Module exact_codec 

Canonical byte encodings of the exact numbers — the one form they cross a task boundary, the wire and storage in.

Every exact type writes its canonical name: an integer its sign and magnitude, a rational its reduced numerator and denominator, a real algebraic number its minimal polynomial and root index, a hyperreal its reduced numerator and denominator in dt, an infinite natural its residue-indexed polynomials, an ordinal its Cantor normal form, a computable real its exact form or the expression that builds it. Lengths are LEB128 varints.

A decoder trusts nothing it reads. ExactCodec::from_bytes accepts an encoding only when it is complete, has no trailing bytes, and is the canonical encoding of the value it decodes to — so a forged non-canonical name (an unreduced fraction, a reducible “minimal polynomial”, a non-real exponential sum) is refused rather than breaking the equality every other part of the system relies on. Sizes are bounded before any work is done: a minimal polynomial’s degree by MAX_WIRE_DEGREE, collection lengths by MAX_WIRE_LEN, magnitudes by MAX_WIRE_MAGNITUDE bytes, and nesting by MAX_WIRE_DEPTH.

Constants

MAX_WIRE_DEGREE

The largest minimal-polynomial degree a decoder accepts: an irreducibility check is a factorization, and this bound keeps a hostile encoding from demanding an expensive one.

MAX_WIRE_DEPTH

The deepest nesting a decoded ordinal or expression may have.

MAX_WIRE_LEN

The most elements a decoded collection may claim.

MAX_WIRE_MAGNITUDE

The most magnitude bytes a decoded integer may claim.

Traits

ExactCodec

A canonical, self-delimiting byte encoding. See the module documentation.

Functions

read_byte

One byte.

read_len

A length, at most bound.

read_list

write_len

write_list

A length-prefixed list.