specs: bench bodies Zig never type-checks (8 specs reject once referenced) #8195

Re-filed from #6268. An attempt claimed that issue on 2026-10-05 and left an empty branch, and a dispatch row that has spent its retry ceiling keeps its issue for good - so no bee could take it again, however much the brief improved. The work below is carried over unchanged.

Problem

Zig analyses a function lazily. A bench body that t27c gen emits is never referenced by a test, so zig test never type-checks it. The bench lane (#6258) referenced every bench with _ = &bench_x; and found that in 8 specs that pass the reference path, Zig rejects the bench body:

• value of type 'u32' ignored (a call result dropped as a statement), in fpga/boards/arty_a7_integration, fpga/boards/qmtech_a100t_integration and fpga/testbench/{cts,partition,power,simulator,router}_tb. Example: estimate_latency(...) in bench_cts.
• unable to resolve comptime value in isa/ternary_gates (a .{...} tuple indexed with a runtime value).

From the same run: 78 of the 116 former-BenchBlock files fail once their benches are referenced, and fpga/testbench/gf16_accel_tb hangs in test_add_zero.

Goal

Every bench body in specs/ type-checks, and a check keeps it that way.

Original checks (kept for history; the Queen does not run these)

• The 8 specs above compile with every bench referenced (_ = &bench_x; or an equivalent t27c option).
• Either t27c gen emits a reference to each bench, or a CI check compiles the bench bodies. Whichever is chosen is written down in the PR.
• The 78-file figure is re-measured and either fixed or listed in a ledger.

Original file list

Spec fixes under specs/, plus t27c's zig backend if the chosen fix is to emit a reference. t27b does not change: it already lowers and compiles each bench body (#6258).

Refs #6063

Queen-runnable restatement (2026-10-05)

The first attempt (branch queen-6268) left no diff against master. Re-measured on master 13e212a by generating each spec, appending test { _ = &bench_x; } for every bench and running zig test --test-no-exec: all 8 specs are still rejected (7 with value of type 'u32' ignored, isa/ternary_gates with unable to resolve comptime value at the bench bodies that are written in Zig syntax - @setEvalBranchQuota, @intCast, for (0..1000)). A spec-only fix is measured to work: writing the dropped call as _ = estimate_latency(...); in cts_tb.t27 removes that error. This restatement covers the spec fixes; the generator option and the CI check from the original goal are hand-written Rust/YAML and are not part of this card.

Boundary

• specs/fpga/boards/arty_a7_integration.t27
• specs/fpga/boards/qmtech_a100t_integration.t27
• specs/fpga/testbench/cts_tb.t27
• specs/fpga/testbench/partition_tb.t27
• specs/fpga/testbench/power_tb.t27
• specs/fpga/testbench/simulator_tb.t27
• specs/fpga/testbench/router_tb.t27
• specs/isa/ternary_gates.t27

User Scenarios & Testing

• Given master, where each listed spec's bench body is rejected by Zig once the bench is referenced, when every dropped call result in a bench is written as _ = call(...); and the ternary_gates benches are rewritten in t27 constructs (no @ builtins, no for (0..n), the operand array indexed by a value the language defines), then zig test accepts each spec with every bench referenced.
• Given the fixed specs, when t27c test-report runs them, then their existing tests still pass.

Requirements

• FR-001: The change MUST make every bench body in the 8 listed specs type-check when referenced, by editing the bench bodies only.
• FR-002: The change MUST NOT change test blocks or function bodies, and MUST keep every bench (no deletion of a bench to silence it).
• FR-003: The change MUST NOT add or modify non-t27 hand-written code (specs/policy/own_language.t27); bootstrap/ and .github/ are out of scope.

Success Criteria

• grep -cE '^ +_ = ' specs/fpga/boards/arty_a7_integration.t27 prints at least 1
• grep -cE '^ +_ = ' specs/fpga/boards/qmtech_a100t_integration.t27 prints at least 1
• grep -cE '^ +_ = ' specs/fpga/testbench/cts_tb.t27 prints at least 1
• grep -cE '^ +_ = ' specs/fpga/testbench/partition_tb.t27 prints at least 1
• grep -cE '^ +_ = ' specs/fpga/testbench/power_tb.t27 prints at least 1
• grep -cE '^ +_ = ' specs/fpga/testbench/simulator_tb.t27 prints at least 1
• grep -cE '^ +_ = ' specs/fpga/testbench/router_tb.t27 prints at least 1
• grep -c '@' specs/isa/ternary_gates.t27 prints 0
• t27c test-report specs/fpga/testbench/cts_tb.t27 | grep -cE '^ FAIL +0$' prints 1
• t27c parse specs/isa/ternary_gates.t27 exits 0 and the spec has at least 2 tests.

The instruments

Every one of these reads and prints; none of them writes. Run them from the repository root, with <spec> replaced by the file in ## Boundary:

.t27 is not Rust and not Zig, and a bee fluent in either writes one by
accident. Measured 2026-09-20: a bee filled eight bodies in
specs/file/watcher.t27 with return Ok(());, Err(FileError::WatcherNotFound)
and for i in 0..watchers.length. The parse ratchet refused the whole file --
Unexpected token in expression: RParen, line 123 -- and a spec that does not
parse generates nothing, so every test it already carried stopped running.

At the top level the parser accepts exactly these eight forms, with an optional
pub, and no others:

const   var   fn   enum   struct   test   invariant   bench

There is no trait, no impl, no type X = ..., no generics, no macro. What
a bee reaches for from another language, and what happens:

• Rust's result sugar (Ok(()), Err(E::V), ?) is not a construct here.
• Rust macros (println!, format!, vec!) are not constructs here.
• A range loop (for i in 0..n) is Rust and Zig, not this language.
• Zig builtins (catch unreachable, @intCast, @constCast) are not constructs here.
• Cross-module reuse (use other::fn;) parses and then generates a comment and an unqualified call, so the Zig fails with use of undeclared identifier (use a::b; generates no import, so no spec can reuse a function from another spec #4298). It does not work yet.

t27c parse <spec> answers in one line whether what you wrote is the language.
Run it before you report, every time.

• t27c spec-status <spec> - the compiler's one-word verdict: IMPLEMENTED, PARTIAL, UNWRITTEN, NOPARSE, NOFN. Exit code is 0 whatever it says, so read the word.
• t27c symbols <spec> - every name the file declares, with its kind. Answers "does this already exist here?" before you add it.
• t27c outline <spec> - per function: its locals, what it calls, what it returns. The contract you are implementing against.
• t27c coverage <spec> - which functions have a test and which do not. The issue asks for tests; this is how you check you wrote them.
• t27c lint <spec> - style and shape warnings. It OVER-REPORTS has no test or invariant: measured 2026-09-20 over 60 specs it printed 677 of those where coverage found 190 untested functions, disagreeing on 59 of the 60 - it warns about bit_to_trit_pair in specs/base/ternary_encoding.t27, which test bit_to_trit_pair_zero calls on line 249. Read it as a hint; coverage is the answer.
• t27c typecheck <spec> - types, before generation. Prints Typecheck OK (0 errors, 0 warnings) or the errors.
• t27c test-report <spec> - builds this spec and runs its own tests, which is what the oracle does. BLOCKED means the generated Zig does not compile, with the error beside it.
• python3 tools/dupe_scan.py --name <function> - where that function already lives, if it does. 576 of 4021 bodies here are byte-identical copies.

The rest of the toolbelt is in docs/BEE_TOOLBELT.md.