nx_gateaddr.nx
buildroot/runtime/nx_gateaddr.nx
about
nx_gateaddr.nx -- CAN THIS GATE'S SUBJECT BE SUBSTITUTED BY A MUTATION HARNESS?
NOT A DUPLICATE OF nx_gatesubj, and the distinction is the whole point: that organ answers WHICH
organ a gate tests (subject IDENTITY, resolved from the gate's own source). This one answers a
different question about the same literal -- WHETHER A HARNESS CAN SWAP THE ARTIFACT THE GATE
EXECUTES. A gate can name its subject perfectly and still be impossible to mutation-test.
WHY IT EXISTS, MEASURED 2026-08-14: nx_gate_bite ran 9 valid mutants against nx_bodyfit_canon_gate
and killed ZERO -- verdict INCONCLUSIVE. The teeth were fine. The gate forked an ABSOLUTE DEPLOYED
path while the harness rebuilds each mutant into the build tree and CORRECTLY refuses to clobber the
live binary. PROVEN, not inferred: the deployed sha256 was byte-identical before and after the whole
sweep, so every mutant ran against pristine code and survived BY CONSTRUCTION.
-- AN END-TO-END GATE THAT PINS ITS SUBJECT TO A DEPLOYED PATH CANNOT BE MUTATION-TESTED, AND ITS
INCONCLUSIVE IS A PROPERTY OF ITS ADDRESSING, NOT OF ITS ASSERTIONS.
That is a measurement-integrity defect: the estate counts such a gate as existing and never collects
its verdict, which is worse than not having it, because its existence is mistaken for coverage.
nx_gateaddr [dir] [deploy-root]
CLASSES -- a PARTITION. The parts are printed and MUST sum to the gate count; an unreconciled
residual is a leak, an explained one is a decision.
PINNED names an absolute <deploy-root> path -> NOT substitutable; a bite here proves nothing
STAGED names a relative _offc/ path -> substitutable; a bite here is meaningful
ELFOTHER names some other .elf -> addressed a third way, REPORTED not guessed
INPROC names no .elf at all (imports subject) -> rebuilt with the gate, bite works
-- UNRECOGNISED SHAPES GET THEIR OWN BUCKET. A shape folded into a known class becomes the number
somebody plans against.
-- COMMENTS ARE STRIPPED BEFORE CLASSIFYING. A scanner that reads comments measures the
DOCUMENTATION, not the code -- and this very header names every pattern it hunts, so an unstripped
scan would classify this organ as its own worst offender.
-- NO SILENT CAPS: it REFUSES at capacity rather than publishing a partial count as a total.
exit 0 census printed | 2 usage | 3 cannot open dir | 4 no gates matched | 5 capacity exceeded
license_tier: ORIGINAL. No hw writes (Rule 26).
dependencies 1 imports · 0 importers
imports: nx_syscalls.nx
imported by: nobody (leaf or entry point)
call flow from main pre-order; caps 40 nodes / depth 6 declared; ↻ = already shown
structs
| none |
consts
| 36 | const GA_DIRBUF: i64 = 262144 |
| 37 | const GA_MAXGATE: i64 = 8192 |
| 39 | const GA_RECLEN_OFF: i64 = 16 |
| 40 | const GA_NAME_OFF: i64 = 19 |
| 41 | const GA_DEFAULT_DIR: *u8 = "buildroot/runtime" |
| 44 | const GA_DEFAULT_DEPLOY: *u8 = "/volume1/homes/elderwesto/nishihost/" |
functions
| 46 | func ga_len(s: *u8) -> i64 { var n: i64 = 0; while s[n] != (0 as u8) { n = n + 1 } return n } |
| 47 | func ga_puts(s: *u8) -> i64 { sys_write(1, s, ga_len(s)); return 0 } |
| 48 | func ga_pn(v: i64) -> i64 |
| 61 | func ga_ends(s: *u8, suf: *u8) -> i64 |
| 76 | func ga_has(h: *u8, n: i64, needle: *u8) -> i64 |
| 97 | func ga_strip(src: *u8, n: i64, dst: *u8) -> i64 called by 1: main |
| 127 | func main(argc: i64, argv: *i64) -> i64 |