Testsuite failures in rtesthyp.mac



Raymond Toy wrote:
> I see this effect too.  I guess that means there's some kind of compile
> time issue with gcl.  I don't know what it is, but gcl is definitely
> computing hgfred([-a,a],[3/2],z) incorrectly after assuming z < 0.
>
>   
It has something to do with the fact that the variable M in STEP4-A is
special.  At the beginning, M = 0.  But just before the call to ALGIII,
it now suddenly has the value -1.

I wish we could change all special variables to use the *foo*
convention.  Or even just declare them special as needed instead of the
declare-top special and declare-top unspecial stuff.

Ray