Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2015-05-07

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

All times shown according to UTC.

Time Nick Message
00:00 sm ok, updated
00:06 maerwald I had trouble with linking the existing account, but creating a new one worked
00:07 sm hm
00:07 sm how can this be tested without nuking one's account(s)
00:07 maerwald sm: can you remove the hasufell1 user? I'll try again then
00:07 maerwald :o
00:07 sm can you not delete it ?
00:08 maerwald sm: how?
00:09 sm I guess you can't, ok
00:10 maerwald yep, I think we also need to fix the scope in "getSyncWithGithubR"
00:10 maerwald that's when you link the acc afterwards
00:11 sm ok, deleting user hasufell from hub.darcs.net (both production and dev instances). I'll probably need to delete your repos too. Ok ?
00:11 maerwald sm: what, wait
00:11 maerwald sm: I meant "hasufell1"
00:11 sm aha
00:12 sm ok, deleting user *hasufell1* from hub.darcs.net (both production and dev instances). And any repos owned by it.
00:12 sm ok ?
00:13 maerwald yes
00:13 sm maerwald: done
00:14 maerwald sm: http://hub.darcs.net/maerwald/darcsden/patch/20150507001259-0bd70 we need this
00:14 sm I've deployed your patch at hub.darcs.net also
00:14 sm ok, stand by
00:14 maerwald I mean "this too"
00:15 sm ok, deployed this on both instances
00:16 maerwald yes, works :)
00:16 sm \o/
00:17 sm maerwald: nice work, thank you
00:18 maerwald sm: I think long-term it would make sense to show in the request handler not only the response we got for a StatusCodeException, but also what was actually fetched. If that would have shown up in the logs, it would have saved us a lot of time
00:19 sm indeed
00:19 maerwald I mean, the broken github link "works", but has no usable data to parse
00:21 sm I'll close this bug after checking with a few other reporters
00:21 maerwald ok, I'll go to bed, thanks for putting up with me :)
00:21 sm thanks for the fix! see you
00:21 maerwald left #darcs
00:33 ichor joined #darcs
01:41 fr33domlover joined #darcs
02:03 Riastradh joined #darcs
02:37 c74d joined #darcs
02:55 mizu_no_oto joined #darcs
03:42 ichor joined #darcs
04:32 c74d joined #darcs
06:13 Heffalump notdan: yes, I think you're right about those messages
06:16 zyla joined #darcs
08:21 amgarchIn9 joined #darcs
08:32 amgarchIn9 joined #darcs
08:57 amgarchIn9 joined #darcs
10:31 haasn`phone joined #darcs
10:43 haasn`phone left #darcs
11:00 mizu_no_oto joined #darcs
11:02 notdan Heffalump: do you want me to submit a patch on the tracker?
12:09 notdan sm: what are the source of icons used on darcsden?
12:18 _loic_ joined #darcs
13:47 Riastradh joined #darcs
13:49 sm notdan: good question
14:21 notdan http://bugs.darcs.net/patch1332
14:21 notdan I tried doing `darcs send --mail', but for some reason it didn't work
14:31 sm huh, I didn't know about darcs dist
14:37 notdan Well that's what I am trying to use for the "download .zip" button :) http://hub.darcs.net/co-dan/darcsden/patch/20150507125131-80de2
14:38 notdan What I really like abouut Haskell is that I have spent some hours writing the code, yet the patch is still tiny!
14:38 sm I noticed that
14:39 sm cool
14:39 sm I tried to find the source of those icons, I'm starting to wonder if alex suraci made them
14:43 sm notdan: is it just me or are the patches buttons failing in your repos ?
14:44 sm branches button I mean
14:49 notdan hm indeed it does
14:49 notdan relative/absolute path issue?
14:50 notdan well I think it is failing in darcsden in general: http://hub.darcs.net/ganesh/darcs-hs-merge/patches
14:50 sm something has deteriorated
14:54 ichor Hi all! Would there be any difference to darcs between a tag and a patch which explicitly depends on all previous patches?
14:54 ichor I noticed that I am not allowed to name such a patch "TAG something".
15:03 sm hi ichor, that's essentially what a tag is
15:04 ichor sm, I see, but does darcs treat tags any differently? Or could one create tags "manually".
15:05 sm I'm guessing not
15:06 sm check what the tag command does
15:07 notdan sm: sorry, I have a question that is not very related to the problem at hand: what do you intend to do with all that #if DARCS28 stuff?
15:08 sm notdan: rip it out! I thought I recently did that
15:09 sm http://hub.darcs.net/simon/darcsden/patch/20150501184047-3c3f9
15:14 ichor sm, thanks! I checked the source and my questions were answered.
15:14 notdan oops, sorry sm, I didn't pull those changes for some reason
15:14 notdan sm: btw, you develop darcsden against darcs-reviewed, right?
