Re: BUG: ngnus-0.10 can't send messages

[Uwe Brauer <oub@mat.ucm.es>]

>>>>> "Katsumi" == Katsumi Yamaoka <yamaoka@jpl.org> writes:

>>>>> Uwe Brauer wrote:
   >> the question still puzzles me, why does everything work with
   >> ngnus-0.6? 


   > Do you know where the regexp "\\<Mc\\([^aeiou]\\)" comes from?
   > Isn't it in your XEmacs init file?  GREP found no file containing
   > "aeiou" in my system.

Still I don't understand why it works in 0.6.
But yes there are files with aeiou on my system

To start with my init file, which has

(defadvice mail-extract-address-components (after capitalize first activate)
   "Capitalize names extracted from first.last@..., first_last@..., and
\"first last\" <...> addresses."
   (let ((case-fold-search nil))
     (if (or (string-match "\\`[a-z---]+\\([._][a-z---]+\\)+@"
			  (ad-get-arg 0)) ; ADDRESS
	    (string-match "\\`\"[a-z---]+\\( [a-z---]+\\)+\" <"
			  (ad-get-arg 0))) ; ADDRESS
	(setcar ad-return-value
		(capitalize (car ad-return-value))))))

(defadvice mail-extract-address-components (after capitalize-mc last activate)
   "Fix capitalization of \"Mc\" names, e.g. \"Mcdonald\" -> \"McDonald\"."
   (let ((full-name (car ad-return-value))
	(case-fold-search nil)
	(start 0))
     (while (and full-name (string-match "\\<Mc\\([^aeiou]\\)" full-name start))
       (aset full-name (match-beginning 1)
	    (upcase (aref full-name (match-beginning 1))))
       (setq start (match-end 0)))))

It is also in 
site-lisp/packages/ngnus-0.10/lisp/gnus-art.el
(defun gnus-button-mid-or-mail-heuristic (mid-or-mail)

In reftex
(defcustom reftex-abbrev-parameters '(4 2 "^aeiou" "aeiou")


The most likely culprit seem to me the defadvice things. I will
comment them out and see what happens, (still insisting that in 0.6 it
works): ok now gnus-0.10 works. The question is, how could this
defadvice function work again under 0.10???

   > In addition, I found in and around `message-bogus-recipient-p'
   > and `mail-extract-address-components' no `string-match' form in
   > which the third argument `0' (or equivalent) is specified.  So,
   > I guess the backtrace might have reported the one caused by the
   > one other than `mail-extract-address-components'.  How about re-
   > reproducing the problem after loading some doubtful el (not elc)
   > files?  Those will be the ones that the following program reports
   > (type C-j at the end of the last line in the *scratch* buffer):
ok I will try that after having tried, that for should me this tricks.



Uwe Brauer