Peter Conrad <[EMAIL PROTECTED]> wrote: >> In case you don't follow all that mumbo-jumbo with multiple SCMs, >> here's a simple, purely tla case. Create a local repo and mirror it >> in a remote repo. Create a text file in one repo and sync it into >> another. Now do this: add a line to the file in the local repo and >> commit. Then go to the remote, ADD THE SAME LINE MANUALLY BEFORE >> SYNCING, so that the files are now in fact identical in source but not >> as recorded in the auxiliary tla directories. Now tla update -- >> you'll get file.orig <=> file, and file.rej showing +<line> in diff. >> If tla were to check that the result of aplying file.rej to the last >> commit would indeed yield the file <=> file.orig, perhaps it might see >> there's no conflict and not bother with it? > > But there *is* a conflict. Maybe you *want* the line to be added twice. > How should tla know?
It could be possible with a three-way merge and a smarter diff3. For example, BK does not report a conflict if the changes relative to the common ancestor are identical. This could also solve the problem with star-merge and cherry-picking (just use the last consecutive common ancestor). -- Catalin _______________________________________________ Gnu-arch-users mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gnu-arch-users GNU arch home page: http://savannah.gnu.org/projects/gnu-arch/
