Re: [patch] get modification-time from vc

[Nicolas Goaziou <mail@nicolasgoaziou.fr>]

Hello,

Rasmus <rasmus@gmx.us> writes:

> The file system modification time as accessed by {{{modification-time}}}
> is often wrong, e.g. when a project is newly cloned or if the file has
> been copied in a sloppy manner.  At least if modification time suggest the
> time that the file was edited by an author.
>
> Sometimes vc can provide a better answer.  The attached patch tries to get
> a sensible time from vc.

Thank you for the patch.

> WDYT?

Well, if you ask, I do not like much when some software tries to
outsmart me (note that I like it even less when it actually succeeds).
So, I think {{{modification-time}}} should expand to modification time.

If you think that is useful, then I vote for a different macro.

Usual nitpicking follows.

> +					       (when (search-forward-regexp "Date:?[ \t]*" nil t)

Any reason not to use the genuine `re-search-forward'?

Is the search case sensitive? I suggest to bind appropriately
`case-fold-search' in any case.

> +						 (let ((time (parse-time-string
> +							      (buffer-substring
> +							       (point) (point-at-eol)))))

Any reason not to use the genuine `line-end-position'?

> +						   (when (org-some 'identity time)

  (cl-some #'identity time)

> +						     (setq date (apply 'encode-time time))))))))

  (setq date (apply #'encode-time time))

> +					  (let ((proc (get-buffer-process buf)))
> +					    (while (and proc (accept-process-output proc .5 nil t))))
> +					  (kill-buffer " *org-vc*")

  (kill-buffer buf)

However, I suggest to use `unwind-protect' for this and kill-buffer
during as an unwind form.

Also, I suggest to factor this code out of
`org-macro-initialize-templates'.


Regards,

-- 
Nicolas Goaziou