Re: x-backspace-delete-keys-p without XKB support

[Stefan Monnier]

>> > Given that the condition that x-backspace-delete-keys-p is quite odd,
>> > and probably unlikely to happen, it is better in the case that we
>> > don't have enough information (because XKB is not supported at
>> > either run time or compile time) to return t. The result is that
>> > [backspace] would delete backwards and [delete] forwards. 
>> 
>> > May I install this patch:
>> 
>> I think a better patch would have the function return `dontknow' when it
>> doesn't know.

> I was thinking about that too, but is there any use for `dontknow'? 
> There's exactly one user of this function, in simple.el, and it only
> cares about x-backspace-delete-keys-p returning false... 
> Wouldn't returning `dontknow' just add complexity for no real gain?

Well, then return `lambda'.  The only user will handle it as if it were
t and the code will be just as simple (just use Qlambda rather than Qt in
those places where we can't figure out what to return).

I.e. no extra complexity.  No concrete gain, other than the fact that the
default behavior in the absence of info can be cleanly chosen in elisp
rather than in C.


        Stefan