[Bug-gsl] questionable results for: gsl_sf_ellint_Kcomp --- redux

[Gerald I. Evenden]

Regarding my first submission I found that if I took the root of the arguments 
then the results agreed with the Abramowitz tables exactly:

#include <gsl_sf_ellint.h>
#include <stdio.h>
#include <math.h>
    int
main(int argc, char **argv) {
    double m;

    for (m = 0.01; m <=0.14; m+=0.01)
        printf("%.2f %.15f %.15f\n", m,
            gsl_sf_ellint_Kcomp(sqrt(m), GSL_PREC_DOUBLE),    <---------
                                            -------
            gsl_sf_ellint_Kcomp(sqrt(1. - m), GSL_PREC_DOUBLE));  <-------
                                            -----------
}

This does not make any sense to me!!

-- 
Jerry and the low-riders: Daisy Mae and Joshua
"Cogito cogito ergo cogito sum"
   Ambrose Bierce, The Devil's Dictionary