bug#32105: 25.2; calendar-read-date should default to today [PATCH INCLUDED]

[Lars Ingebrigtsen <larsi@gnus.org>]

Boruch Baum <boruch_baum@gmx.com> writes:

> The function is inconsistent in that it uses function `calendar-read'
> for the year and day values, but `completing-read' for the month value.
> Should `calendar-read' behave like `completing-read'? Maybe it should
> have additional optional arguments for everything required by
> `completing-read', and then just call `completing-read'? Or just not use
> `calendar-read' at all, and deprecate it?

Hm...  well, calendar-read seems nice, since it validates the numbers...

> If you `completing-read' -type behavior for entry of the year field, how
> many history entries are you going to give the user? You could use
> `history-length', like so:
>
> (let* ((n-year (calendar-extract-year (calendar-current-date)))
>        (a-year (- n-year (/ history-length 2)))
>        (z-year (+ n-year (/ history-length 2))))
>   (number-sequence a-year z-year))
>
> This has a disadvantage that for certain use-cases future years might
> not make sense.

I think just putting the current year in M-n is fine -- we don't have to
mess with the history at all.

That is, something conceptually like:

(read-from-minibuffer "Year: " nil nil t nil "2019")

Then 2019 is in M-n and can be edited, and just hitting RET will also
return 2019.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no