v0.2.0

Microfloats v0.2.0

Diff since v0.1.1

• Add @microfloat macro
• Add NonFiniteBehavior trait
• Avoid private Base names, exposing new exponent_bits, significand_bits functions.
• Use Republic.jl for marking public API
• Fix world age issue (caused by generated function) when converting from types defined outside package

Breaking:

• Make Microfloat abstract
• Make OverflowPolicy a trait inherent to the type instead of conversion argument
• Rename types and remove MX_* aliases
• Fix Float8_E8M0FNU (previously MX_E8M0) rounding and remove zero value
• Remove BitPackingExt

Merged pull requests:

• Bump actions/checkout from 5 to 6 (#18) (@dependabot[bot])
• Bump julia-actions/cache from 2 to 3 (#19) (@dependabot[bot])
• Bump codecov/codecov-action from 5 to 6 (#20) (@dependabot[bot])
• Bump julia-actions/setup-julia from 2 to 3 (#21) (@dependabot[bot])
• v0.2 (#22) (@AntonOresten)

Closed issues:

• MX_E8M0 zeros (#11)
• Abstract Microfloat type (#12)
• Opt in to specific saturation/overflow modes by overdubbing methods (#16)