Hi all,

I've found a problem/challenge with the conflictor picture, and think I
have a solution, but am unsure whether this might run into trouble.  The
problem (which we've seen before hypothetically) is when we've got four
dependent patches ABCD, and we try to commute them into the order
B'D''A''C'.  The current code barfs on this, because there's no
representation of D'' or A'' in terms of a set of conflicting
patches--because this is an artificial sitution that can't happen as a
result of merges.

So my thought is that this shouldn't be allowed.  We can commute ABCD to
B'A'D'C', but the commutation of A'D' will fail.  This breaks the old idea
of conflictors as "everything commutes", but should keep the actually
desired behavior (which is that every *merge* succeeds).

Thoughts, questions, etc welcome.  Mostly this email is for Ganesh, who's
been working with me on this, but I thought others might enjoy reading
along or contributing.
-- 
David Roundy
Department of Physics
Oregon State University
_______________________________________________
darcs-devel mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-devel

Reply via email to