Re: emacs-19.el disturbs gnus

sandy@ibm550.sissa.it
Wed, 23 Mar 1994 10:08:59 +0100


Joerg-Martin,

On Tue, 22 Mar 1994 you wrote:

 >   emacs-19.el defines a function current-time.
 >  
 >   gnuspost.el, in function gnus-inews-valid-date, tests for
 >   the existence of current-time and then assumes that the
 >   function current-time-string can take an optional argument:

 > ;; Author: Masanobu UMEDA <umerin@flab.flab.fujitsu.junet>
 > ;; Version: !Header: /home/fsf/rms/e19/lisp/RCS/gnuspost.el,v 1.16 1993/11/22 06:44:12 rms Exp !
 > (defun gnus-inews-valid-date (&optional time zone)
 >   "A date string that represents TIME and conforms to the Usenet standard.
 > TIME is optional and defaults to the current time.
 > Some older versions of Emacs always act as if TIME is nil.
 > The optional argument ZONE specifies the local time zone (default GMT)."
 >   (timezone-make-date-arpa-standard
 >    (if (and (fboundp 'current-time) (not featurep 'emacs-19))
 >        (current-time-string time)
 >      (current-time-string))
 >    zone "GMT"))

Looks to me like you are trying to run the V19 gnus in emacs 18.
emacs-19.el shouldn't be loading otherwise.  If gnus is going to try
to be V18 backwards compatible, it should do a better job of grokking
the emacs version; something like

(string-equal (substring emacs-version 0 2) "19")

Rather than a doubtful heuristic like (fboundp 'current-time).

By the way, above you should have (not (featurep 'emacs-19))

I can't change the name of the function current-time in emacs-19,
because it needs to have that name for backwards compatibility.  Nor
do I want to add the extra arg to current-time-string, because if I go
down that road, I'll soon be rewriting emacs 19 in emacs-19.el.  I
would say that the fix should be made in gnus.

--sandy