tur/bits

(bit-shr [x :int n :int] :int)

logical (unsigned) right shift of x by n bits.

(bit-shr -1 1)  ; => 9223372036854775807  (fills with 0, not sign bit)

Since: Phase B1

(float->bits [x :float] :int)

reinterpret a float's IEEE-754 bit pattern as an :int.

This is the WRITE half of the int64 carrier round-trip. A cons cell, a variadic rest list and a HAMT slot all hold their elements in an :int-typed word, so a float stored there travels as its bits, not as a number.

Not a conversion: (float->bits 1.5) is 4609434218613702656, not 1. Use int->float / float->int (stdlib/math.tur) when you want the NUMBER. :: between an int and a float kind used to pick between these two readings silently and is now a compile error -- see docs/archive/ascribe-int-to-float-expression-ambiguity.md.

(bits->float (float->bits 1.5))  ; => 1.5

Since: 2026-08-29

(bits->float [b :int] :float)

reinterpret an :int's bit pattern as a float.

The READ half of the carrier round-trip, and the inverse of float->bits. Reading a float back out of a cons cell or a variadic rest list is exactly this: (bits->float (list-head c)).

Not a conversion: (bits->float 1) is 5e-324, a denormal, not 1.0. A denormal contributes nothing to a sum, which is what made the old (:: <int expr> :float) spelling of this misread as a DROPPED term rather than a garbage one. Use int->float (stdlib/math.tur) for the number.

(bits->float (float->bits 2.25))  ; => 2.25

Since: 2026-08-29

(println-float [x :float decimals :int] :nil)

print a float with the specified number of decimal places.

(println-float 3.14159 6)  ; prints "3.141590"

Since: Phase B1