# IRC log for #darcs-theory, 2008-12-03

All times shown according to UTC.

Time Nick Message
09:15 kowey joined #darcs-theory
09:25 me-so-stupid joined #darcs-theory
09:46 kowey joined #darcs-theory
14:26 Igloo_ joined #darcs-theory
16:10 arjanb joined #darcs-theory
16:55 semka joined #darcs-theory
18:02 kowey joined #darcs-theory
19:59 ordnungswidrig joined #darcs-theory
20:00 ordnungswidrig /whois #darcs-theory
20:00 ordnungswidrig left #darcs-theory
20:35 kowey joined #darcs-theory
21:42 Heffalump btw, I've had another go at working out the weakest hunk-hunk commute conditions and although I'm not certain, I have a reasonable bet that it's (there is a gap between the patches || one of the patches both adds and removes lines || both of the patches add lines || both of the patches remove lines)
21:44 Igloo "Both add lines" is what I disproved last time, isn't it?
21:45 Igloo That might be right if you don't want to use commute for merging patches
21:46 Heffalump that's for commute, not for merge
21:46 Heffalump your counterexample was where both patches being merged added lines
21:46 Heffalump (I think, I don't have it to hand)
21:47 Heffalump my last two conditions mean that you can
21:47 Heffalump sorry
21:47 Heffalump my last two disjuncts mean that you can merge a deletion and an addition that touch
21:49 Igloo That might work, but it would certainly need different axioms
21:56 Heffalump why?
21:58 Heffalump AFAICT it's unambiguous to do that.
21:58 Igloo It fails  pq <-> q'p'   =>   q'^ p <-> p' q^
21:59 Igloo when p and q both only add lines
22:00 Heffalump ah, grmph, ta
22:14 Heffalump I guess the condition for touching patches is that at most one of the 4 chunks is empty, then. But that's ugly :-(