Re: format-spec.el: feature wish

[Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann)]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:
>
>> Right now, it appears that the SPECIFICATION argument of FORMAT-SPEC
>> should be an alist where the keys are characters and the values are
>> strings.
>>
>> WIBNI the values could be something else, too, which would then be
>> evaluated?
>>
>> Since anything but strings as values aren't useful at the moment, this
>> would be upward compatible.
>
> How would you use this?  I'm having difficulty envisioning what you
> could express with this that you can't do (easier) with the specs
> themselves...

(defvar silly-format-spec '((?t . (current-time-string))
                            (?l . (user-login-name)))
  "foo")

(defvar silly-format "%l on %t"
  "bar")

(defun silly-insert (msg)
  (insert (format-spec silly-format silly-format-spec)
          " " msg))

Well.  Hm.  Maybe this is indeed very silly, but I hope you get the
point.  With the current mechanism, I have to construct the format
spec on the fly, ie, on every invocation of silly-insert.  But with
my new proposed mechanism, silly-format-spec could be a variable.

kai
-- 
Simplification good!  Oversimplification bad!  (Larry Wall)