----- Original Message -----
From: <hjoksch at att.net>
> What is wrong with pointing out
> that a certain function in package A is slower than in package B?
There is nothing wrong with this;
however, there is nothing wrong with saying
"It is probably not relevant" :)
Usually, the programmers of package A would like to find out how B does it
better, and try to improve their program if that can be done without other
disadvantages.
Only if the programmers consider it worth their time.
>
>>
>
> Derive 4 is a very old package, written primarily for pedagogical
> purposes. I would have expected the much more sophisticated Maxima to be
> simply more efficient.
This would seem to make sense, except that Derive and Macsyma also have
different objectives.
Sometimes Deriv is faster, and even gets "more correct" answers. Sometimes
it is quite unable
to do something that Macsyma can do. It depends on what the Derive authors
had in mind
to include and to exclude from that package.
That its performance should be critically dependent on the Lisp dialect
used surprises me very much; after all, some running times differed
dramatically between the Maxima and Derive.
The times between Lisp implementations can vary dramatically; even the same
"brand" can
become twice as fast from one version to another, depending on how much
effort was
spent in improving some part of the compiler.
>
> If that is indeed the case, including some specifics about which Lisp
> dialect does what better
> in a README for Maxima file would be worthwhile.
This would be difficult to predict, though my guess is ... generally CMU-CL
and its descendants are
faster for numerical computation, and CLISP is smaller but perhaps somewhat
slower.
Perhaps others on this mailing list have experience to offer.
RJF