On Thu, Apr 04, 2019 at 08:27:22AM -0400, Jeff King wrote: > IMHO the difference between those cases is that it is very easy to type > something natural to get a tag of tag, like: > > git tag -m foo mytag > # oops, try again > git tag -m bar -f mytag mytag Could it be that the semantics of `-f` are unintuitive and the real culprit? The documentation talks about replacing the tag, which at least naively to me has the sense of "well, the old tag object gets garbage collected sometimes". But if the tag is still referenced, the old tag hangs around. And this gets quite confusing. For example: git tag -m 'Message' -a myTag git tag -m 'Message' -a myOtherTag myTag git tag -m 'Message' -f -a myTag otherBranch Now a `git show myOtherTag` gives me the old myTag - and labels it as myTag. Copy and pasting this directs me to somewhere completely else, though. Wouldn't it be the more sane approach here to fail, when the old tag is (or as in your example will) still be referenced? Or make some effort with git show that it is clear that the referenced tag can no longer be referenced with its original name? Greetings, Eckhard
