garretrieger
commented
1 month ago - Slightly increase the tolerance to account for differences in floating point math across platforms.
- The specific issue is a fused multiply add (d = d1 + (x - x1) * scale; in hb-subset-instancer-iup.cc)
- Also fix more implicit double conversion warnings.