Repository navigation
Tags
Tags
Release 2.91 fix t/affine.t: tolerate long-double precision drift affineInvert(scale(2,3)) computes 1/3 in extended precision on -Duselongdouble perls, so the exact is_deeply comparison against a plain-double 1/3 failed on freebsd/linux with such perls (GH #68). Compare with a 1e-6 epsilon instead. Also bump $VERSION to 2.91 in GD.pm, and in GD::Image, GD::Image_pm.PL and GD::Polygon, which were left un-bumped when 2.90 was cut. Co-authored-by: omp Sonnet 5 <sonnet-5@anthropic.com>
Release 2.91 fix t/affine.t: tolerate long-double precision drift affineInvert(scale(2,3)) computes 1/3 in extended precision on -Duselongdouble perls, so the exact is_deeply comparison against a plain-double 1/3 failed on freebsd/linux with such perls (GH #68). Compare with a 1e-6 epsilon instead. Also bump $VERSION to 2.91 in GD.pm, and in GD::Image, GD::Image_pm.PL and GD::Polygon, which were left un-bumped when 2.90 was cut. Co-authored-by: omp Sonnet 5 <sonnet-5@anthropic.com>
PreviousNext