Camelia, the Perl 6 bug

IRC log for #darcs, 2011-05-02

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

All times shown according to UTC.

Time Nick Message
00:18 kowey left #darcs
00:38 navaati left #darcs
00:51 gal_bolle left #darcs
02:08 lispy joined #darcs
03:01 gbeshers left #darcs
03:01 owst left #darcs
04:33 secorp left #darcs
04:46 secorp joined #darcs
04:57 secorp left #darcs
05:22 secorp joined #darcs
05:27 bsrk2 left #darcs
05:27 bsrk2 joined #darcs
06:10 jeltsch joined #darcs
06:14 Jaak joined #darcs
06:17 jcpetruzza joined #darcs
06:19 jcpetruzza left #darcs
07:25 jderque joined #darcs
07:38 balor__ joined #darcs
07:44 Weltraumschaf joined #darcs
07:52 balor__ left #darcs
08:14 bsrk2 left #darcs
08:19 waern__ left #darcs
08:21 waern joined #darcs
08:35 jderque left #darcs
08:38 bsrk2 joined #darcs
08:47 waern left #darcs
08:47 waern joined #darcs
08:51 shenshei joined #darcs
09:03 jderque joined #darcs
09:31 secorp left #darcs
09:34 secorp joined #darcs
09:35 shenshei left #darcs
09:40 shenshei joined #darcs
09:48 gbeshers joined #darcs
09:55 etarasov joined #darcs
09:57 gbeshers left #darcs
10:03 lispy left #darcs
10:04 bsrk2 left #darcs
10:33 secorp left #darcs
10:34 secorp joined #darcs
10:41 shenshei left #darcs
10:43 shenshei joined #darcs
10:49 secorp left #darcs
10:51 secorp joined #darcs
11:02 owst joined #darcs
12:42 jeltsch left #darcs
12:43 jeltsch joined #darcs
13:14 lelit left #darcs
13:14 lelit joined #darcs
13:26 intripoon_ joined #darcs
13:28 jderque left #darcs
13:29 intripoon left #darcs
14:07 copumpkin left #darcs
14:33 gwern left #darcs
14:50 gwern joined #darcs
14:50 gwern left #darcs
14:50 gwern joined #darcs
14:58 secorp left #darcs
15:03 Weltraumschaf left #darcs
15:13 shenshei left #darcs
15:30 raichoo joined #darcs
15:49 jeltsch left #darcs
15:50 shenshei joined #darcs
16:02 jderque joined #darcs
16:20 Heffalump left #darcs
16:47 raichoo left #darcs
16:49 raichoo joined #darcs
17:01 Heffalump joined #darcs
17:11 dcolish left #darcs
17:11 preflex left #darcs
17:11 Igloo left #darcs
17:11 idnar left #darcs
17:12 dcolish joined #darcs
17:12 preflex joined #darcs
17:12 Igloo joined #darcs
17:12 idnar joined #darcs
17:17 shenshei left #darcs
17:19 shenshei joined #darcs
17:27 balor__ joined #darcs
17:36 balor__ left #darcs
17:42 owst left #darcs
17:55 shenshei left #darcs
17:57 lispy joined #darcs
18:16 balor__ joined #darcs
18:27 owst joined #darcs
18:42 secorp joined #darcs
18:43 balor__ left #darcs
18:43 balor__ joined #darcs
18:49 lispy left #darcs
18:52 balor__ left #darcs
18:52 balor__ joined #darcs
18:55 Heffalump mornfall: do you think darcs init should generate a fresh uuid for the root, or not? I think there are some interesting trade-offs here.
18:57 shenshei joined #darcs
18:59 owst What are the pros/cons?
19:00 Heffalump if you generate a fresh uuid, then pulling between repos that were init'd separately becomes tricky
19:01 Heffalump and in particular darcs init ; darcs pull -a is no longer the same as darcs get
19:02 Heffalump if you don't, then rename the root operations don't work too well
19:02 owst Right. Why would you want to use a fresh uuid then?
19:02 Heffalump because of wanting to rename the root
19:03 owst Oh, sorry I didn't see that
19:03 owst :)
19:03 Heffalump suppose we want to merge project X and project Y
19:03 Heffalump a natural starting point would be pulling them together, putting the files from X in a directory X, and Y in a directory Y
19:23 owst Is this right: inventories are a chained list of patches (for each tag?) with a list of the patches added since the previous inventory and their corresponding hashes (giving the filename of the patch)?
