CODECS (NUM-008). InferenceFloat32 and ModelWord32 are structurally the
same 4-byte record (this family's own docstring: "lifted VERBATIM ...
next to ModelFloat64Bits"), but Alpha families are nominal, so the two
conversions below are the only glue needed to reuse Data.Bytes' single,
KAT-proven Word32 codec (evidence/language-testing/L11n-codec-owner-audit.md)
for float bits -- never a second byte-assembly/decode implementation.
98def inferenceFloat32AsModelWord32 =
99 (lambda unrestricted value : (family InferenceFloat32) .
100 (eliminate
101 InferenceFloat32
102 (lambda unrestricted current : (family InferenceFloat32) . (family ModelWord32))
103 value
104 (branch InferenceFloat32Bits byte0 byte1 byte2 byte3 . (modelWord32 byte0 byte1 byte2 byte3))))The compiler supplied declaration spans and resolved links from this source snapshot. This page does not assert that this file belongs to a checked closure.