[math-fun] Property lists on INT's and FLONUM's ???

Oops!  Sent to the wrong email list!

At 09:13 PM 3/9/2013, Henry Baker wrote:
>It is interesting to me that MIT Lisp didn't use a true hash table for its "OBLIST" until
>the late 1960's with Maclisp.  This is true even though Ershov (in his _1958_ CACM paper)
>showed how to do _Hash CONS_ (!!)
>
>http://home.pipeline.com/~hbaker1/othergc/Ershov-Hash.txt
>
>Maclisp was able to uniquize "smallnums" by simply allocating them in a fixed portion
>of the address space, but Lisps basically abandoned uniquizing fixnums and bignums, even
>though it could have been done with some reasonable efficiency using hash consing.
>
>I guess that the property lists for integers (and floats) were never used very much, so
>the ratio of work for uniquizing was never paid back by increased efficiency of accessing
>the property list.
>
>In modern Common Lisp, one can create one's own hash tables indexed by integers, so one
>can create one's own "property lists" for integers if one so desires.
>
>At 05:56 PM 3/9/2013, Whitfield Diffie wrote:
>>> I just came across the 1960 "Lisp 1 Programmer's Manual", which I'd never seen before:
>>> As this manual clearly describes, both integers and floats had _property lists_ and were _uniquized_.
>>
>>> Q: When did Lisp numbers lose their property lists ?
>>> Q: Did any other Lisps keep property lists for numbers ?
>>> Q: Were Lisp property lists ever used for interesting things: e.g., storing primality, factors, etc. ?
>>
>>    I had Lisp in a programming course as a freshman in 1961 and began
>>programming seriously in Lisp under Roland Silver in 1965.  I do not
>>recall property lists on numbers.  I suspect it was a quick
>>implementation hack for the first version and was eliminated in favor
>>of some faster way of doing arithmetic.