Re: master 1d91d9b717d: project-vc-dir: Use truenames

On Fri, Jul 24, 2026 at 1:33 AM João Távora <[email protected]> wrote:

> On Fri, Jul 24, 2026, 03:51 Dmitry Gutov <[email protected]> wrote:
>
>> Hi,
>>
>> On 23/07/2026 19:42, João Távora wrote:
>> >
>> > under which it does. Maybe the fact that it fooled two programmers
>> > already justifies a comment?
>>
>> Anyway, how about this? Does it feel a bit repeating itself? I would
>> rather remove the last sentence but as the author I'm probably not the
>> best judge for that.
>>
>>
>> -    ;;
>> -                                       nil))
>> +    (let ((dir (read-directory-name "VC status for directory: "
>> +                                    (vc-root-dir) nil t
>> +                                    nil))
>> +          truename)
>> +      ;; Try to match the result of `vc-refresh-state' in a file buffer.
>> +      ;; Otherwise if you do C-x v d -> C-x C-f -> C-x v d you may get a
>> +      ;; new *vc-dir* buffer, different from the original.
>> +      ;; We use the truename if the DIR has no VC backend detected, but
>> +      ;; its truename differs, and has a VC backend.
>
>
> If I follow correctly,  this is the grammatically simpler:
>
> "If DIR has no VC backend but its truename does, use that instead of DIR."
>
> Right? If so, then I think it's an improvement. Except I don't understand
> when it can ever be true, but that's my fault for not understanding the
> precise semantics of "having a VC backend", and explaining those probably
> has no place in this comment.
>

Not following this thread 100% having been on the road for a while, so
apologies if I'm missing something key...

I don't understand how a truename can have a back end but a symlink to it
cannot even if the truename is a directory.  It is transparently the same
under almost all contexts (under Unix-like systems).