how about we get rid of "fast arrays"



On 2012-07-13, Richard Fateman <fateman at eecs.berkeley.edu> wrote:

> Where did fast arrays come from? Who did this?

Can't be sure at this late date, but I'm pretty sure it was Bill
Schelter. use_fast_arrays is present in the Maxima 5.0 tarball at
Sourceforge, which comprises stuff Bill worked on by himself, if I'm not
mistaken. The timestamp on src/mlisp.lisp is 1994-09-28. use_fast_arrays
is not present in the Macsyma version 302 (1983) reference manual which
was bestowed on me by Judah Milgram (he rescued it from the library --
it was headed for the recycling bin -- thanks Judah!). 

Incidentally the reference manual was published by "MIT/LCS". Wasn't MIT
out of the Macsyma business by then? I guess not -- looks like I have
gotten the chronology mixed up. 

> Does anyone use this?

Dunno. Seems unlikely, to be honest. All the questions and discussion
have been among the developers.

> If we are going to start removing features because we don't think they 
> are very good and "so far as we know"  no one is using them, seems
> like a slippery path.

Slippery path, eh? Me first, me first! WHEEEEEEEEEEEEE ....

> Putting an array in a value cell may be important to someone.

Agreed. That should still be a feature. I am inclined to do it via a
function (make_hash, say) which returns a hash table. I'll write a more
detailed proposal in a day or two.

best

Robert Dodier