nx_inflate_gate.nx

nx_inflate_gate.nx -- PROVE THE DECOMPRESSOR AGAINST AN INDEPENDENT IMPLEMENTATION. ★A ROUND-TRIP AGAINST OUR OWN COMPRESSOR WOULD VALIDATE ITS OWN BUGS. If nx_gzip and nx_inflate shared a misreading of RFC 1951 they would agree perfectly and both be wrong -- the banked "a round-trip equivalence gate cannot kill a symmetric mutant" law, applied to compression. So the fixture below is a REAL gzip stream produced by an INDEPENDENT implementation (system gzip), embedded byte-for-byte, together with the exact plaintext it must yield. Agreement here is cross-implementation evidence, not self-consistency. The CRC32 tooth uses the PUBLISHED IEEE check value for "123456789" (0xCBF43926) -- a constant from the standard itself, not a number we produced and then declared correct. NEGATIVE CONTROLS ARE THE POINT. A decompressor that silently truncates on corruption is worse than one that is absent, because it hands you plausible partial data. Every wall is fired in-run. license_tier: ORIGINAL No hw writes (Rule 26).