15:15 notdan Because I make some modifications to darcs itself I guess I will have to figure a way to develop against screened
15:15 notdan :(
15:15 sm notdan: prior to 2.10, I was mostly building against darcs-screened. Using 2.10 right now
15:15 sm I don't bother with darcs-reviewed
15:16 notdan ah, fair enough. I guess the question is: do you want to see patches that work against screened?
15:18 sm I suppose you'll need to depend on screened sometimes. I'd like to be able to merge those in trunk too, but keeping darcsden installable with darcs 2.10 for the foreseeable future, so I suppose we'll need flags here and there
15:18 sm actually, is it possible to do your gsoc and have it still depend on a released darcs ?
15:18 sm flags are a pain
15:21 notdan sm: hm, well definitely not for everythingg
15:22 notdan This "download archive" thing, for example, required some modifications to darcs
15:22 sm maybe we can get it out as a darcs minor release
15:22 notdan and I fully agree that flags are pain in the arse!
15:23 notdan yeah, but that should be finilize towards the end of the soc
15:24 sm I should make a darcsden release asap, depending on darcs 2.10 only
15:25 sm then I think we probably do want a flag for darcs-screened-specific stuff, only to reduce the headaches for people trying to hack on darcsden
15:26 sm they should be able to build with darcs 2.10 if they want
15:27 notdan word
15:28 notdan sm: OK, regarding the forks error: it seems weird, but if you actually check, the _darcs/inventory file does not exist! (at least in my sstem)
15:28 sm does not surprise me :)
15:29 * sm got desensitized to missing file complaints from darcs' subsystems
15:30 notdan http://hub.darcs.net/darcs/darcs-screened/browse/src/Darcs/Repository/Format.hs#122 this seems to be the source of the problem
15:31 notdan This function is kinda weird: it  never returns Darcs2 as a format..
15:31 notdan oh wait nvm
15:31 sm there should be a nice accurate reference doc describing each version of darcs' repo file structure
15:32 notdan I agree
15:40 maerwald joined #darcs
15:45 notdan hm, as it turns out, the "real" error is that darcsden cannot cess te format file
15:48 notdan ah, which is totally understandable, because /forks calls 'getForkChanges' from Forks.hs, which changes the home directory!
15:49 notdan s/home directory/current directory/
15:49 notdan this is why i am suspicious about depending on the current directory so much
15:54 notdan http://hub.darcs.net/simon/darcsden/issue/133
16:25 mal`` joined #darcs
16:33 sm notdan: thanks!
16:33 sm is this due to a change in darcs, or darcsden ?
16:33 notdan Oh, this I don't know
16:35 sm darcsden is built non-threaded to work around libdarcs's tendency to change the current directory
16:37 mankaev joined #darcs
16:38 maerwald unfortunately darcsden doesn't build on FreeBSD due to some linking problem with system-uuid http://vpaste.net/pZGtT
16:42 notdan Well, the error was definitely much clearer in darcs 2.8 :[
16:42 sm I think I've built on osx
16:42 notdan http://hackage.haskell.org/package/darcs-2.8.5/docs/src/Darcs-Repository-Format.html#tryIdentifyRepoFormat
16:43 notdan maerwald: try switching from system-uuid to uuid
16:44 maerwald notdan: what do you mean with "switch"? Just change system-uuid to uuid in the cabal file?
16:45 notdan Yaeah
16:45 notdan actually no, sorry, that wouldn't work
16:47 notdan maerwald: also replace Data.UUID.V4 for System.UUID.V4
16:47 notdan in src/DarcsDen/State/Session.hs
16:47 notdan and replace uuid with nextRandom
16:51 maerwald notdan: it seems to compile at least... given that system-uuid is an outdated package, shouldn't darcsden switch?
16:52 notdan Probably. is it very outdated?
16:52 notdan tbh, I have no idea about that package
16:52 sm sounds good
16:52 notdan it just makes a lot of foreign C calls, judging from the source  code
16:53 notdan whereas the uuid package is a pure haskell implementiation of the UUID spec
16:53 sm maerwald: actually, why do you say system-uuid is outdated ? http://hackage.haskell.org/package/system-uuid doesn't say that
16:54 maerwald sm: last commit in 2011 and causes linking errors under FreeBSD... I consider that outdated, laxly spoken
16:54 sm yes I see
16:56 notdan http://hub.darcs.net/simon/darcsden/issue/99
16:56 notdan maerwald: I mean it still works. It's just a bunch of simple bindings. The problem is that it does not work under FreeBSD, it's just outside of the scope of the packagee
16:56 notdan anyway, +1 on switching to uuid
16:57 notdan sm: I have the power to close issues on darcsden's issue tracker. Is that intended?
