On Tue, Sep 19, 2017 at 08:52:24AM +0900, Junio C Hamano wrote: > I think you can use skip_prefix() to avoid counting the length of > the prefix yourself, i.e. Thanks, will use it. > The hardcoded +10 for "is_tag" case assumes that anything other than > "refs/tags/something" would ever be used to call into this function > when is_tag is true, and that may well be true in the current code > and have been so ever since the original code was written, but it > still smells like an invitation for future bugs. is_tag is used later. I'll chance it so that it does not rely on it to match, but it still has to produce it. > Was there a reason why A and c are in different cases? Are we > worried about case insensitive filesystems or something? The tags have been there of different case already. I don't know why. I'll change the branch names but I'm reluctant to touch existing tests. -- Max
