"brian m. carlson" <sandals@xxxxxxxxxxxxxxxxxxxx> writes: > On 2020-06-12 at 01:55:56, dwh@xxxxxxxxxxxxxxxxxxx wrote: >> I now think even that proposal is overly complicated. I think the >> easiest solution is to simply standardize the existing pipe-fork >> interface as the way GPG talks to all signing tools. For signing tools >> that have different command line interfaces than GPG, we can create >> adapter scripts. Tools that want to be compatible can adapt. > > This becomes pretty tricky because Git parses OpenPGP headers in a > variety of places (e.g., at the end of tags). If your proposal is to > wrap new formats in a fake OpenPGP format, like some existing tools do, > then that would be viable, but otherwise you're going to require either > Git to know about your signing format specifically (which is not a > sustainable approach) or some sort of configuration framework like has > been previously discussed. > > If you're going to wrap things in a fake OpenPGP format, then you don't > actually need to send any patches to Git at all; you can simply set > gpg.program and continue. True enough ;-) Thanks for a concise summary of the situation.
