Edwin Leuven <[EMAIL PROTECTED]> writes: > i read that, and i find this "wildcard" notion very strange. perhaps i > am overseeing something, but why don't we have an interface where > > \bind "" "buffer-close" > simply means "there is no key binding associated with buffer-close" etc?
I really prefer \bind "C-w" "" meaning ``do not bind C-w to (presumably) buffer-close''. > the fact that the gui becomes very complicated also suggest that > something is wrong since this stuff shouldn't be complicated. Here s my view on how it could work (in horrible pseudo-code): 1/ the reference binding set (e.g. cua) is loaded and saved. toplevelKeyMap().load(bind_file); KeyMap refkeymap = topLevelKeyMap(); 2/ then the user bindings are loaded on top toplevelKeyMap().load(user_file); 3/ there is only one possible operation possible in the user bind file: override a binding. When overriding a binding with empty string (AKA LFUN_NOACTION), the binding remains in the table, but it is equivalent to removing it. 4/ since the two tables are in sync (each entry of refkeymap correspond to an entry in user_keymap at the _same_ location, because their is no deletion and insertion is done at the end of the table), it is possible quickly to make a diff between the two tables, either for the purpose of outputting user.bind or for display purpose in the UI (just put the entry in bold). Binding a function to something which is already a prefix can be forbidden for now, as it has always been. Bo, what is the part that would not work in this scheme? JMarc
