[rescue] Sick Dreamcast hacker

[Joshua D Boyd]

On Tue, Jun 18, 2002 at 01:12:37PM -0400, Dave McGuire wrote:
> On June 18, Joshua D Boyd wrote:
> > And there is no need to steal the screen from a palm.  We can buy them
> > easily enough.
> 
>   Yes.  Take a peek at this link...this group has found a very nice
> display, and the LCD supplier for my current project has something
> similar as well.  They're even relatively easy to drive.
>  http://openhardware.net/Dragonix/

Spiffy.
 
>   I'd be up for designing the hardware and maybe some of the low-level
> software (it's very closely related to the stuff I'm doing for work
> now anyway) if someone else can do the high-level software...my brain
> isn't mathematically-oriented...

Well, I know how do things like do simple UI programming, and I know
how to write all the needed math code (well, a lot of it would be
fairly poor compared to what is used commercially, but here poor
mainly means slow, not incorrect).  I think I probably could write a
scheme interpreter, but going from writing one to implementing one in
an FPGA some how...

The tasks that would need to be done, as I see it (in no particular
order) would be:

1) define the lisp environment (ie, the specifics of the language to be
   used)
2) define the user environment (what buttons?  where, physical layout?
   How do buttons react?  etc)
3) Write all the math code.
4) implement the lisp environment in hardware and software.  Taking
   Sexps and parsing them into whatever the best form is (I'd have to
   look up what the concensus was among the various lisp machines, if
   there was one) a software job.  Executing that results of the parse
   is the hardwares job.  Memory mangement, device management is
   implemented as is usually done.
5) implement the UI in hardware and software.

6) (optional) create a simulated lisp and user environment on a unix
   machine so that the details of the UI can be hammered out and the
   math library developed.

Now, step 4 is probably by far the hardest.  Doing step 4 in a manor
that it can boot from flash storage and communicate to TTY would be a
worthwhile project by itself, and perhaps it would be best to do it
that way initially.

If we really wanted a lisp based calculator (say as a commercial
product), it would probably be best to start with a 68k [1] (or perhaps
something with an MMU) base and not bother with a true hardware lisp
environment initially.  In building a true hardware lisp machine, we
can draw on some papers written about the old machines, and once we
have most of the work done, we can start borrowing lisp or scheme code
from various places.  If we start on a 68k base, we don't have to do
anything from scratch except what we really want to.

-- 
Joshua D. Boyd


[1] It could be anything, preferably 32bits with FPU though.  I just
    happen to like the 68k and related chips.