krm (12/08/82)
I'm looking for a personal computer sometime next semester but I think that I need lisp. It seems I've become addicted. Does anyone have opinions on currently available lisp systems? I have seen promo literature for Life- boat's stiff upper and mu/LISP, and Datapoint? app-L-LISP and Datapoint?'s version of this for the Atari. Why doesn't someone use a LISP instead of a BASIC as a first interpreter for their hardware? Isn't LISP more intuitive and powerful for about the same amount of resident space? 'chard. decvax!cwruecmp!krm
mwm@Okc-Unix@sri-unix (12/10/82)
From: Mike Meyer <mwm@Okc-Unix>
I've been trying to find a LISP for my CP/M system for about a year or so.
They aren't easy to come by.
General comments: If you can't get to more than 64K, you will be in trouble.
So don't think about the Apple/Atari/etc. as lisp systems. (Caveat: if the
lisp system is the whole system, things may be better. But I wouldn't
bet on it). In fact, you probably want an 8086/8 system, or a 68000. I haven't
looked at Lisps for those, but it would seem you ought to be able to get a
Lis for the IBM PC. Try talking to John R. Allan (author of Anatomy of a
Lisp) at 1-408-353-3857.
For CP/M-80 [I have 400+K w/ caching, so it ought to be useable. Only one way
to find out.], avoid the Supershaft LISP. It doesn't have a pretty-printer,
because it doesn't have the features you need to write one. Like the ability
to print a '('. 'm not familiar w/ Stiff Upper Lismanaged to overlook that). MuLisp/muStar is not unreasonable. Comes with a
visual editor/development system (muStare), that is sourced in Lisp. However,
it has some (mis)features that bother me. Something called `autoquoting.'
This means that an atom is bound, by default, to 'atom. I've never tried
working on a Lisp system with this feature, but I've spent enough time chasing
bugs in FORTRAN/BASIC caused by typeos in a name to not like it. There are
others: implicit sequencing & a default cond/exit mechanism. These result in
any input with balanced parens being a valid function definition. Sounds like
trouble to me. Finally, for cheapies, there's LISP/80 from the software
toolworks. This sells for $50, and includes lot's of nice goodies. It's big
and slow, to the point that it's author calls it unsuitable for anything more
than playing around with LISP.
For the winner, there is TLC LISP, written by John R . Allen. This is billed
as a "extended MacLisp." Includes read tables, autoloading functions, etc.
Sold for $250. Unfortunately, the distribution rights (rites?) for CP/M have
been sold to DRI, so he can't sell me a copy. DRI doesn't know that they have
it yet, and is liable to change the packaging, price, whatever. This could be
good, in the the compiler & swill be, and it could be bad in that the price will go UP.
As for LISP as the basic system, I agree. You ought to be able to get a LISP
system that is both smaller and faster than the BASICs in most micros. I once
wandered the same thing about Tiny-C. If you are adventurous, you could try
putting your own in a ROM and replacing the one the system is supplied with,
but . . .
Let me know what you wind up doing.
<mike
~avsdS:nelson (12/10/82)
There are at least 3 LISPs for CP/M. They were reviewed over a year ago in Byte. mu/LISP comes from Microsoft, but may not have been written by them; a package called mu/MATH for solving equations was written by some people in Hawaii and comes with mu/LISP for under $300. There is a public domain LISP also - CP/M User's Group has it - contact via Lifeboat in NY. Why isn't LISP used as the resident interpreter instead of BASIC? First of all, the existing LISPs take up considerably more space to my knowledge. Second, the public would probably have a hard time understanding it. Third, available software written in BASIC vs. LISP. There is one computer that I know of that has resident FORTH (well, it's more like LISP than BASIC is). That is the Panasonic hand-held, but hardly a computer you'd want as your only one. Buy an Apple, Atari, or CP/M system (consider Morrow Micro-Decision at $1000 with disk) and you'll probably find what you want. Glenn Nelson, Ampex, Redwood City
RG.JMTURN@Mit-Mc@sri-unix (12/11/82)
Lisp on non-tag machines is a real memory eater, by virtue of the more onerous storage scheme. MacLisp gets away with it by having a Dec 10 worth of memory behind it, and sharing areas. I doubt you'll ever see a reasonable Lisp on a 64k system. Saying that Lisp shouldn't be that much harder to implement than BASIC is an interesting statement. The current Lisp Machine virtual load is about 14Mbytes (4.5 MWords). Discounting utilities, the basic system is more like 4Mbytes, and this is on a tag machine. The kind of Lisp you'ld get if it was implemented on a BASIC level (as most are now for micros) is a package with little utility support, extremely limited stack levels, slow as sin, and you can forget Flavors or anything else hairy. If we could design a good Lisp processor for the kind of money you're talking about, it would be built. But it won't happen for at least another N years where N is 2 generations of hardware. What you've got until then is toys and snails. If you won't shell out at least Vax or LispM money, you're not going to get a reasonable Lisp. James Turner