nx_gatequality_test.nx

nx_gatequality_test.nx -- the gate for the coverage measurer. A coverage tool that nothing checks is the joke it exists to prevent, and worse, its failure mode is silence: an over-reporting coverage number reads as reassurance and nobody goes looking. The three properties below are the ones whose absence would make the measurement anti-correlated with the truth rather than merely imprecise: T1 COMMENTS STRIPPED. Gates here carry long headers that NAME the functions they discuss. Matching raw text would mark a function covered because someone wrote prose about it -- and the worse the coverage, the likelier that prose exists. T2 TOKEN-EXACT. Substring matching would credit si_class to a mention of si_class_of, which is how a coverage tool starts reporting numbers that are reassuring and false. T3 TRANSITIVE. Found by verifying a row rather than trusting it: the management API's gate drives it over HTTP and calls one dispatcher directly, so direct-reach scored it 15 permil. Reachability through the organ's own calls is the honest measure. expect_exit: 0 license_tier: ORIGINAL