check_core_leaves.nika passed a ref Ty? from a block arm, and rustc refused the lowered code. About 15 turns of probes (q2–q10) narrowed it to a_lent_part_is_handed_on_as_it_is, which removes the argument from lent_args in block arms. Next: bootstrapping. The old binary lowered my own fix wrongly, so I had to build once without that line, put it back and lower again. Third: the brief’s planned fix (the type in the cast) does not work in rustc. That cost one full build cycle before a 10-line rustc check showed it.self.facts, no Rust adapter);integer_named (in check_names.nika) and that Ty::Fn.result is Ty?;emit_flow_result’s neighbours, and the lent_args / views_handed_on code in the checker and emitter.That scripts/commit.nika reads the attribution only from <report minus .report.md>, so a .stepN.report.md gets the wrong co-author and session. The brief could have named that, plus integer_named.
nikaia run probes; one plain rustc probe; the guard set, clippy and pre-push-check; tests/compiler at both settings (411 green).as cast’s type and a 10-second rustc probe would have shown it first. Also a second build that failed because of the bootstrap problem..rs file and compile it with rustc.” That would have caught the wrong cast plan before any build. Second: in builder-facts, “a fix whose own tools code needs the fix to lower: build once without the caller, then add it back”, and that step reports must commit through the main report name.