> > Proposed giiSplitInputs posted to ml, ggiSplitInputs still todo.
Huh - have I missed it ?
> there is something in the API which I find extremely unpleasant.
> cube3d.c line 109 illustrates my point:
> it appears giiJoinInputs is used (with a NULL argument) to return
> the ggi_input_t. That's very misleading.
Yes - it's a bad hack, and I'm responsible for it ...
> I think the least to do is to add am accessor function, something like
> 'giiGetInput(ggi_visual_t)'. That seems trivial to add, but makes the API
> much more meaningful. (Of course, ggiJoinInputs can then use this new
> function, instead of the other way around).
Hmm - actually I do not want to encourage getting the input without also
_detaching_ it from thee visual. I'd rather like to "force" detaching the
input when querying for it. That way people won't be able to do the mistake
ggiJoinInputs now does by using giiGetInput in the wrong way.
Is there a reasonable use for getting the input without also detaching it ?
> I don't quite understand why kgicon is part of this project at all.
> Didn't Steffen set up a project on sourceforge a long time ago just
> for kgicon ?
There are 2 "versions" of KGI ... Steffen's "real" one, and an "ad hoc hack,
works o.k. for now" version I used to take care for. The latter is what is
in the tree.
CU, ANdy
--
= Andreas Beck | Email : <[EMAIL PROTECTED]> =
