Camelia, the Perl 6 bug

IRC log for #darcs, 2013-10-24

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

All times shown according to UTC.

Time Nick Message
01:26 intripoon joined #darcs
01:45 arpunk joined #darcs
01:45 stepkut joined #darcs
02:51 mizu_no_oto joined #darcs
02:53 qinglingquan joined #darcs
02:54 qinglingquan left #darcs
03:13 preflex joined #darcs
03:16 mulander joined #darcs
03:20 sm notdan: a patch! oh, bless you
03:21 sm I'd forgotten what they looked like
03:21 stepkut` joined #darcs
03:22 sm your idea sounds great. Are there some whitespace changes in this patch ? it looks that way
03:28 sm what's the second hunk about ? changing getBox r True to False
03:31 sm also it doesn't compile.. needs a little more work
03:36 preflex_ joined #darcs
03:38 gh_ joined #darcs
04:18 notdan sm: oops sorry
04:19 notdan sm: yeah, I am a bad person and I sent in something untested... :(
04:19 notdan I am going to resolve the cabal hell that I am in now and try to compile everything again
04:29 gh_ sm, about http://irclog.perlgeek.de/​darcs/2013-10-23#i_7750700 , there is a patch in  screened, which is now at "version" 2.9.7 (+151 patches)
04:38 mizu_no_oto joined #darcs
05:16 kowey joined #darcs
06:16 lelit joined #darcs
06:42 arpunk_ joined #darcs
07:17 raichoo joined #darcs
07:29 raichoo joined #darcs
09:13 schlaftier joined #darcs
10:13 owst joined #darcs
10:26 donri joined #darcs
11:42 arpunk joined #darcs
11:46 mizu_no_oto joined #darcs
12:30 mizu_no_oto joined #darcs
13:49 kowey argh, darcs changes | less is still interactive on my Mac :-(
13:56 dolio Also, darcs log --interactive is not interactive. :)
13:56 dolio Actually, the help is just wrong, I guess. It still says it's an alias for changes. :)
14:15 mizu_no_oto joined #darcs
14:23 kowey don't we have tty detection via haskeline?
15:38 sm g'day all. Thanks gh_! I found it
15:39 sm notdan: if you need cabal help, shout
15:46 sm I closed a darcs hub issue last night, first change in a month
15:46 sm there's a lot of easy stuff in the tracker right now if anyone gets the urge
15:47 sm http://hub.darcs.net/simon/darcsden/issues
15:47 sm I'll match any bugfix or patch with a closed issue
15:48 sm this week, at least
15:52 gh_ hi
15:58 arpunk hi sm, i'm trying to build latest darcsden with ghc 7.6.3 but i'm getting errors on hsp-0.7.3 and ssh-0.2.12, shall i report them or are you aware of them?
15:59 sm hi arpunk, let me check
15:59 arpunk i looked up at the issues but didn't found anything related
16:00 sm that seems to be what I'm building with
16:00 gh_ kowey, ouch.. I've been thinking recently that changes should be non-interactive by default. Will contribute to the thread on the mailing list.
16:00 gh_ (after being a proponent of it being interactive by default, yes...)
16:02 sm arpunk: what errors ?
16:02 kowey it's troubling
16:03 arpunk sm: sorry, telephone. Here they are http://lpaste.net/94747
16:03 kowey I see us trying to do terminal detection on stdout
16:03 whaletechno joined #darcs
16:05 sm arpunk: ah, darcsden needs one extra hsp patch updating 0.7.x, you need to use http://hub.darcs.net/simon/hsp
16:05 gh_ kowey, yes: http://hub.darcs.net/darcs/darcs-screened/b​rowse/src/Darcs/UI/Commands/Changes.hs#148
16:06 sm and to get trhsx, I think install the latest hsx package. You could try adding that to the dependencies, I'm not sure why it's not one
16:06 arpunk ok, trying
16:08 sm although.. stepcut did release a 0.7.3, not sure what's up
16:10 arpunk sm: shall i get ssh from your repo aswell?
16:11 sm pretty sure I did upload my latest ssh as 0.2.12
16:11 sm none of this is quite making sense
16:12 arpunk for what it matters, i've tried on archlinux and fedora 19, 20
16:12 arpunk same results
16:13 arpunk hsp 0.9.2 builds fine
16:13 sm thanks, no need to try more I'll get to the bottom of this
16:14 sm darcsden requires hsp 0.7.x plus ghc update patch, 0.8 and up have a different API
16:14 gh_ n-dolio, about http://irclog.perlgeek.de/​darcs/2013-10-22#i_7748774 , see http://bugs.darcs.net/patch1088
16:15 arpunk alright, glad i could help. What about the ssh package?
16:17 n-dolio gh_: If you want to allow empty names for whatever reason, the bzr approach of making sure after the name is empty is probably sufficient.
16:17 gh_ n-dolio, yes, I've just tried with git and bzr, and I think the bzr approach is sensible
16:18 n-dolio Maybe the prompt should even have an (e)dit option.
16:19 n-dolio So if you accidentally record -m '' somehow, you can fix it up after you've picked patches without amend-record or something.
16:22 sm arpunk: ssh needed an upper bound on cryptohash, I've just released ssh 0.2.12.1. Thanks for the report
16:24 arpunk no problem!
