[plt-scheme] Native code generation and immutable pairs

All-

On Mar 11, 2006, at 10:05 AM, Matthias Felleisen wrote:

>
> On Mar 11, 2006, at 1:20 AM, Jim Blandy wrote:
>
>> On 3/10/06, Lauri Alanko <la at iki.fi> wrote:
>>> Also, at some point eq? on pairs needs to be unspecified. Once pairs
>>> have no state, they don't need identity, either.
>>
>> I like Henry Baker's paper about this:
>> http://home.pipeline.com/~hbaker1/ObjectIdentity.html
>>
>> It bothers me that the more I think about this stuff, the more I
>> appreciate ML...
>
> The abstract suggests that what he wrote up in 1990-92 in this  
> essay had been well-known in the Scheme and OpSem community.  
> Indeed, I was teaching
>
>   (define (my-eq? x y) ... (set-car! x 0) ... (set-car! y 1) ... (=  
> (car y) 1) ...)
>
> in the freshmen course at Rice back then. My own knowledge goes  
> back to Indiana, probably some Steele-Sussman paper. The op-sem  
> people proved things about this eq? as early as 1986 LiCS (Ian  
> Mason, advised by Carolyn Talcott).
>
> Computer science, the art of multiple rediscovery.

Also, Gerry Sussman gives a very direct illustration of the idea with  
a pair of pieces of chalk in the SICP videos (circa 1986).

((some links below; the easiest thing to do, for iTunes users at  
least, is to use the feed: link))

http://lambda-the-ultimate.org/node/1096
feed://feeds.feedburner.com/SICP
http://podhostess.com/SICP/
http://swiss.csail.mit.edu/classes/6.001/abelson-sussman-lectures/


Fast foward to maybe four-fifths of the way through and you'll see  
Gerry talking philosophy about two pieces of chalk.  (Yes, it comes  
back to set! and eq? in the end!)

-Felix