Collecting random inputs



C Y wrote:
> 
> --- Richard Fateman <fateman@cs.berkeley.edu> wrote:
> 
> > I also have integration look-up code, though it is not
> > really macsyma-oriented, rather it is a web server.  The
> > idea is: if you have a problem, look it up in our table.
> > (20 millisecs).  If you find it, go away happy.  If you don't
> > find it, run some algorithms.
> 
> I'd noticed that - one of the cooler websites I've seen lately :-)  Has
> anyone tested maxima against that huge list of intergals?
> 
> > I would still like to figure some way of getting the
> > commercial macsyma improvements into maxima.
> 
> Hear hear.  Unfortunately, that would mean finding it, and convincing
> whoever has it to let go of it.  If I recall correctly, Richard Petti
> can't disclose the buyer, and the buyer sure hasn't disclosed
> him/herself, even by marketing the product or trying to sell it. 

It's not quite THAT mysterious. I think the buyer was
Andrew Topping.  Why he let all the people go and stopped
selling it is not clear.

> Also,
> since that person probably paid a good chunck of cash for it, they are
> also probably going to be reluctant to let go of it for free.

I don't know how much he paid for it.  What would you
pay for a company that loses money each month, and has
done so for years?  (I know that also characterizes
Amazon.com)
> 
> > In any of these efforts (my own included)  there is always
> > the danger of someone saying "I have a really good replacement
> > for the X module"  when in reality it is in almost all
> > ways inferior.  A valuable component of this group
> > effort would be thorough testing.  Presumably the
> > commercial people developed test suites.
> 
> If not, maybe we could start with the make test stuff and develop our
> own, adding each case anyone has trouble with in old or new code and
> building up from there

Sounds good.

>