16:25 sm so my comment about needing special hsp branch may be wrong, just installing hsx may be enough
16:25 gh_ n-dolio, yes good idea. The patch name check happens *after* the interactive hunk selection, so it would be nice to not loses someone's selection just because they wanted an empty patch name.
16:26 sm arpunk: though hsx 0.7.3 on hackage depends on hsx, so you should already have ~/.cabal/bin/trhsx
16:26 sm if so, it's probably not in PATH ?
16:29 sm empty patch name ? is it good to allow those ?
16:30 Heffalump no, it's not :-)
16:30 Heffalump IMO
16:31 sm I'm conflicted. I can see it being ok with a zillion automated changes eg from a wiki
16:31 arpunk sm: you are right, i did not have trhsx on my $HOME/.cabal/bin, retrying properly now
16:31 sm though I think gitit adds a message if user leaves it blank
16:34 gh_ sm, out of consistency with other RCS' we can safely prohibit empty patch names
16:36 gh_ I'm not 100% convinced empty patch names are inherently bad, since you can match a patch by date for instance. and patches always have a date.
16:37 sm though conceivably you could have two with the same date
16:37 sm and even the same hash ? dunno
16:38 mtp if they have the same date, same hash, and different size/content, that is probably bad
16:39 gh_ there is a random hash added to every patch to ensure they are different ( the Ignore-this: thing)
16:39 sm ah
16:40 gh_ because indeed, the timestamp only has a granularity to the second (automated tools can possibly create several patches with the same timestamp)
16:40 gh_ if I read timestamp correctly  (eg: 20130615195330 )
16:42 arpunk_ joined #darcs
16:49 arpunk_ sm: the deps are now all installing fine, however i'm getting this: http://lpaste.net/94749
16:49 notdan sm: thanks! I particularly don't know what to do about this error: cabal: ../../../projects/darcsden/cabal​-dev/bin/make-pandoc-man-pages: already
16:50 notdan arpunk_: oh, you too are trying to install darcsden :)
16:50 arpunk_ yea :)
16:51 n-dolio I expect empty patch descriptions are bad enough that you could probably justify requiring a flag to enable them.
16:54 arpunk joined #darcs
16:54 sm arpunk_: hmm. I just added that to darcsden to fix a warning, but it probably needs to be #ifdef DARCS28-ified
16:55 sm fix on the way
16:56 lelit joined #darcs
16:58 n-dolio I haven't checked our team's policy, but I bet empty patch names are not acceptable here. :)
17:00 gh_ n-dolio, bzr accepts empty commit names with « -m "" »; git does not. we seem to have a choice there.
17:01 n-dolio Anyhow, I'm not really invested in how that part resolves. I just don't want to say, "I'm sure I want to commit the things I just said I wanted to commit." :)
17:01 sm when can I depend on darcs 2.9 anyway ??
17:04 n-dolio And, from what I hear, if you're going to copy someone, bzr is a pretty good choice.
17:08 sm mercurial too, probably
17:09 n-dolio hg commit -m '' opens an editor.
17:09 n-dolio So apparently -m '' is a no-op there.
17:22 sm arpunk: darcsden head should build with darcs 2.8 again, thanks for the bug report
17:23 sm notdan: hi, huh that's weird
17:24 sm paste the full error ?
17:32 schlaftier joined #darcs
17:37 arpunk sm: thank you! it now compiles and install fine :), time to read the code
17:38 notdan sm: http://pastebin.com/iwJdMJCc
17:38 notdan It does not exist though, which is strange
17:39 notdan Hm now I restarted it and it gives me a different error, can't build ssh
17:40 notdan oh, need to update the package database maybe
17:44 * sm would blow away something
18:01 notdan Is it ok if 'darcs check' takes more than a minute?
18:02 gh_ on relatively big repos (like darcs' for instance), yes
18:03 notdan ok
18:07 marmoute n-dolio: mercuriel refuse to commit a changeset with empty description.
18:16 gbeshers joined #darcs
18:19 notdan sm: there is an inconsistent use of tabs vs spaces in the code. May I replace the tabs with spaces in the places where I edit the files or should I just leave it?
18:29 amgarchIn9 joined #darcs
18:31 sm notdan: whitespace fixes are very welcome. I usually try to put them in their own patch
18:31 sm what's changing getBox r True to False.. is that a bugfix ?
18:32 sm (the second hunk in yesterday's patch)
18:32 * sm wonders what to think when darcs check says the repository is consistent! and Hash mismatch(es)!
18:37 Heffalump hash problems are separate - you have to use a separate command for them (which I forget off the top of my head)
18:38 Heffalump darcs gzcrcs --check / darcs gzcrcs --repair
18:38 Heffalump oh, hash mismatches, not crc ones
18:42 notdan sm: hm it doesnt hacnge True to False
18:42 notdan *change
18:43 sm if I misread it, nm
18:51 notdan Ok, it should compile now
