On Mon, Oct 22, 2001 at 10:36:15AM +0200, Lars Gullik Bjønnes wrote:
> | For completion purposes the space could simply be ignored.
>
> the process of supplying what to complete on seems to be a bit more
> complicated...
Why?
> | Splitting the names would make the actual interface much slimmer, and
> | extending a "module" (read "mathed" or some other "complex" inset like
> | tabular) does not require changes to the "core" (and I tend to consider the
> | current need to do so a fundamental architectural flaw of LFUNS)
>
> ortogonal.
Could you expand this to a full sentence? Of course, putting in some
"autoregistration" feature for LFUNS would be another solution. Do you mean
something like that with "orthogonal"?
> | Note that e.g. tabular already does exactly this (even if not officially
> | called so) with its 'tabular-feature' LFUN.
>
> then perhaps badly so.
Well, perhaps. But unless "module changes" require "core changes" it is
more than a valid work-around IMNSHO. On the other hand, code like
switch (action) {
case LFUN_WRITE_A:
write('A');
break;
case LFUN_WRITE_B:
write('B');
break;
/*...*/
case LFUN_WRITE_Z:
write('Z');
break;
}
does not look nice, either. I think there are quite a few things that
possess some kind of "natural hierarchy", and things like
tabular->feature->row->add or math->accent->circumflex are two of them.
Andre'
--
André Pönitz .............................................. [EMAIL PROTECTED]
