On Tue, 22 Mar 2005 17:10:28 +0100, Peter Conrad <[EMAIL PROTECTED]> wrote: > But there *is* a conflict. Maybe you *want* the line to be added twice. > How should tla know?
I believe that's the real problem with the locality of relative diffs tla uses. When I let tla go first, I was stunned that it just did that -- added the line twice! It flew in the face of the intuitive expectation that, if the most recently committed remote file and the modified local file conflict only in the same diff location which turn out to be modified _identically_, they are, well (duh!) identical, and _therefore_ there's no conflict! I think it'a a drastic demonstration why diff is not the best thing to use. E.g., darcs does detect the conflict but in such a way as to show it's still the same line. It doesn't just go add the line twice -- why, it was not added twice into the original commit! The result of the propagation of the change, a file with the two identical lines, is a heretofore unseen beast, and tla does that silently to you. Hmm... Cheers, Alexy _______________________________________________ Gnu-arch-users mailing list Gnu-arch-users@gnu.org http://lists.gnu.org/mailman/listinfo/gnu-arch-users GNU arch home page: http://savannah.gnu.org/projects/gnu-arch/
