From: Andreas Beck <becka@rz.uni-duesseldorf.de>
  To  : ggi-develop@eskimo.com
  Date: Tue, 9 Nov 1999 23:09:50 +0100

Re: Problems over matrox fb

Hi !

> What I wrote in there was that after looking at the Matrox docs
> I found out that the drawing registers indeed take signed
> coordinates, so all we have to do is properly cast the coordinates
> to the right size. 

Ah - perfect. 

> I've already commited this change to the Matrox sublib, so please test 
> if it works now.

Great.

> After some thinking I'd believe that most other cards does this too,

Yes. I know at least one example. 

> and therefore there's no longer any reason to hesitate about supporting
> negative coordinates. I would hovever like to propose that we document
> coordinates as having a valid range of -32768 - 32767. 

Yes. This sounds reasonable. If you have way-off coordinates, clipping gets
very tricky even in software, and there will rarely be cases, where stuff
that is that far off really matters.

We should try to handle other cases as gracefully as possible, though.
The only real issue is for drawline anyway, other cases are trivially
handled by standard clipping code.

> This is not even a limitation, as int may very well be 16 bits accordning to C
> standards.

According to ANSI-C, only -32767 - 32767 (mind the symmetric range !) is
guaranteed for int.

CU, Andy

-- 
= Andreas Beck                    |  Email :  <andreas.beck@ggi-project.org> =