Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2014-05-01

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

All times shown according to UTC.

Time Nick Message
01:12 favonia joined #darcs
01:45 edwardk joined #darcs
02:28 mizu_no_oto joined #darcs
02:45 favonia joined #darcs
03:59 edwardk joined #darcs
06:44 raichoo joined #darcs
07:00 slyfox joined #darcs
07:00 slyfox joined #darcs
09:28 lelit joined #darcs
09:44 slyfox joined #darcs
09:44 slyfox joined #darcs
09:50 xstill joined #darcs
10:10 xstill joined #darcs
11:29 amgarchIn9 joined #darcs
11:35 raichoo joined #darcs
14:37 nomeata joined #darcs
14:48 siel joined #darcs
15:08 mizu_no_oto joined #darcs
15:20 gh_ joined #darcs
15:56 stepcut joined #darcs
16:57 edwardk joined #darcs
17:32 dino- joined #darcs
17:33 dino- I'm maintaining the darcs 2.8.4 build for Arch and just got a report from someone who's having terminfo dep problems..
17:34 dino- I'm doing it in a cabal-sandbox and am surprised that it's not just installing its own copy of terminfo-0.3.x.x (as the .cabal file asks for)
17:34 dino- This user has ghc 7.8.2 installed which has terminfo-0.4.0.0
17:35 dino- Starting to suspect I don't understand how this works with sandbox exactly.
17:36 dino- Does this sound familiar to anyone? Problems with recent ghc and terminfo
17:43 dino- Aha, I just found this thread on the darcs-users ml: http://comments.gmane.org/gmane.comp.version-control.darcs.user/26889
17:44 dino- So, we don't support ghc 7.8 yet
17:49 Heffalump is there a strong need to support GHC 7.8 with darcs 2.8?
17:49 dino- I doubt it. This is just one user I'm hearing from
17:49 Heffalump my general feeling is that supporting a new compiler is beneficial for a library's users, but not for an executable's users
17:50 Heffalump ok. There are a few complaints we've heard, so it may still be worth doing.
17:50 dino- Ok, I haven't been following the ml
17:54 owst joined #darcs
17:55 favonia joined #darcs
17:56 owst Hmm. Any reason there isn't a witness parameter for pending?
17:56 owst I'd like to make a function that adds to pending, but doesn't need the patches to be commuted past working-changes first.
17:57 owst But to make that witness-y I'd need a witness for the pending state.
18:02 owst My usecase is to allow renames of a file to a file that is deleted in working - we need to add patches that remove the recorded contents of that file, before rmfile, before adding a move patch. However, if we try and commute a hunk/rmfile patch like that past the changes w.r.t. working, we fail, since the workign changes are the same (leading to pending containing two copies of a "hunk and rmfile")
18:05 owst Err, I think I was being stupid - I need the remove patch to just be w.r.t recorded, so no need for a pending witness - move along, nothing to see here!
18:11 owst Oh, it's nothing like what I thought. I need my function to take a patch p w.r.t recorded, and the recorded->pending patches toPending, and then push p into toPending as far as possible, and dump any remaining (conflicting) pending that remains.
18:11 shlevy joined #darcs
18:11 shlevy Does darcs have tags or some other way of signifying a "version"?
18:13 owst e.g. if I have replace PATH foo bar and I then delete PATH in working and do darcs mv OLD_PATH PATH, I need to make pending look like: hunk -PATHCONTENTS ; rmfile PATH ; move OLD_PATH PATH
18:13 owst I.e. removing the replace patch
18:13 owst shlevy: yes, darcs has tags, they record a set of patches
18:15 shlevy owst: Ah great
18:15 shlevy hmm I see it in man darcs now
18:15 shlevy I must have made a typo in my search before :D
18:17 amgarchIn9 joined #darcs
18:51 carter Q: is darcs buildable bug free on 7.8 yet
18:51 carter ?
18:51 carter :)
18:53 Heffalump why do you ask?
19:05 sm good, good, that seems to have worked
19:05 Heffalump ?
19:09 sm carter has stopped asking that inconsiderate question
