Why LispOS?

[cosc19z5@bayou.uh.edu]

> Perhaps this has been discussed before, but I'm curious about why
> people want LispOS.  What do we expect it to do for us that just
> running some lisp system on unix won't?

I like this question.  If for nothing else it will get people
to look at what they want, and what they expect.  This will
help form a goal for the ultra-rare possibility that a
LispOS (from this mailing list) will become a reality.

I've posted some ideas about what a first step should be
(the topic was "CL OS Ideas" or something similar).  I think
reading that message will give you my answer to your question.

To summarize it however, what I expect a LispOS to give me
that a standard Lisp system on Unix won't is a completely
Lisp-Centric view of the computing environment.

I want Lisp for commands, Lisp for the application/scripting
language, I want Lisp objects for the "file system", I
want "files" to consist of Lisp objects.  "Files" (or
rather Persistent Storage Objects (PSOs)) would ideally
be structured so that more sense can be made of them
and to ease in their manipulation (makes things higher
level).

I want Lisp objects to be the result of all commands, so
that I can use, parse, and evaluate them as I would any
Lisp objects.

I want the equivalent of "ls" to give me a list that I can
happily use.

THAT is what I want a LispOS to give me.  True, that this
can be accomplished by a Lisp shell riding on top of Unix,
or Winblows or whatever.

When I turn on my computer, I want to do everything I need
to do without ever resorting to anything other than Lisp.

This is my dream.


> 
> My own interests are for a more integrated lisp development
> environment.  For example, I still remember using Masterscope on the
> Dandelion; I could display a function call tree, click on the nodes
> and get source, do all kinds of queries as to who called or used what,
> and so on.

As part of the LispOS/LispShell project this would definitely
be a priority in my ideal world.  Having the entire computing
environment in Lisp would be just one step -- enhanced tools
for Lisp development would be implemented.


> 
> But this doesn't require LispOS.

No, but having a bona fide Lisp-friendly kernel could make things
more efficient.  Maybe.

Having a LispOS/LispShell (something that appears like an O/S to
the end user anyway) would fit in very nicely with such tools.
Your computing environment is your development environment.
I like the idea.  The distinction between application/scripting/
Commands is blurred.  Whether you are writing a simple shell
automation script, entering a command, or writing a videogame,
the process is the same -- it appears like you are programming
in the native O/S language.  Not some low level machine code,
or a third rate excuse for a "structured" language.  No,
a language that is the essence of your operating environment.

Add in some other niceties, like maybe a Lisp-isized database
query language (one that you can load into the environment
and use from the command line or within your own code) and
I see a computing environment that can't be beat.

A nice, intelligently designed, Lisp-Centric windowing system
would do great.  No one would have anything on us then!

Of course, the ability to run other programs is important, and
if this is simply a shell, then compiled code can be run, but
it can be made to fit as seamlessly as possible in the
shell. 

Ideally, applications will eventually be re-written/modified
to be as lisp-centric as possible.

Lisp will be all that there is.


> 
> I'm also interested in a top-to-bottom environment that exposes more
> of the machine at the lisp level.  This does seem to be something that
> LispOS is needed for.

Again, as part of a shell project, an API would be provided.
Everything that can be accessed from the O/S / Hardware will
have a Lisp-friendly wrapper.  Lisp will ideally do EVERYTHING
you need.


> 
> Another question that arises is whether there's anyone with the time
> and desire to actually code this up.  Personally I have neither
> (because I'm currently not interested in groveling around operating
> system internals and such).  I'm more interested in working at a
> higher level, perhaps working on Masterscope-like tools.  I'm also
> limited in time---if I really get going I can spend 3 or 4 hours a
> week.

I would not mind working on this, but there are a few problems
involved:
	1) I am not using a Unix system at home, and I have
	   no desire to change.  Since many people seem to
	   think that Linux or FreeBSD should be used, this
	   may knock me out of the running (before I've started
	   I would say!).

	2) I know nothing about O/S internals, and I have no
	   desire to code in anything other than Lisp.  Be it
	   Common Lisp (preferably) or Scheme.  I do not want
	   to soil myself by hacking C or C++ or assembly.  I
	   used to be willing to do this, but frankly I'm
	   more sick of C and C++ than I ever was.

	3) My time is limited, but I will happily give what time
	   I can over to this project.



> I'd love to be able to slip a CD in my Intel box and an hour later
> have a Symbolics look-alike.  But I can get along with what I have.
> Is there anyone on this list who is just foaming at the mouth :-) to
> create the Symbolics look-alike?

If I knew what Symbolics was like, maybe I'd be foaming right about
now :).  From all the praise I've heard heaped upon this system,
I would be very interested in seeing a Symbolics clone and would
happily try it out (if a demo were available).


> 
> -Fred Gilham   gilham@csl.sri.com
> 

Regards,
Ahmed