Junio C Hamano <gitster@xxxxxxxxx> writes: > How about doing it this way? We know and trust that existing > revision traversal machinery is doing the right thing, and it is > only that the clear_commit_marks() calls are botched. Another alternative may be to allow any object to clear_commit_marks() and have the callee dereference as needed. After all, the revision walking machinery does such a dereferencing when leaving these marks that the function wants to clear, so it might make sense from that point of view. A quick "git grep clear_commit_marks()" tells me that most of the codepaths do make sure the object is a commit when they cast their first argument to (struct commit *) when calling this function, but some of them do look suspicous. -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html