19:14 n-dolio Answer is no. darcs has some bugs on all versions of ghc, I'm sure. :)
19:30 Heffalump :-)
19:31 Heffalump I would like to hear if/why people need 2.8 on 7.8, as I'm sure it would be doable with some effort.
19:33 n-dolio Well, 2.9 was broken on 7.8, too.
19:33 n-dolio Still is, I think?
19:33 n-dolio Or was the race condition all of it?
19:33 Heffalump yes, though slyfox sent in a patch which I think we just need to apply
19:33 Heffalump that I'm still not quite clear on
19:34 schlaftier joined #darcs
19:34 schlaftier joined #darcs
19:35 carter the issue with the caches in particular
19:36 carter i've been using 2.9 for repoconversion purposes only so far, which evades the caches issue
19:38 sm carter: maybe you can be guinea pig for slyfox's patch
19:38 carter could be
19:38 sm http://bugs.darcs.net/issue2364
19:39 Heffalump carter: specifically I'm interested in why you need to use GHC 7.8
19:39 carter Heffalump: because all my dev is using 7.8
19:39 Heffalump and you don't have GHC 7.6? We could probably supply a binary.
19:40 carter i have a 2.8 binary on my mac just fine
19:40 Heffalump I mean of 2.9
19:40 carter well... i have a 2.9 binary built with 7.8, just it has that caches bug :)
19:40 carter sure
19:40 carter that'd work too
19:40 carter (2.9 isn't quite released yet is it)
19:40 Heffalump no
19:41 Heffalump ah, MacOS, I don't have that avaiable to build on
19:41 carter Heffalump: would the slyfox  patch fix things allegedly?
19:41 n-dolio Whenever arch gets around to shipping 7.8, it'll be annoying to have 7.6 around, too.
19:42 Heffalump n-dolio: right, so distros need consistency to keep things sane
19:42 n-dolio Arch doesn't ship darcs.
19:42 n-dolio You have to build it yourself.
19:42 amgarchIn9 joined #darcs
19:43 Heffalump slyfox's description of the problem: http://bugs.darcs.net/patch1149
19:43 dino- n-dolio: I'm keeping one of the AUR builds maintained for darcs, the one that builds 2.8.4
19:44 Heffalump which does suggest he is going to send a further patch
19:44 dino- Just went through this with a GHC 7.8 user and I ended up giving them my bin package I had kept around
19:44 carter what software runs the bug tracker http://bugs.darcs.net/ ?
19:44 Heffalump carter: roundup
19:57 amgarchIn9 joined #darcs
20:44 raichoo joined #darcs
20:55 mizu_no_oto joined #darcs
21:14 c74d joined #darcs
21:28 favonia joined #darcs
22:46 owst joined #darcs
22:47 owst What's a good name for a function: that commutes a FL p past a p: FL p :> p -> Maybe (p :> FL p)
22:47 owst We have commuteFL for the symmetric case
22:47 owst commuteFLSym? commuteFLSingle? commuteFLp?
22:52 owst Heffalump: can you recall a function that does something like a commute of a patch past an FL, but doesn't stop at the first commute failure - I want to remove all patches that don't commute, and make sure I do commute past every patch I can
22:52 owst I'm sure such a function must exist! :-)
23:23 owst Heffalump: just a thought, I'll give you some context of what I'm trying. When I add a move patch that moves into a deleted-in-working path, I need to delete the existing contents of the path. That's easy, just delete the path in recorded and grab a diff. Then, I need to add those patches to pending, such that any existing changes to the path are dropped from pending - I thought I could use commuteWhatWeCanFL
23:23 owst to push the patches inverse as far into pending as possible, and drop the rest, but this is wrong because there might be other changes that I need to keep after the first patch that wont commute. Therefore I think I need somesort of clever thingy that I asked about earlier, that'll allow me to commute past everything I can, but drop anything from pending that conflicts with my removal patches.
23:23 owst Phew. Bed!
23:35 edwardk joined #darcs

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