19:25 Heffalump yes
19:25 Heffalump I'm not quite sure when they are broken up - clean tags are a common time to do it
19:25 owst Right, cool.
19:26 owst And then hashed_inventory gives the hash of the current pristine, and a 'base' inventory, and a list of patches on top of that?
19:28 Heffalump yes
19:30 owst Of course, I did the stupid thing and didn't read the wiki first.
19:43 owst The index file just saves you from having to recursively read all the pristine files?
19:47 Heffalump yes
19:48 owst Ok, I think I understand what everything is for now :). So, when you say uuid for the root, do you mean the pristine hash given in hashed_inventory?
19:55 Heffalump no. I mean in the darcs 3 design work.
19:56 owst aha, right!
20:02 secorp left #darcs
20:06 balor__ left #darcs
20:06 JaffaCake joined #darcs
20:07 secorp joined #darcs
20:09 JaffaCake1 left #darcs
20:12 balor__ joined #darcs
20:23 balor__ left #darcs
20:44 mornfall Heffalump: Well, you don't have to generate the root at darcs init time.
20:50 Heffalump lazily on first record?
20:52 mornfall Presumably.
20:52 mornfall It's kind of a corner-case, granted. But probably reasonable?
20:54 mornfall About inventories, it is safe to assume that they are never guaranteed to be broken up, but if they are, it is always on a clean tag.
20:54 Heffalump I think it would just move the corner-case along a bit - now if you record a new patch then pull you get a problem
20:57 mornfall Well, yes. Hard to tell.
20:58 mornfall I still think it's more elegant that way, overall.
20:58 Heffalump anyway, I was wondering about allowing multiple UUIDs to be mapped to the same directory name
20:58 Heffalump (including root)
20:58 mornfall And merge them?
20:58 Heffalump yes
20:58 mornfall You could, I suppose.
20:58 Heffalump it then gives you a problem about which UUID to use to record new changes.
20:59 mornfall You can also record a merge patch when that happens (semi-automatically, I guess).
20:59 mornfall I mean, when you pull two roots into a single repo, ask whether to merge them.
20:59 mornfall Or something.
21:00 mornfall The main question is probably, how often this will come up in practice.
21:00 mornfall And the solution will depend on that.
21:00 mornfall The possibility to cleanly merge repositories is tempting.
21:00 mornfall Unrelated, even.
21:01 mornfall And hardcoded IDs will likely compromise that, to some extent.
21:01 Heffalump I don't like focusing too much on "how often this will come up in practice", though of course it's important to consider. But darcs is supposed to be elegant, too.
21:01 mornfall Of course, but sometimes you have to cut corners, I guess.
21:02 mornfall And conceptually, I think hardcoding certain object IDs is the less elegant option.
21:02 Heffalump I still think it's the right thing for prefs :-)
21:02 mornfall Sure, if we can come up with a fresh-uuid solution that also addresses the init issue, that's best.
21:02 owst Sorry to jump in, is this stuff (design ideas) written anywhere? I'd like to start reading something to catch up with the ideas you're discussing :)
21:02 mornfall http://wiki.darcs.net/Development/PrimsV3 is about as much as you get
21:03 mornfall And the backlog from two (?) days ago.
21:03 mornfall From here.
21:03 Heffalump and Ideas/AddAddConflicts
21:03 owst Cool, thanks, I'll have a peek at some point soon.
21:03 Heffalump we should keep links to the relevant logs on the wikipage
21:27 iago joined #darcs
21:37 jderque left #darcs
22:10 raichoo left #darcs
22:33 shenshei left #darcs

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