Help-char bogosoties [was Re: delete-key-deletes-forward]
Jan Vroonhof <vroonhof(a)math.ethz.ch> writes:
[Glyn's fix to (remprop 'backspace ascii-char)]
It does seem to work. However as you suggest it may not be the right
fix. Maybe event loop key look-up is slightly bogus when in the
minibuffer.
Looking for other fixes I found that the problem lies in
event_matches_key_specifier_p
which is used to check whether help-char is bound. (To see the problem
do a (setq help-form "ouch!") and then you git it even without
read-minibuffer.
If the key specifier to be matched is a char then the event is forced
to char before comparison (using the 'ascii-char property if need be).
However if the key specifier is a normal specifier then nothing is
done. So we can also fix the problem by setting
(setq help-char '(control h))
However for some reason this does not work on TTY's (Backspace is till
considered help-char with the ascii-character property set). The in
code event_matches_key_specifier_p seems indeed to handle things
differently for ttys.
Jan
P.S. Note with (remprop 'backspace 'ascii-char) 'C-q Backspace' no
longer insets a ^H, I am not sure that is a loss.