16:57 maerwald notdan: given that there is a pull request not answered after ~3 years https://github.com/solidsnack/system-uuid/pull/3 I will not open a bug report upstream
16:57 sm notdan: hmm.. seems a bit odd
16:58 maerwald I also see the "and close" button
16:58 maerwald but I didn't click it yet
16:59 sm need some notifications from that bug tracker
17:00 notdan All right, I need to stop hacking
17:01 sm me too. Later notdan, maerwald
17:28 maerwald mh, darcsden really seems to have trouble with couchdb-1.6.1
17:28 maerwald no useful error message
17:39 lambdabot joined #darcs
17:41 Riastradh joined #darcs
18:26 maerwald can I set a non-default port for pushing to a darcsden server?
18:29 maerwald what would be the syntax for darcs push then?
18:31 sm maerwald: something like SSH_PORT=8922 darcs push ... maybe
18:38 maerwald sm: do you have an idea https://gist.github.com/hasufell/5f162e540851c3d1eeaa <- happens on push
18:39 maerwald there is a "inventories" directory, but not "inventory"
18:46 sm maerwald: something to do with darcs-1 vs darcs-2 repo ?
18:46 sm I have darcs-1 support disabled in Production.hs, and all my repos are darcs-2
18:46 sm just a wild guess
18:53 maerwald sm: oh, I think it is a problem that the darcsden repository clone is set as my homedir, so the user repositories are NESTED repositories
18:58 maerwald mh no, doesn't work
19:01 notdan that inventory message is hella confusing
19:01 notdan maerwald: it is  a problem with paths/directory
19:01 maerwald well, it's FreeBSD, so it's well possible yes
19:02 notdan Are you running darcsden in production or in dev mode?
19:02 maerwald production
19:03 notdan do you have the correct directory structure in /home/darcsden?
19:04 maerwald notdan: can you be more specific? I have cloned darcsden in my homedir, build it and left the homeDir variable at ".", so it is executed from /usr/home/darcsden/darcsden
19:05 maerwald the web page works fine
19:05 notdan hm, sorry, a stupid question: are you running darcsden from hackage or from HEAD?
19:05 maerwald HEAD
19:06 maerwald http://darcs.cs-minden.de/
19:07 notdan oh nvm then, sorry
19:07 notdan can you check that /usr/homedarcsden/darcsden/users/hasufell/test is really a darcs repo?
19:07 notdan i.e. can you push to it directly?
19:11 maerwald notdan: you mean "darcs push /usr/home/darcsden/darcsden/users/hasufell/test"?
19:11 maerwald from within the server ofc
19:12 maerwald it wants to push and then fails with: darcs: darcs: createProcess: runInteractiveProcess: exec: does not exist (No such file or directory)
19:13 notdan :|
19:13 notdan wow this looks pretty bad
19:16 maerwald those could be two different errors though
19:16 maerwald maybe darcs depends on specific binaries hardcoded somewhere?
19:16 maerwald I've had something similar with gitlab
19:20 maerwald yep
19:20 maerwald at least something was missing, works now
19:20 maerwald so I can push locally
19:22 maerwald ok, all works now... I installed darcs via the package manager as well and it probably pulled something in that was missing, no idea what exactly
19:22 maerwald maybe hs-unix-compat
19:23 maerwald http://darcs.cs-minden.de/ if someone wants to test
19:28 sm looks good
19:29 sm I see the signup flow is a little confusing. Eg it shouldn't offer to sign up with google if that's not configured
19:29 maerwald ah
19:30 sm and when you first sign in using github, it shows a user creation form with "sign me up", "sign up with github", "sign up with google" buttons, should show only the first
19:31 sm and maybe the register button can be dropped
19:33 amgarchIn9 joined #darcs
19:33 maerwald sm: is that different on hubs.darcs.net?
