> > BTW, in regard to the pow.t tests, the following is weird: > > my @pow = ([3,30,1e-14], > [4,32,0], > [5,20,1e-14], > [2.5, 10,,1e-14], > [-2, 69,0], > [-3, 30,0], > ); > > In the above, the test for powers of 3 has an epsilon of 1e-14 (which all of > the pow.t test failures seen are well below). But the test for powers of -3 > has an epsilon of 0. This is inconsistent. Why should -3**$x be more precisely > determined than 3**$x? The above was added with change 20169, BTW. The [2.5, 10,,1e-14] test, with its extra comma, is particularly weird :-) -- jplThread Previous