Mark the seven resolved-call-emission witnesses test fn so the floor enrolls them (main is red)

[gunbai-bot]

Main has been red since 23:47 on a refusal that is correct and is catching a real defect:

REQUIRED-FLOOR REFUSAL cause=BarrenTestSidecar count=1 — a `*_test.dag` entry must
declare at least one `test fn`; the required floor enrolls nothing from:
dag/test/claim/resolved_call_emission_identity_witness_test.dag

The defect

The file declares seven w_*() -> Bool witnesses, each calling compile_dag_rust_emit_check with positive and negative traps — and every one is a plain fn. Floor discovery selects TestMarkedDecl, so a nullary Bool is a shape, not an assertion. All seven have been dead since they landed.

Marking them is the state the author plainly intended: the w_ names, the -> Bool return, and the file's own note describing them as a PERMANENT REGRESSION CONTROL all say so.

How main broke — neither PR did anything wrong on its own base

time	PR
18:43	#9436	lands the file with seven unmarked witnesses
18:56	#9499	lands the barren wall, 13 minutes later

Each was green when it merged. Squash-merge without rebase never compiles siblings together, so two disjoint green PRs broke main on contact — and the second is the wall that makes the first's defect visible. This is the wall working, not misfiring.

What this does not claim

These seven have never executed. This change enrolls them for the first time and cannot promise they pass. If one is red, that is a real finding about post-resolution callable identity that has been invisible since #9436, and it should be read as one rather than as a regression introduced here.

Checked

• No other module calls any of the seven
• No roster (expected-red, route-gap, non-verdict) names them

so marking them changes no existing consumer.

Found while investigating why the floor refused on gunbc#9517; that PR inherited this same failure through its merge of main, and is blocked behind this.

🤖 Generated with Claude Code

[gunbai-bot]

Closing as a duplicate of #9535, which is byte-identical (same file, same 7 fn → test fn lines) and was opened ten minutes earlier. I found the same failure while investigating why the floor refused on #9517 and did not check for an existing fix before opening this — that check is on me.

Deferring entirely to #9535. One piece of context from my investigation that may be worth carrying there, since it explains why this was invisible rather than careless: the file landed at 18:43 (#9436) and the barren wall landed at 18:56 (#9499), thirteen minutes apart. Each was green on its own base; squash-merge without rebase never compiles siblings together, so two disjoint green PRs broke main on contact. The wall is working, not misfiring.

Also worth flagging for whoever picks it up: these seven have never executed, so enrolling them cannot promise they pass. A red there is a real finding about post-resolution callable identity that has been invisible since #9436, not a regression #9535 introduced.

— sent from witty-wren-148