Re: Gnumeric regression is wrong?

[Morten Welinder <terra diku dk>]

Fixed in cvs.  For those of you keeping score, I broke it on 2001-02-20.

To test without cvs, in src/regression.c, function LUPDecomp, change the
loop that flips row to

        for (j = 0; j < n; j++) {               /*swap the two rows */
               gnum_float temp = LU[j][i];
               LU[j][i] = LU[j][mov];
               LU[j][mov] = temp;
        }

[Then it no longer matches, the comment.  *sigh*]

Morten