Fix revert again

[Aidan Kehoe]

 Ar an t-ochtú lá is fiche de mí na Samhain, scríobh Michael Sperber: 

 > Aidan's recent fix (of 2006-10-27) worked, except for the "binary"
 > coding system, if that was determined implicitly.  The reason is the
 > handling of buffer-local variables.  (First against the wall when the
 > revolution comes.)
 >
 > Aidan, could you have a look at this before I commit?

I don’t understand how my initial version eliminated the problem at all,
given that :-(. But then I never did see the issue on my machine. Besides
being a huge pain in the arse when it comes to modelling and optimising
Lisp, are buffer-local variables in general untrustworthy? I saw odd
behaviour in 17346.16021.770786.578415 at parhasard.net , for example, but I
wasn’t aware of more general issues. 

 > Index: lisp/ChangeLog
 > ===================================================================
 > RCS file: /pack/xemacscvs/XEmacs/xemacs/lisp/ChangeLog,v
 > retrieving revision 1.771
 > diff -u -r1.771 ChangeLog
 > 2006-11-28  Mike Sperber  <mike at xemacs.org>
 > 
 > 	* files.el (revert-buffer-internal): Determine the coding system
 > 	while still in the original buffer, rather than the new one.

-- 
Santa Maradona, priez pour moi!