Hi, [Regarding gsl-1.8]
When using a compiler with a division by zero trap, I have found a *possible* bug in gsl_sf_legendre_sphPlm_e function, in legendre_poly.c file. The line where the error occurs is 575: result->err += fabs(y_mm_err/y_mm) * fabs(y_ell) Since I need to compile with these checks enabled, I have substituted it by: if(fabs(y_mm)>GSL_DBL_EPSILON) result->err += fabs(y_mm_err/y_mm) * fabs(y_ell); which seems to work. But I am not sure if it is the proper solution, I have no experience looking in the insides of gsl. Alberto. _______________________________________________ Bug-gsl mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-gsl
