[Bug-gsl] Bug in gsl_sf_coupling_3j_e

[Martin Landriau]

Dear developers,

I found a problem with the function gsl_sf_coupling_3j_e located in the file 
specfunc/coupling.c : the tests should pick up that 

(l1 l2 l3) = 0
(0  0  0 )  

if l1+l2+l3 is not an even integer and it doesn't.  It instead attempts to 
evaluate it and sometimes returns a non-zero value, albeit one with an error 
larger than the value itself.

Cheers,
Martin

P.S. I am using 1.15 on Mac Os X 10.6 and all tests were passed.

---
Martin Landriau
Max-Planck-Institut für extraterrestrische Physik
Giessenbachstraße 1
85748 Garching
Germany
---
T: +49 89 30000 3735
F: +49 89 30000 3965
---