On Mon, Nov 10, 2003 at 02:41:49PM +0200, Eero Pajarre wrote:
> Richard Rauch wrote:
> >Resizing to 0,0 could be undesirable for some programs that may
> >propogate resizes to children windows. (I recall getting errors
> >about 0-sized sub-windows.)
> >Also, detecting 0,0 as a special case for resize strikes me as out of
> >GLUT's character of directness. We have a visibility flag (and for
> >GLUT compatibility we should support it better). That flag is both
> >a natural way to handle iconized/minimized windows, and sufficient
> >so far as I can see. (Though we may have to support resizing them,
> >too, as legacy compatibility for whatever GLUT/freeglut you are using.)
> >IMHO, if we generate 0,0 resize events, we need to support 0-sized
> Ok, I am pretty sure that the 0,0 resize happened also with Glut,
Hm. I thought that I replied to this, but it's marked un-replied-to
Anyway, if MS-WINDOWS version of old GLUT did this, then it is window-
system dependant. (My sneaking suspicion is that MS-WINDOWS itself may
have sent the resize evet, which was just forwarded by GLUT---and by
> (Is there anybody on the list with Windows + Glut (classic) setup,
> who could test this? I don't have Glut installed at the moment)
> I think I actually ended special casing 0,0 size because otherwise
> setting the projection matrix could bomb because of division by zero.
Hm. But it wouldn't be required for ortho projections (e.g., for GUI
systems layered over GLUT/freeglut).
> I am not 100% sure if we need to be compatible with classic Glut in this,
> especially if the behavior is different with Windows / X11.
> (I would need to update my programs for this though)
You would need to for old GLUT or (up through current) freeglut on X11.
That may not be a huge issue for you, though.
I think that we should go through the standard visibility mechanism,
and make that work for both. We can do the resize to 0,0 if need be,
but I'd consider that a legacy compatibility issue at best and would
still rather push visibility as the "standard" way to do this.
If we do go with supporting 0,0 sized windows, we need to support
such sub-window sizes/resizes from applications. (That could be
nasty of the host environment doesn't support that directly.)
I think that on X11, map/unmap notifies will probably take care of
it. I trust that no one has any actual objections to at least adding
that. If I don't hear anything by "sometime soon" (maybe this evening,
maybe in a day or so), I'll go ahead and do that for UNIX_X11.
"I probably don't know what I'm talking about." http://www.olib.org/~rkr/