[SGVLUG] Fwd: MITx is open for enrollment

[Dustin Laurence]

On 02/13/2012 07:20 PM, Claude Felizardo wrote:

> AI languages eh?

Anyone at MIT in that time was a lisp person for the same reason that 
Italians grow up to love pasta.

But Gerry Sussman may not want to be pigeonholed as an "AI guy".  It 
turns out he did some fairly hairy numerical work demonstrating chaos in 
planetary orbits *on supercomputing hardware of his own design.*  I 
doubt he thinks there's much he can't tackle with any kind of software 
toolset he chooses, and he's probably right.

He also turns out to be one of the original Free Software Foundation 
board members, so he knows all about our little C-based free unix world. :-)

> ...I used Smalltalk at a dot-com and at JPL and LISP
> was used to extend AutoCAD years ago but I have never had the occasion
> to use just Scheme other than in a class.

For what little it's worth, here are some complaints I've gathered from 
actual lispers.

The problem with Scheme is that in order to keep it a small teaching 
language, they didn't standardize a wide and deep library for all the 
normal practical computing tasks.  Of course people still wrote tons of 
libraries for practical work--over and over again, in different 
incompatible ways.  I think that was a fundamental mistake, but 
hey--nobody asks me.  The result, I gather from people who know better 
than I, is that scheme isn't as useful as a general-purpose language as 
it should be.  Sometimes, the function you need isn't there.

Common Lisp, of course, took the opposite approach, with a standard 
document for libraries and extensions totaling several planetary masses. 
:-)  You can't say CL isn't there to hold your hand, but it holds 
*everyones* hand by remaining backwards-compatible with code written for 
zillions(TM) of dialects and libraries.  *The function is there*--but 
which of the dozens of subtly different tools is best for your job is a 
matter best discussed between you and your psychic advisor.

Most of the real work I hear of being done in lisp (OK, OK, so I've only 
got a couple of examples--*work* with me here) seems to be in CL, which 
sure isn't going to be what was in autocad.

I'm not sure which I'd choose if I was told I had to do something in 
lisp.  I think I'd try to measure whether the lisp or the scheme 
communities are more annoying about rejecting the world that rejected 
them, and then use the other one just to annoy them. ;-)

Seriously, I'd be *happy* to have paid work in lisp so I would really 
learn it, but it isn't going to happen.

 > ...The "Wizard Book" was after
> my time but I still have my "Red Dragon book" somewhere...

Totally different books--*the* Dragon book was for implementing 
compilers (especially for big hairy compilers for static Algolish 
languages, IIRC), and maybe graduate level?  The Wizard Book was MIT's 
freshman programming text for an age or two.

But there is an odd overlap--the Wizard Book spent a lot of time making 
sure you knew how to interpret lisp in lisp itself (OK, scheme in scheme 
itself), so it was interested in certain language translator issues. 
That's actually a sane thing to do in lisp, for suitably chosen values 
of "sane."

I'm quite tempted to sign up for the hardware class.  Where, oh where, 
could I find the time? :-(

Dustin