Robert Dodier <robert.dodier at gmail.com> writes:
> On 2013-09-30, Wilhelm Haager <wilhelm.haager at htlstp.ac.at> wrote:
>
>> I think, a new bug has been introduced in Maxima 5.31.1-1:
>> realroots hangs when trying to calculate the real roots
>> even of a simple polynomial:
>>
>> (%i1) p:x^3+2*x^2-2*x-1;
>> (%o1) x^3+2*x^2-2*x-1
>> (%i2) realroots(p); .... HANGUP!
>
> The problem is that recent versions of GCL parse 1e-6 and 1e-7
> incorrectly in compiled code. I've reported it to the GCL project.
> In the meantime I've replaced those literals by (float <rational>)
> (there are only a couple of them).
>
> A user can set rootsepsilon:1e-7 to get the expected behavior.
>
> Of course one may wonder if there are other problematic literals ...
> I didn't find any but I didn't look very hard either.
Ah, I bumped into this GCL "feature" myself. There's a fix for a
testsuite failure in
f43ba19d5f1897263478678e313cdba587095ef8
which I think is basically the same problem.
David's work on getting SBCL+Windows is fantastic - maybe we can
eventually get the majority of our users on a slightly less
idiosyncratic lisp implementation :-)
Rupert
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 315 bytes
Desc: not available
URL: <http://www.math.utexas.edu/pipermail/maxima/attachments/20131002/c87bac72/attachment.pgp>;