Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2014-04-15

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

All times shown according to UTC.

Time Nick Message
00:41 dolio joined #darcs
01:19 adnap joined #darcs
01:24 adnap joined #darcs
01:29 adnap joined #darcs
01:34 adnap joined #darcs
03:19 mizu_no_oto joined #darcs
03:41 mizu_no_oto joined #darcs
04:04 c74d joined #darcs
04:10 c74d joined #darcs
04:23 c74d joined #darcs
05:45 lelit joined #darcs
07:01 notdan joined #darcs
07:12 raichoo joined #darcs
07:14 nomeata joined #darcs
07:21 raichoo joined #darcs
08:20 MasseR joined #darcs
09:41 byorgey joined #darcs
12:12 alexei joined #darcs
12:41 edwardk joined #darcs
12:51 mizu_no_oto joined #darcs
13:26 mizu_no_oto joined #darcs
13:35 favonia joined #darcs
14:33 alexsuraci_ joined #darcs
14:33 edwardk joined #darcs
14:34 c74d3 joined #darcs
14:38 IbnFirnas_ joined #darcs
14:43 raichoo joined #darcs
15:13 nomeata joined #darcs
15:55 edwardk joined #darcs
17:07 favonia joined #darcs
17:17 mornfall joined #darcs
17:36 lelit joined #darcs
17:37 dolio joined #darcs
17:47 gh_ joined #darcs
17:54 rdesfo joined #darcs
17:54 gh_ hi
17:54 rdesfo hello
18:16 Heffalump hi
18:31 gh_ I'm looking at repository cloning with packs. I wonder what would be a good heuristic to detect when packs getting + missing patches is less eficient that normal getting
18:33 gh_ counting number of patches of current repo vs packs? The problem is that 1 "heavy" patch can be worse than 50 simple ones..
18:35 gh_ one radical solution: only get with packs when packs are up-to-date
18:37 alegadea joined #darcs
18:50 gh_ except if all new patches are tags...
19:02 Heffalump is pack cloning working properly now?
19:03 Heffalump also, why would using a pack if present ever be less efficient?
19:03 gh_ I just cloned darcs.net, which has a quite old pack, and it works
19:03 Heffalump cool, just wasn't sure if whatever problem that caused us to abandon packs previously had been fixed or not
19:03 gh_ Heffalump, pristine getting, if pack is outdated, requires getting the pack then applying missing patches
19:04 Heffalump or could we calculate which files were out of date and get those?
19:05 gh_ Heffalump, I guess by getting missing patches (which can take some time) and doing a listTouchedFiles on them..
19:05 Heffalump we need to get those patches anyway though
19:05 Heffalump unless we're trying to make a lazy repo, at which point I'm not sure a pack makes sense
19:05 Heffalump though perhaps it does
19:06 gh_ when pack is up to date, it does
19:07 gh_ yeah, updating pristine is doe by getting missing patches and doing "tentativelyMergePatches" on them
19:07 gh_ *done
19:09 markstos joined #darcs
19:09 gh_ so it's more than just getting new pristine files. I wonder why we can't just grab current inventories and pristine files?
19:10 Heffalump well, without listTouchedFiles, how would we know which to grab?
19:11 gh_ by looking at pristine hash of remote hashed_inventories, and getting it
19:11 gh_ redundant files will not be downloaded again since we use the cache
19:13 n-dolio Do packs ever work on hub, by the way? I always see errors about packs not working.
19:13 n-dolio I've tried enabling packs on all my repos.
19:13 gal_bolle joined #darcs
19:13 gh_ and replace local hashed_inventory by remote one. none of this involve downloading patches.
19:13 markstos left #darcs
19:14 sm n-dolio: no they don't IIRC (I don't update them, and they gaves errors or no speedup last I tried them)
19:14 n-dolio Ah, okay.
19:14 sm packs are still WIP
19:16 gh_ WI"P" :-)
19:18 gh_ really I want to get the fixed for 2.10
19:18 gh_ *tem
19:18 gh_ *them
19:23 raichoo joined #darcs
19:53 gh_ I suggest a little change of format:
19:53 gh_ * when creating packs, copy pristine hash to _darcs/packs/pristine
19:53 gh_ * when getting, compare remote _darcs/packs/pristine to pristine hash of _darcs/hashed_inventory
19:53 gh_ * if _darcs/packs/pristine does not exist, or hash is different, get normally, otherwise get with packs (function copyPackedRepository2)
20:01 gh_ if we want to be more clever in the future and still get the patches archive in case of different pristines, we can do it, but I don't feel like changing so much code
20:02 sm the less cleverness the better, perhaps
20:06 Heffalump sounds reasonable
20:06 Heffalump given what sm said about errors, do we know if they work in general or not?
20:09 whaletechno joined #darcs
20:10 whaletechno joined #darcs
20:24 nomeata joined #darcs
20:27 gh_ joined #darcs
20:37 gh_ Heffalump, I don't know. I suspects errors to happen in cases when packs are outdated since there is much more going on in those cases. We need to test.
20:43 rdesfo joined #darcs
20:59 rdesfo left #darcs
23:02 favonia joined #darcs

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