On Saturday, August 10, 2002, at 01:10 PM, Tim Rowledge wrote:

> reic0024 at d.umn.edu is claimed by the authorities to have written:
>
>
>> much of anything about framebuffers or VMs. Jay Carlson actually got a
>> working one up- not sure if it's based on yours.  Works on the regular
>> Linux framebuffer. Found it in:
>> http://vhl-tools.sourceforge.net/ftp/squeak/preview/

Yeah, I ended up writing that from scratch from sqX.  But Tim's code 
certainly counts as an inspiration.

> A quick skim suggests it will need some cleanup before being useful for
> current vms. Right at the top it needs the #include of the joystick and
> serial stuff pulling out, along with any calls to them. Aside from that
> it ought to work.

Aside from being hardwired to 1bpp (trivially fixable), little-endian 
displays (easily fixable) and stuck doing all its own pixel remapping 
(uh, I forget if LE displays and/or generic code to remap the 4bpp/8bpp 
Squeak palette is around anywhere)....

Oh, and there's the small problem of Squeak's two window systems being 
in open form---MVC in particular will eat all your batteries unless your 
VM attempts to guess when MVC is just spinning on user input.  Morphic 
is slightly better in practice (as most people don't set their step time 
to 1 microsec) but worse in design (elevating spinning to a design 
principle).

Somewhere around here I have a version of sqFb  with code for MVC spin 
detection and initial image load via XModem too.

Jay