This alpha release introduces hardware single-precision floating-point support on the RP2350, STM32F407, STM32L476, and STM32F746. It is mostly working in its current state, even though the string conversion routines could possibly be improved upon. If anyone has any input on how to practically improve them without resorting to things like bignums, it would be greatly appreciated. This release also fixes a bug where literals immediately before assigning their value to double-cell local variables would not be compiled properly.