I'd recommend timing 'git diff-index -z HEAD -- <file>' vs 'git status
--short <file>' vs 'git status -z --porcelain <file>' in the console on
your real repository and posting the numbers to the bug 8288.
And maybe contributing an example patch that works for you. The function
responsible for this is `vc-git-state'.
Both `git diff-index` and `git status` were comparably slow at least to my subjective perception. After archiving old branches and a garbage clean -- `git tag`, `git branch -d`, `git gc` -- both perform fast enough again. For the time being, I think I retract my earlier belief that I was seeing the same thing as described in bug 8288.
Thanks for your help.