Camelia, the Perl 6 bug

IRC log for #darcs-theory, 2009-10-18

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

All times shown according to UTC.

Time Nick Message
10:42 iago joined #darcs-theory
11:43 tux_rocker joined #darcs-theory
13:29 tux_rocker i don't understand why in Judah Jacobson's paper, he assumes that non-sensible patches have a zero effect
13:30 tux_rocker the effect of a hunk to a non-existent file seems undefined to me
13:30 tux_rocker if it's undefined, it's not a 0
13:30 tux_rocker but IANAM (i am not a mathematician)
15:48 mornfall tux_rocker: It makes perfect sense to do that -- a conflict is a non-sensible sequence, eg.
15:49 mornfall Also, adding undefineds to a theory usually just confuses things, and 0 effect works just as well in this case.
15:50 mornfall Or so I guess, anyway.
15:50 mornfall I should really just read the paper again.
15:51 tux_rocker i only read up to the beginning of part 3
15:52 tux_rocker but anyway, for 0 it should hold that x0 = 0x = 0
15:52 tux_rocker but doesn't that imply that a conflicted repository should have an entirely empty pristine?
15:53 tux_rocker if the repo is a sequence of patches xyz0, you have z0 = 0, y0 = 0, x0 =0, so the entire sequence is equal to 0
15:53 tux_rocker or, well, perhaps he answers this in the section on conflicts
15:54 mornfall Yes, that's probably true.
15:56 mornfall I think that he just doesn't admit a "conflict" state -- like darcs does not. You need mergers/conflictors to allow such repositories.
15:56 mornfall (Which form sensible sequences.)
15:57 mornfall My comment about conflicts and 0 was probably too hasted.
15:57 tux_rocker hmm, darcs allows conflicted states, doesn't it?
15:57 mornfall But he does seem to use the 0 in some of the proofs.
15:58 tux_rocker but the effect of a patch sequence with conflicts does not include the effects of the conflicting patches
15:58 tux_rocker that's how I understood it
15:58 mornfall tux_rocker: Well, for darcs "conflicted" really means "there's a conflictor on which nothing depends".
15:58 mornfall tux_rocker: Which is different from "this sequence is not sensible".
15:59 tux_rocker ah, right
16:00 tux_rocker so darcs creates the conflictor when you pull conflicting patches, not when you resolve the conflict?
16:00 tux_rocker (at least theoretically)
16:00 mornfall Yes.
16:00 mornfall You can't even represent the repository otherwise, under linear contexts.
16:00 mornfall (I.e. you really would get something akin to completely empty repo, since it cannot be represented at all...)
16:01 mornfall The resolution patch is usually a normal hunk, no mention of conflictors in that.
16:01 mornfall It only resolves the conflict by virtue of depending on that conflictor that represents the original conflict.
16:02 mornfall Which may be a little dubious in itself, but seems to work most of the time.
16:04 tux_rocker ok, it makes sense to me
18:08 arjanb joined #darcs-theory

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