Re: Hardware Descriptions

[Roger Gregory <roger>]

Actually as I see it the technical problems and choices 
are a little different than you have stated.

1) Mac BE may not require a MacII.  Except for VM the Mac's
are all the same to us.  Of course for speed reasons it would be
silly to run on a Mac+. However if someone wants to do it
bad enough, they should be able to.  Memory is another matter,
the small Mac's have limited ammounts of ram.  Certainly
putting Xanadu on an SE30 is interesting to us as a portable
demo unit.  It should be no different than a MacII (Hugh, any
opinion).

2) Suns:  We will not support the Sun1 or the Sun2 except under
special arrangement.  It would require a recompile, and perhaps a
new compiler, as we haven't bothered to get C++2.0 ported to the Sun2. 

   Sun3 & Sun4 are no problem, we may have some required system rev
level (i.e. 4.0.3 or newer) but this requirement could be lifted if
required.  Many Sun3 users still use SunOS3.5, and at the very least
we would have to test under that environment.  The most we would have
to do is generate a C++ that will run on 3.5 and recompile.
Fortunately Autodesk will probably have some idea of what is in the field.

  The Sun386i port may or maynot be out at the sametime as the other
Sun's, depending on the compiler being fixed enough to compile our
code.  This is a low priority, though as soon as we have a system that
works again we should test the 386i so we can send the bug report to
sun with some expectation that they will fix it by the time we need
a working compiler.

  The window systems for the sun are part of the frontend, so
3rd party developers shouldn't be overly concerned with our choice
here.  Users will be,  my strategy here is to build the first FE
for portability basing it on the window systems that will become
standard on Sun's eventually.  If we find that we hit that market
before XView really penetrates that market, we will have to do a
Sunview version as well.

3) PCoid machines.  Here I agree with marcs, mostly.  A straight
DOS seems so difficult as to be impossible, however a DOS extender
version might be relatively simple.  This is one that really needs
market feedback.  I can't understand why someone would want to run
DOS rather than OS/2 on a dedicated backend server, but if a lot
of them do, it would be nice to know how difficult it will be to
have something to sell to them.