Re: master 1d91d9b717d: project-vc-dir: Use truenames
Dmitry Gutov <[email protected]>
| Newsgroups | gmane.emacs.devel |
|---|---|
| Message-ID | <[email protected]> |
Hi Joao, On 15/07/2026 03:09, João Távora wrote: >> Could you or him point to a specific scenario/step which doesn't match >> their preferences when my patch is applied? > > I've just tried your patch. I didn't give it much testing, but it seems > to work exactly like I think it should. I.e. C-x v d RET when > visiting any file in a symlinked repo seems to behave exactly like > C-x p v. I guess that's because dir defaults to 'vc-root-dir's value, > which is also good. Excellent! > I thought your earlier reply about vc-responsible-backend meant > that if I gave it a subdir it would revert to following the link, which > would again seem like a mistake to me. But I see that even if > I give it a subdir, vc-responsible-backend still returns 'Git. Right: this is about what the symlinked directory points to, not about which directory we pass to vc-dir. Maybe there are other practical examples of when vc-refresh-state follows the link, but this is the main one I've managed to reproduce: repository's subdirectory symlinked externally. Actually in this scenario vc-refresh-state does not prompt even when this variable's value is 'ask' - because file-chase-links does not account for symlink parent directories - it just follows the links with a message. For compatibility's sake vc-dir should not prompt either (unless we change the condition inside vc-refresh-state). > So everything's fine from my side. I'll still use C-x p v since I have > muscle memory for that, but at least one of my new colleagues > who's somehow still using Emacs after 2 months (the poor soul, > now even wants a proper keyboard) can start using vc confidently. Our prayers are with him/her. Sean, this should go to the master branch, right?