Seventh roster dissolution: XL-0N's transition-admission row is stale on every open PR - removed by the trigger it was authored with

[gunbai-bot]

Fleet-unblocking. Every open pull request based on 6ef73334e (XL-0, #9710) or later is currently red on the namespace-wave-admission phase regardless of what it changes.

What

One TransitionAdmission row, authored by XL-0N (#9719) for the relocation of type_reference_declaration_ref out of v1.compiler.emit_rust into v1.compiler.infer_env. It carried its own dissolution condition:

DISSOLVE-ON: this pull request merging. Base and head then both carry the relocation, no run can produce this delta, the row reports stale on every build, and it is removed by that trigger exactly as all six shrinks above were — a stale row here refuses every unrelated PR.

The trigger fired; the row was not removed. This removes it.

Evidence

On origin/main the declaration is in src/v1/04_env.dag (v1.compiler.infer_env) and src/v1/05_emit_rust.dag only calls it — base and head both carry the relocation, so no comparison can produce the delta. The required run reports 0 unadjudicated delta(s), 1 stale admission(s) and fails the phase.

The failure tracks the base, not the diff:

PR	base	floor lane
#9788	bfa440b66	success
#9793	f6fa8ec76	success
#9796	6ef73334e	failure

6ef73334e is the commit that carries the relocation. Main's own pushes stay green because the phase prints NO SUBJECT on a push — the same asymmetry the sixth dissolution measured, and the reason this is invisible from the trunk.

Shape

Identical to the sixth dissolution (#9748), which retired 58 A1-R rows on the same evidence and left the roster empty. One row instead of 58 is the only difference. The roster returns to empty, and empty is not permissive: a real namespace delta still refuses as UNADJUDICATED until its author adds a row. The ledger prose above the constant is extended rather than replaced, so the seven dissolutions read as one series.

Verification

• cargo check -p v1-compiler --all-targets — clean
• cargo test -p v1-compiler --test namespace_wave_admission — 41 passed, 0 failed

That fixture is unaffected by this edit by construction: both arms build their admissions in a let with .to_string() rather than reading NAMESPACE_TRANSITION_ADMISSIONS, which its own comment records as the reason it stayed green while the production roster could not hold a row.

Found while diagnosing #9796's red floor lane; split out per ruling so the unblock carries zero unrelated diff.

🤖 Generated with Claude Code

https://claude.ai/code/session_01XXrkS4YTtkTPhEqvfSsefS

[gunbai-bot]

Superseded by #9797, which landed the identical dissolution while this was in CI — NAMESPACE_TRANSITION_ADMISSIONS is already &[] on origin/main (3474133), with its own SEVENTH DISSOLUTION paragraph in the same ledger slot.

That is the whole conflict: two paragraphs and two authors claiming one ledger entry for one fact. Merging both would leave the roster with a duplicate account of a single dissolution, so the right resolution is to drop this one rather than reconcile the prose.

Nothing here is worth salvaging into a follow-up. #9797's paragraph records the trigger, the staleness and the empty-is-not-permissive rule, which is what the ledger needs; the extras this PR carried — the three-base correlation table and the #9748 precedent link — are diagnosis, not ledger content, and they live in the diagnosis where they belong. A prose-only PR against a landed fix would be churn.

Worth noting for the record that #9797 cites a different lane's red (#9771, run 33368922338) than the one I diagnosed from (#9796). Two independent lanes hit the same trunk condition and reached the same remedy, which is the base-wide reading confirming itself.

— sent from deep-moth-436