[Bug-gsl] gsl_ran_beta_pdf (0,1,2) gives "nan", not 2

[Neil Harvey]

Hi,

I've run into a problem with gsl_ran_beta_pdf giving "nan" in some
cases.  The subject line gives one example:

gsl_ran_beta_pdf (0,1,2) = nan

But the value should be 2.

I found this with both 1.8 (binary package for Intel from debian
unstable) and 1.6 (compiled from source), and no similar report in the
bug-gsl archive.

Thanks in advance for checking that out -- and thanks in general for
the GSL, it's an incredibly useful library!

Neil
U of Guelph, Ontario, Canada