Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2016-01-22

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

All times shown according to UTC.

Time Nick Message
02:23 drostie joined #darcs
02:49 mizu_no_oto joined #darcs
03:39 mizu_no_oto joined #darcs
04:43 Riastradh joined #darcs
04:46 mizu_no_oto joined #darcs
05:21 Big_G joined #darcs
05:55 Big_G joined #darcs
09:00 smidge joined #darcs
09:05 smidge joined #darcs
09:30 gal_bolle joined #darcs
10:31 gh_ joined #darcs
10:50 smidge_ joined #darcs
11:01 smidge joined #darcs
11:09 smidge joined #darcs
11:11 smidge joined #darcs
12:03 mizu_no_oto joined #darcs
12:27 gh_ damn, just when I finish writing a test about filepaths conversion between darcs and git, I remember about owen's darcs-bridge repository, and I see it's full of tests
14:20 Riastradh joined #darcs
14:50 drostie joined #darcs
15:08 sm my god! it's full of tests!
15:12 sm http://img.picturequotes.com/2/19/18158/my-god-its-full-of-stars-quote-1.jpg
15:19 gh_ :)
15:42 gh_ so I ported 2 pieces of code from owen's darcs-bridge: escaping/unescaping of filenames http://bugs.darcs.net/patch1436 and deletion of empty directories on import http://bugs.darcs.net/patch1438
15:42 gh_ and I believe they should also go to branch 2.10
15:43 gh_ there are other changes that could be ported from darcs-bridge. support of file moves for convert import can be nice to have.
15:44 gh_ git does not support file moves but git fast-export has a flag that can detect them based on file contents similarity, and mark them
16:14 gh_ ok file move support is cool, we need it
16:18 sm go gh_
16:26 owst joined #darcs
16:26 owst Hi :-)
16:39 gh_ hey owst :)
16:39 gh_ it's been a while
16:39 owst Indeed, long time no speak!
16:40 gal_bolle hi owen
16:40 owst I have been keeping my eye on the mailing lists and noticed you're doing heroic work porting some darcs-bridge things over gh_
16:40 owst Hey gal_bolle
16:41 owst I'm not around this weekend, but I'll probably come and hang out on IRC next week - I need to get back into some darcs work :-)
16:41 gh_ owst, I'm looking at porting some stuff from your darcs-bridge code to darcs, but in a short-term way
16:42 owst Very interested by the pijul work you've been doing gal_bolle - I'd like to understand the theory a bit more
16:42 gh_ owst, nice :)
16:42 gh_ owst, I see that to support renames (or moves) importing, I would need to change the type of InCommit..
16:42 * owst takes a look at my darcs-bridge repo
16:43 gh_ which currently, in darcs itself, is mosty code taken from petr's darcs-fastconvert
16:43 owst Crickey. I do *not* remember this code!
16:44 gh_ now, file move import would be very nice to have since every relatively big git repo I have on my machine has losts of moves in the output of "git fast-export -M"
16:44 owst Hmm!
16:44 gh_ :)
16:45 owst gh_: can you point me to where InCommit is defined in the screened repo?
16:46 gh_ looks like an important change is "Build up a RL of patches when processing commit contents"
16:46 gh_ *patch
16:46 owst Ah yes, that sounds vaguely familiar
16:46 gh_ (in darcs-bridge)
16:46 gh_ yes
16:47 gh_ http://hub.darcs.net/darcs/darcs-screened/browse/src/Darcs/UI/Commands/Convert.hs#689
16:48 gh_ indeed the patch "Handle renames/copies when importing" comes right after
16:49 mizu_no__ joined #darcs
16:49 gh_ and then there are 240 more patches in darcs-bridge
16:49 gh_ !
16:50 owst Heh, I generated many patches, but not quite a working bridge!
16:54 gh_ owst, if you're still a student this year and want to apply for gsoc again we may fix that! :)
16:55 * gh_ looks at "darcs diff --from-hash b138854b72  --to-hash dd804b514f0" inside of the darcs bridge repo
16:55 owst Alas I am not, I've been in industry for 1.5 years!
16:55 owst But, my interested has been piqued, I'll certainly take a look at porting over the rename stuff (at least) to the import code in darcs next week/the following week
16:57 gh_ owst, oh great :)
16:58 gh_ I like that function in darcs-bridge, that attempts to recognize unquoted copy/renames filenames in the fastimport stream by trying every possible combinations when filenames have spaces in them :)
