From: Marcus Sundberg <e94_msu@elixir.e.kth.se>
  To  : ggi-develop@eskimo.com
  Date: Tue, 11 Aug 1998 13:54:16 +0200

Re: Big changes to CVS tree

> > Hmm, the LIBGGI_FB_WIDTH/HEIGHT were refering to entries of the old ggi_info
> > struct. We should probably remove them completely to prevent confusion.
> 
> Why?  It is just shorthand for vis->mode->visible.x/y.  I complained because
> it was set visible.x/y rather than virt.x/y, which was obviously wrong.

If we remove it (or rather rename it, as it might be useful in some places)
it's easyer to make sure that no libs are using LIBGGI_FB_WIDTH as stride.

> > Please adopt Andrew's soon to be used sheme and use:
> > vis->[drw]_frame->buffer.plb.stride
> > ( [drw] as in regexp, not indexing an array ;)
> 
> You mean LIBGGI_FB_[RW]_STRIDE ?

Yes, something like that.

> But all frames are assumed to have the same mode anyway.
> Just vis->w_frame->buffer.plb.stride is sufficient.

With the current implementations yes, but it will probably not always
be true, and it's better to use the read/write frames properly from
the beginning.

> The other thing I found out about stride... it is not easy to make
> driver-libs use it properly, they all do
> 
> (uint16*)LIBGGI_CURWRITE(vis) + y*LIBGGI_FB_WIDTH(vis) + x....

Well, noone said it would be easy. ;)
I'll give it a try right away.

//Marcus