Skip to content

Draft: Fix #275: Signal underflow when reading small floats

Raymond Toy requested to merge issue-275-signal-underflow-in-float-reader into master

Previously, small floats that were smaller than least-positive float would signal a generic reader error that the number was not representable. And we explicitly disabled the underflow trap.

Instead, allow underflow traps and catch it and re-signal it with a restart to flush the value to 0.

Merge request reports