18:54 notdan Urgh, I screwed it up :( I've introduced an unrelated change in the whitespace commit http://hub.darcs.net/co-dan/darc​sden/patch/20131024184759-e7f45
18:54 notdan sorry I am still not proficient in darcs
18:54 sm no problem, no hurry
18:55 sm you can review the change locally before recording with darcs whatsnew, or darcsum-mode in emacs (the latter is good at highlighting whitespace)
18:55 notdan What should I do? Should I try to delete the patches I made (obliterate) and start over?
18:55 notdan Will I be able to "force push" the patches to darcsden if I do that?
18:55 sm yes, I'd obliterate the pushed patch and unrecord the local one and edit/and or record again
18:56 sm ssh hub.darcs.net obliterate darcsen
18:56 sm with a d
18:56 sm then darcs unrecord (the most recent)
18:56 notdan wow , I didn't know you can do that :D
18:57 notdan (ssh + obliterate)
18:57 sm yup. It's fine for dev branches like this, but best avoided with public repos people are pulling from
19:06 notdan I am curious, how does I reference a particular patch in the world of darcs? In git, there are sha hashes for that..
19:08 sm similar.. use darcs changes --xml to see hashes and --match 'hash xxxxx' (eg) to match on it
19:09 sm but normally in darcs land we just use the patch name or part of it; this matches the most recent patch name by regexp. darcs help patterns gives more ways
19:10 sm eg darcs changes -p 'my recent patch' -v
19:10 notdan oh, I haven't tried looking at xml output :P
19:10 notdan thanks
19:10 sm darcs log, I mean. Stupid interactive changes. :)
19:11 notdan ok I think this should be good now: http://hub.darcs.net/co-dan/darcsden/changes
19:19 sm notdan: where in the UI does the new text show up ?
19:23 sm oops I jumped to the wrong port
19:23 sm http://hub.darcs.net:PORT/simon/darcsden shows it not working.. PORT is 8902
19:24 sm (obfuscated to keep bots out)
19:24 sm notdan: it's close. I have to run out right now
19:25 sm running a local instance will help you a lot, or we can look at it later
19:25 sm o/
19:50 whaletechno joined #darcs
20:10 mizu_no_oto joined #darcs
21:42 arpunk_ joined #darcs
22:03 Mdx joined #darcs
22:04 Mdx Hey, anyone with a minute to spent? Having some trouble with darcs and windows
22:11 Heffalump sure
22:12 * Heffalump actually uses darcs on windows so has some hope of being useful
22:25 amgarchIn9 joined #darcs
22:41 Mdx_ joined #darcs
22:41 Heffalump Mdx_: I can try to help with darcs on windows
22:42 Mdx_ I resolved my problem, but maybe I should report what happened
22:42 Mdx_ oh sorry, didn~t saw you sent a message before
22:42 Heffalump np
22:43 Mdx_ So, first of, darcs 2.8.1 (that I got from the website) crashes when I try to do darc get.  using 2.5.1 worked
22:44 Mdx_ but before trying to get the binaries, I tried to compile the code, but cabal install is not working
22:44 Mdx_ this happens http://lpaste.net/94768
22:44 Heffalump eek
22:44 Mdx_ Idk if this is a cabal bug or a specific cabal bug
22:44 Mdx_ ops
22:44 Heffalump did you try cabal update?
22:44 Heffalump I don't think I've ever seen it before
22:44 Mdx_ specific to darcs*
22:44 Mdx_ yes, i tried
22:45 Mdx_ let me try once more
22:45 Heffalump does something like "cabal install rmonad" work? (Just looking for something simple to try)
22:45 Mdx_ funny thing is, if i go to hackage and download manually, i can unpack it with 7zip
22:45 Mdx_ lol, it didnt work
22:45 Mdx_ (rmonad)
22:46 Heffalump same error?
22:46 Mdx_ yes
22:46 Mdx_ wth happened, cabal was working yesterday
22:46 Heffalump go to %APPDATA%\cabal and see if the tarball there corresponds to the one you can download manually
22:46 mizu_no_oto joined #darcs
22:47 Heffalump (are you generally familiar with the Haskell ecosystem?)
22:47 Mdx_ not really, but i guess is the folder on AppData\Roaming\cabal
22:48 Mdx_ tarball is on this path: cabal\packages\hackage.haskell.org\darcs\2.8.4
22:48 Mdx_ deleting it doesnt fix it, just forces to redownload
22:49 Heffalump and is it the same contents as the one you can download manually and unpack?
22:49 Heffalump perhaps try downloading manually and putting in that folder, then seeing what happens
22:49 Mdx_ wait, what I've said is wrong, it's not actually forcing to redownload
22:50 Mdx_ downloading manually, you mean by hackage or using cabal?
22:50 Heffalump doing the same thing you said above where you could unpack with 7zip
22:51 Mdx_ one sec
22:54 Mdx_ same error
22:54 Mdx_ but it's happening to whatever isntall comand i use... darn
22:56 Mdx_ I'll go back to #haskell and ask there
23:48 sm notdan: I think the correct syntax is value=(cloneUrl owner)

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