mirror of
https://github.com/block/buzz.git
synced 2026-10-27 20:41:57 +00:00
a3fb049a09
Two residuals from the gate-satisfiability lap, both single-clause, both found by a two-encoding model of the landed bytes: Release-gate conjunct 1 still required the claim tuple to pass every applicable oracle. FI-CONF-INTEROP-EXIT is applicable by construction -- the deferred exception presupposes it, and not-applicable requires behavioral proof of an absent surface, which a missing second implementation is not. Deferred is neither pass nor not-applicable, so under the literal reading the gate still failed on its first conjunct for exactly the shape the deferral exists to permit; only a reader supplying the unwritten carve made it pass. The conjunct gains the same validly-deferred pointer as the other two. This residual was created by this lap: the sentence is byte-identical across heads, but "deferred" entered the vocabulary it quantifies over. The pending-canonical-fixture reserved value was valid only alongside a deferred exit result, but no listed oracle detected a violation -- FI-CONF-CLAIM-COMPLETE enumerated its rejection grounds and tuple-value validity was not among them, admitting the one reachable lying-tuple state: exit run and passed against the published fixture while the tuple still records the placeholder, defeating the tuple's purpose of naming the inputs actually run. The row gains that rejection ground. Verified: two-encoding gate model at the amended text reads literal one-implementation admission 0 -> 3 with the strict/literal delta closing to 0, attribution control varying only the conjunct-1 carve (arms disagree), exemption inert with a second implementation, no leak to ordinary oracles; fixture-misuse enumeration has the old row admitting both reachable misuse states and the amended row zero, with every legal shape kept and an over-broad clause that kills the legitimate deferred-plus-placeholder shape as the positive control. Oracle census unchanged at 58. Co-authored-by: Eva <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> Signed-off-by: Eva <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz>