[LMH] Re: [LMH]Genera world load format?

[Mike McDonald]

>To: Mike McDonald <mikemac@mikemac.com>
>Date: Fri, 04 Jun 2004 17:49:08 -0400
>From: Brad Parker
>
>
>Hi!
>
>I was looking over your code.  Wow!  nice work.

  Aw, shucks. Thanks. Unfortunately, I haven't had a chance to look at
your code yet due to ever present work problems. I did grab a copy
though.

>It looks like you where trying to emulate the i-machine.  I've been
>working on the l-machine.
>
>One reason is that I have not found any docs on the i-machine
>instruction set.  But I must admit I never looked (so far anyway) at the
>doc files on my lisp machine... (oops)

  I picked the i-machine for two reasons. One, like you, I have a
refurbished XL1200 from Schmidt. I figured that having a running
example would help in researching the desired behaviour. Second, there
are world loads that are going to be a lot easier to get to run for
the i-machine than the l-machine, namely the ux400. It use RPC to
access most of the I/O devices. The sources for the Sun side of the
RPC is also available. Emulating all of the custom hardware so that
the existing microcode can't tell the difference is going to be a
pain.

>Where did you get info on the macroinstructions?

  Genera 8.3 included a lot of the sources not normally released,
including the l_sys and i_sys directories. sysdef and sysdf1 contain a
lot of the information. Add in the sources for the debugger and the
compiler got me as far as I am.

>On the l-machine I've been using the microcode definitions, which are in
>lisp and include a rudimentary simulator...
>
>There certainly seems to be a large amount of overlap with the l-machine
>macroinstructions.
>
>-brad

  Yes, there is. I think your work on the l-machine microcode will
help me a lot on the i-machine macrocode. It'd probably even help the
TI LispM folks if they bothered to look at it.

  One of the things you've mentioned on your website is going some/all
of the LispM code to run under Franz Lisp. Can you tell me more about
this? One approach to getting a virtual LispM running I've considered
is porting the sources to an existing CL system. I was thinking of
using Franz ACL or CMUCL.

  Mike McDonald
  mikemac@mikemac.com