Source/Reference

FloatLiteralSpec

reference/numeric/FloatLiteralSpec.alpha

243 lines42 declarations13.2 KiBSHA-256 136c61d4a2e6

field · lines 24–24

specDenominator

Full file
The N5 SPECIFICATION of decimal -> IEEE-754 conversion as an .alpha algorithm over compact naturals (Language & Testing Evolution L12c, PRD 08 N5, NUM-002, REG-021). Every arithmetic step is one of the L12k compile-time heads (nat-add / nat-multiply / nat-subtract / nat-divide / nat-modulo) or the core nat-less-than, so on closed inputs the reference evaluator folds it in O(digits) per step. This module is a REFERENCE (never in a package root and never built): a definition holding compile-time arithmetic with open operands is refused at erasure by design, and this module holds nothing else. It is evaluated under the draft edition alpha-2027 by debug/float-spec-check.py, which compares it row by row with reference/numeric/float-literal-kat.tsv (the exact-rational authority) and with the bootstrap's literal elaboration. Inputs are naturals only: (negative flag, significand digits, exponent magnitude, exponent-negative flag) = (-1)^negative x digits x 10^(+-exponent). Signed binary exponents are carried with the offset specOffset (= 4096) so every intermediate stays a natural. numerator / denominator of the exact value
24field unrestricted specDenominator : Nat

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.