Molt handling of floating point numbers is lacking. The expr command doesn't understand NaN, Inf, or -Inf, so floating point errors are mishandled. Also, floating point output of whole numbers loses the ".0", and so floats are magically transferred into integers. This area needs to be looked at, and fixed. Probably this issue should result in a number of more specific issues.
As of commit c3e9fda7cf5a3d8f54039cfa79ef7b239b561edb, expr understands and outputs Inf, -Inf, and NaN. Issue #29 covers the numeric formatting problem.
Molt handling of floating point numbers is lacking. The
expr
command doesn't understand NaN, Inf, or -Inf, so floating point errors are mishandled. Also, floating point output of whole numbers loses the ".0", and so floats are magically transferred into integers. This area needs to be looked at, and fixed. Probably this issue should result in a number of more specific issues.