You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
t27b: refuse signed % on an untyped local, as the reference does (verify_multitarget) #6593
After #6584, verify_multitarget.t27 compiles under t27b and fails, while the reference (t27c gen-zig + zig test) refuses it, because of a signed % on an untyped local. t27b should refuse it at the same stage, so the verdicts agree.
A conformance spec under specs/tri/t27b/conformance/ with a signed % on an untyped local is refused by t27b with a typed reason.
verify_multitarget moves from fail to blocked in the lab, and reference_disagree stays 0.
Boundary
specs/port/tools/verify_multitarget.t27
User Scenarios
Given a t27 spec containing a signed % operation on an untyped local variable, when t27b compiles it, then t27b refuses compilation with a typed reason at the same stage as the reference compiler.
Requirements
FR-001: t27b MUST reject a signed % on an untyped local during type checking with a typed diagnostic.
FR-002: The rejection MUST occur at the same compilation stage as the reference (t27c gen-zig + zig test) so that verdicts agree.
Success Criteria
t27b specs/port/tools/verify_multitarget.t27 exits with non-zero and emits a typed error about signed % on untyped local.
A new conformance spec under specs/tri/t27b/conformance/ containing a signed % on an untyped local is refused by t27b with a typed reason (exit 0 for the test runner asserting refusal).
verify_multitarget moves from fail to blocked in the lab, and reference_disagree stays 0 (measured by lab metrics).
An attempt claimed this issue on 2026-10-06 and left an empty branch. A dispatch row that has spent its retry ceiling is released once and then keeps its issue for good, which is the right rule for a third identical failure and the wrong outcome for an issue nobody ever attempted twice. The new issue carries the same work, the same boundary, and today's brief.
Closing this one so the swarm can take the work again.
Activity
github-actions commentedon Oct 11, 2026
Re-filed as #9048.
An attempt claimed this issue on 2026-10-06 and left an empty branch. A dispatch row that has spent its retry ceiling is released once and then keeps its issue for good, which is the right rule for a third identical failure and the wrong outcome for an issue nobody ever attempted twice. The new issue carries the same work, the same boundary, and today's brief.
Closing this one so the swarm can take the work again.