19:35 sm no it's the same there
19:36 maerwald also, it would probably be cool to be able to pass or configure the ssh port for darcs on push
19:36 maerwald and then darcsden should show the appropriate command
19:36 sm SSH_PORT worked, right
19:36 maerwald yes
19:37 sm that could be nice info to have on the default front page
19:37 maerwald it'd be better to be able to configure it per-repository though
19:37 sm woah
19:38 maerwald in git it's part of the remote url
19:41 maerwald but for now I'll just hack src/DarcsDen/Pages/RepoPages.hs to include the port
19:48 maerwald mh, the ssh port should probably not be an argument for darcsden-ssh, but be configured in Settings/Production.hs... otherwise you cannot read it from within the code
19:51 sm on the other hand, we'd like to move settings out of code and into a config file/options for usability
19:51 sm rebuilding just to change your url or port sucks
19:51 maerwald sm: that would be the next step :P
19:51 sm I feel this step is always a bit dicey with haskell :)
19:58 _loic_ left #darcs
20:19 maerwald sm: I have it... you'll see the new message when you create an empty repository
20:21 sm nice
20:25 maerwald sm: http://hub.darcs.net/maerwald/darcsden/patch/20150507202206-0bd70
20:26 maerwald I hope <% "" %> doesn't do anything weird
20:27 maerwald this is if portMsg evaluates to ""
21:40 Heffalump notdan: yes please (re the wrong messages)
21:42 Heffalump sm/notdan: I'd expect to get small API changes needed for darcsden or other clients into the 2.10 series
21:43 sm maerwald: that
21:49 maerwald where did the ($-) in Darcs.Repository go? I'm trying to compile darcs-fastconvert against 2.8.5
21:50 Heffalump I think it might be redundant and you can just use ($) now
21:50 Heffalump try that anyway
21:50 Heffalump the functions to run on a repository are now wrapped up in an explicit "Job" type
21:55 maerwald and "identity" seems to be gone as well
21:56 maerwald previously from Darcs.Patch
21:56 Heffalump I remember removing that on purpose. But I can't remember why.
21:57 Heffalump I think darcs didn't actually use it for anything and it was cluttering up the Patch type
21:57 Heffalump do you know what darcs-fastconvert wants it for?
21:58 sm maerwald: that (and the uuid patch) look good, thanks! merged and deployed
21:58 maerwald adddeps (infopatch info ident) deps... where ident = identity :: FL RealPatch
21:59 maerwald sm: the branch site is broken for me, how about you?
22:00 maerwald http://hub.darcs.net/simon/darcsden/patches
22:00 sm yes we opened an issue for it today
22:00 sm quite inconvenient
22:00 maerwald ok
22:00 sm for the ssh message, even better if it could be aware of whether an ssh server is running or not
22:01 Heffalump if we can get a test for the branch issue we might be able to track down what introduced it
22:02 Heffalump maerwald: oh, if it's an FL RealPatch then it can just be replaced by NilFL
22:03 Heffalump but what version are you hacking on if the type is RealPatch rather than RealPatch Prim or with some other argument?
22:03 * sm works on the latest darcsden-ssh deployment problem
22:03 maerwald Heffalump: was identity some sort of mempty?
22:04 Heffalump well, kind of, but it required an explicit constructor in things like Patch itself
22:04 Heffalump or rather in primitive patch types
22:04 maerwald another occurence is (identity :: RealPatch)
22:04 Heffalump anyway, FL is just a list type, so NilFL (i.e. []) is fine
22:05 Heffalump can you point me at the repo you're using?
22:05 maerwald http://repos.mornfall.net/darcs-fastconvert/
22:06 Heffalump http://hub.darcs.net/mornfall/darcs-fastconvert seems to e one more patch up to date
22:06 maerwald meh, then the hackage site is out of date
22:07 Heffalump yeah :-(
22:07 Heffalump you know some of the import/export code from fastconvert is now in darcs proper? (I can't remember how much)
22:07 sm isn't darcs-fastconvert effectively built in to darcs 2.10 ?
22:07 Heffalump I don't think it was completely done
22:08 sm maerwald: darcs convert works very well, for one git branch at a time
22:09 maerwald maybe this is a delicate questions, but where are the haddock comments in darcs :P
22:10 maerwald I don't think I'd be able to understand this in less than 3 Months
22:11 sm a good question. darcs originally used a lot of literate documentation (.lhs files, now all separated I think), and it might have got going before haddock was much used
22:12 Heffalump missing, I'm afraid. If you ask me enough questions you might be able to prod me into writing some more.
22:14 sm maybe some of the wiki docs could be moved into haddock as well
22:14 * Heffalump dodges the political excitement by going to bed
22:15 sm oh ?
22:16 sm you mean wiki vs haddock vs lhs, or something even more exciting ? :)
22:16 sm well good night
22:17 sm so, my latest darcsden-ssh binaries refuse client authorization because "Roaming not allowed by server"
22:19 sm also, a dev build of darcsden-ssh segfaults when I try to connect to it
22:20 sm ssh is finicky
22:22 maerwald sm: does he mean loitering not allowed?
22:34 maerwald now the type checkers is going insane, I give up
22:46 maerwald darcs convert doesn't seem to work with git repositories
22:47 sm sure it does
22:47 maerwald oh, I was reading http://darcs.net/DarcsBridgeUsage
22:48 sm everything you read about converting darcs repos other than convert's help is probably out of date
22:49 maerwald there is not a single word about git in converts help
22:49 sm darcs 2.10 ?
22:49 maerwald oh, I guess it's still 2.8.5
22:50 sm yes this came in 2.10
23:20 sm Heffalump: http://hub.darcs.net/ganesh/ssh/issue/3 , also I think you might have a few ssh patches unpushed to hub

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