surprised if that changed  things.
00:55 dolio Anyhow, assuming that there's nothing you can do to change this...
00:55 dolio One thing you can do is change the pkgbuild to require all dependencies of darcs to be installed via pacman.
00:56 dolio And actually use dynamic linking, I guess.
00:56 dolio And those will all be installed globally.
00:56 dolio I don't know if that's an acceptable solution.
01:09 dino- dolio: Well, it's a lot of stuff. I've been trying to get away from arch packages for Haskell stuff, except for ghc and cabal (sometimes)
01:09 dino- bleh
01:10 dino- Thanks for working on this with me. Going to attack it more tomorrow.
01:11 dolio No problem. You might want to ask in #ghc, though.
01:11 dino- But had this vague hope that we could still have binary things made with Haskell as Arch packages ^
01:12 dino- yes, this is all about the ghc really
01:29 gh joined #darcs
01:39 Riastradh joined #darcs
02:22 Meeh joined #darcs
03:14 dolio dino-: FYI, cabal 1.22.3.0 still creates a big rpath.
03:53 dolio So, do other people see errors when trying to get some of my repos? vector-algorithms, CC-delcont, logict, bytestring-show
03:55 dolio Complaints about pristine.hashed involving filenames from inventory.
03:55 dolio I think those are all repositories I've converted from old darcs formats.
03:56 dolio (They're on hub.darcs.net; I forgot to mention.)
07:43 Wu joined #darcs
08:33 dino- joined #darcs
10:38 c74d joined #darcs
11:25 mizu_no_oto joined #darcs
12:46 mizu_no_oto joined #darcs
12:48 Wu joined #darcs
14:00 sm dolio: yes
14:01 sm darcs get http://hub.darcs.net/dolio/vector-algorithms -> Couldn't fetch `0000005842-8dd07a0f85e758f6b6341777fa785cf692e42d43a012f69a804374fe8f4f9a82' in subdir pristine.hashed from sources
14:02 sm thisrepo:/tmp/vector-algorithms  cache:/home/simon/.cache/darcs  readonly:/home/simon/.darcs/cache  repo:http://hub.darcs.net/dolio/vector-algorithms
15:52 dolio sm: Right.
15:52 dolio And not with --lazy
15:57 dolio It doesn't affect all the repos I've converted, though, so I guess it's not that.
15:57 dolio pts and upts have been converted from older formats.
16:36 lambdabot joined #darcs
16:38 dolio Those barely have any patches beyond the convert, though.
16:56 lambdabot joined #darcs
17:26 dolio Apparently the 2.9 darcs I was using before did make the same complaint, but 2.8.4 does not.
17:48 sm dolio: anything related in the bug tracker
17:52 dolio Not that I can see.
18:25 fr33domlover joined #darcs
18:32 sm what convention do you use for keeping branches in the working dir for easy access.. maybe darcs get . _BRANCHNAME ?
18:44 dolio I have, say, vector-algorithms/{head,release,some-work,...}
18:48 dolio I don't tend to nest, though.
18:48 dolio Just have things parallel.
18:49 sm nod
18:49 sm I'm dealing with that version number conflict between branches you were asking about the other day
18:50 sm maybe --skip-conflicts sometimes helps there
18:52 sm also I'm finding it frustrating/puzzling to a typical user that a one-line change adding ".1" results in a seemingly arbitrarily large & complex conflict when you pull --mark-conflicts
19:16 sm is it true that you can't push any conflicting patch to darcs hub, even together with a resolution patch ? I think so
19:18 sm note to self: push to VCS before uploading release
19:18 dolio I think I did it the other day.
19:18 dolio No, I pushed the resolution to my version bump and such.
19:19 sm after pull --mark-conflicts, I should have *amended* the patch, not record a new one
19:23 dolio I used a patch...
19:24 dolio Or, recorded a new one, that is.
19:31 sm then I'm totally confused
19:35 sm I fear I'm struggling to do a simple release management task with darcs[hub] and would have been finished an hour ago with git
20:07 sm it was probably foolish of me to try and shoehorn darcsden 1.1 -> 1.1.1 changes into darcsden head, as if they had been made in the past. If I want those in public history, I have to just publish a separate branch/repo
20:08 sm got carried away with the darcs magic
20:09 dolio It would be cool if darcsden could organize branches for a particular project via internal subdirectories.
20:09 dolio I think I filed a wish item long ago about that.
20:10 sm I was also thinking today it should consider repos with the same name prefix as related (branches)
20:11 dolio Probably want some indicator besides mere prefix.
20:11 dolio Like, not everything prefixed with 'vim-' is a branch of one project. :)
20:12 dolio Something more illegal. Like /.
20:26 sm yeah I suppose
20:32 dolio I'd probably prefer it to manage branches internally, though. Because depending on the required naming scheme, I might not think to put something in that format until it's too late.
20:33 dolio Unless foo and foo/branch were recognized as foo being the main branch or something.
20:34 sm sounds good to me
20:34 sm except having / in names is not ideal
20:35 sm how about :
20:35 sm or .
20:36 dolio Not sure. . might have uses in actual project names. Wouldn't surprise me.
20:36 dolio : is probably safer, unless you're looking to recruit perl users or something.
20:36 sm it wouldn't have to be perfect, just a convenience for hub users
20:39 sm separately, should it automatically treat user1/somerepo and user2/somerepo as "branches" (forks, related repos)
20:41 sm also what about user1/somerepo[:topic1] and user2/somerepo:topic2
20:42 sm I think those'd be reasonable
20:42 dolio Dunno. Launchpad does that, I think, and it's pretty cool.
20:42 dolio Tracks all forks everywhere of a project as the same project.
20:42 dolio And can show you how they relate.
20:43 dolio It's much more project oriented than github and the like (which are user-oriented).
20:45 dolio That approach avoids certain pitfalls.
23:13 povman joined #darcs
23:14 mal`` joined #darcs
23:14 povman joined #darcs
23:14 povman Morning
23:15 povman Will there be a hackathon some time this year?
23:45 povman I particularly want to help speed up `darcs rec -l` when there are a large number of files.
23:48 povman In particular, `git add -A && git commit -a` takes 1m15s, while `darcs rec -la` takes 5m34s

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