Stephen J. Turnbull writes: CW> auto-detect (if the info files are correct) and auto-detect CW> when saving the buffer means XEmacs' way which is the Unix CW> file ending. Agreed. OK, I wait with my patch until this has been fixed. CW> Thus, I fail to see where "toggling" from nil to crlf is bad. Well, I also use XEmacs on Unix. The eol type "toggling" there: a. w/o my change: nil -> nil -> ... b. w/ my change: nil -> crlf -> cr -> crlf -> ... Still fail to see why b is worse than a. Toggling from nil to cr would also be OK with me... (if nil doesn't mean the Unix way ;-)) CW> Talking of patches, xemacs-patches is again broken. I got no CW> robot reply for a couple of patches I can understand that. But given the fact, that XEmacs' former mailbot (2yrs ago) made senders of mails to xemacs-patches work for /dev/null, it is probably better to send a mailbot reply. I now know that it is not again broken... CW> * "Toggling" Unix -> Dos -> Mac -> ... doesn't seem to be CW> what people really need. No, combined with my cryptic "OK, there could be a variable", my above sentence meant that button2 should IMHO toggle between 2 eol types (and these 2 are controlled by the variable) and not all 3 (OK, if we have that variable, there could also be an option for it) as probably not too many people work in an environment where all three systems are used. OK, for all 3 we can have a menu... - Christoph