On Wed, 2004-10-06 at 11:53, Barton Willis wrote:
> Do we need to downcase the rtest files?
That may or may not solve the problem. Now that the "trivial" exercise
of downcasing the code itself is done, the real problem to be solved is
making maxima truly case-sensitive. I intend to start working on that
problem soon.
In that context, downcasing the rtest files seems premature. I am not
even sure if downcasing everything would work at this point. Feel free
to try it if you like it.
> When my file "known-bugs.mac"
> contains the test
>
> /* bug 1036546 */
>
> realpart( (%I*%E^-(%I/2)-%I*%E^(%I/2))^a );
> sin(1/2)^a*2^a$
>
> we get trouble later on in rtest6a and rtest9. If I
> partially uppercase my test
>
> realpart( (%I*%E^-(%I/2)-%I*%E^(%I/2))^A);
> sin(1/2)^A*2^A$
>
> then no problem....
Excellent! The behavior you see is a perfect example of the subtle
long-range correlations in maxima code introduced by the current
case-determining algorithm.
--Jim