Suppose I fork someone's git repo and make commits A, B, C, and D. The persom who I forked from then cherry-picks A and C, which therefore become A' and C'. He also makes commits X, Y, and Z of his own. So after all this, my branch has A B C D, and his has A' C' X Y Z. Assume that both branches are published, so rebase is not an attractive option. Also assume that X Y Z don't conflict with any of A B C D. How do I merge these two branches in a useful way? Should I simply merge and then manually resolve everything? Is there anything I can do about duplicate commit messages in the log of the merged head? Are these two branches doomed to sync only by cherry-picking each other's commits from now on?
A straight merge should work. Git should notice the identical change sets and will not attempt to merge A/A' or C/C'. If X, Y, or Z touched the same code that A'/C' changed, you may have to resolve conflicts there, especially if B or D touched the same code.
You can always git merge --no-commit and inspect the result to see if it's what you expected.