17:02 gh_ ok, looks like file move import is not going into 2.10.3, it mamy go into 2.12 then
17:03 owst Cool - don't let me hold you up if you want to go ahead and port it before 2.10.3
17:03 owst left #darcs
17:03 gh_ s/mamy/may
17:03 gh_ alright owst
17:04 owst joined #darcs
17:04 gh_ hmm, kiwiirc user. he indeed works in the industry.
17:04 owst Heh, step 1: install osx IRC client :-)
17:05 owst (work laptop!)
17:05 gh_ :)
17:09 gh_ owst, I think I'm going to port file move now
17:09 owst Cool, good luck!
17:10 gh_ port or attempt porting
17:22 gh_ owst, did you manage to make darcs-bridge export "faithfully" repositories with conflicts?
17:22 gh_ that is, in a way that the information of the conflicting patch is not lost
17:23 owst Hmm, no I believe not. I think I got stuck when hitting a bug in conflictors (i.e. trying to commute out the conflict into several "branches")
17:26 gh_ was your objective to put the conflicting patches in separate branches?
17:28 owst I think so yes, otherwise you (must?) lose information
17:31 gh_ yes, well that seems the most elegant way
17:32 gh_ okay, the other RL present in InComment was to parse darcs patches inside of git commits.. wow :)
17:32 owst Hmm, what on earth was that for?!
17:32 owst It was something to do with making *something* round-trip-able
17:32 gh_ you did it!
17:33 gh_ the patch is "Recover darcs-patches encoded in commit message"
17:37 owst Hmm. Something to do with replace patches
17:37 owst http://hub.darcs.net/owst/darcs-bridge-export-branch/patch/56a96d3145e108fb93a2c6be30ffd93e79b446e5
17:38 gh_ nice
17:40 gh_ owst, we now have code to detect replaces that we could also use during git import
17:43 owst Oh, cool - that'd be nice. Where is it?
17:49 gh_ http://hub.darcs.net/darcs/darcs-screened/browse/src/Darcs/Repository/Util.hs#218
18:13 Riastradh joined #darcs
18:38 owst Nice - right, I'm off for the weekend. Good to chat gh_ - I should be around next week, see you then!
18:40 Heffalump hi owst, bye owst :-)
18:40 * owst waves goodbye
18:40 gh_ owst, see you soon! I'm going to need you for the patch review :)
18:46 gh_ hi Heffalump
18:48 mizu_no__ joined #darcs
19:16 gh_ I'm having some issues while porting file move handling to current HEAD, will upload my current patch soon
19:27 gh_ I've sent http://bugs.darcs.net/patch1439 , currently WIP because of a couple of compile errors I could not sort out
19:27 gh_ (which I pasted)
19:29 Heffalump could you put up the source of the file that doesn't compile?
19:30 gh_ Heffalump, ok
19:32 gh_ Heffalump, http://hub.darcs.net/gh/darcs-screened_1/browse/src/Darcs/UI/Commands/Convert.hs
19:32 gh_ (I put it on the patch tracker too)
19:39 Heffalump for 797, you'd better give diffCurrent a type signature
19:39 Heffalump to help track it down
19:40 gh_ ouch, the last 2 errors is because I used the die function
19:40 gh_ instead of error for instance
19:40 gh_ Heffalump, ok
19:41 Heffalump but I suspect the fundamental problem is you are trying to do the reverseFL diff +<+ outside the unFreeLeft
19:41 Heffalump the unFreeLeft is (from memory) turning a witness that you can exercise some control over, into one that is opaque (the Sealed)
19:43 gh_ does it have anything to do with the change in RL that ben did?
19:44 Heffalump oh, maybe, you probably just need to swap the arguments to +<+
19:45 gh_ I'm not sure yet, I thought the change was only in the constructors of RL.. let me see
19:46 Heffalump no, I think +<+ must have swapped too, otherwise it'd be really confusing
19:47 gh_ "This means backward lists (RL) are constructed by adding
19:47 gh_ elements to the right (xs :<: x). Concatenating backward lists (xs +<+ ys)
19:47 gh_ means the xs are earlier than the ys, in symmetry with (xs +>+ ys)."
19:47 gh_ ok yes, and it does make sense in the code too
19:48 gh_ (that was the long description of Ben's patch)
19:48 gh_ aand.. it compiles :)
19:50 mizu_no__ joined #darcs
19:59 gh_ but breaks test on deleting empty directories during import.
19:59 gh_ ok, enough for today!
19:59 gh_ thanks Heffalump!
23:00 owst joined #darcs
23:20 owst joined #darcs

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