<?xml version="1.0" encoding="utf-8"?>
<entry xmlns="http://www.w3.org/2005/Atom" xml:base="https://docs.spec-kitty.ai/adr/3.x/2026-09-06-2-runtime-event-emitter-disposition/">
  <id>https://docs.spec-kitty.ai/adr/3.x/2026-09-06-2-runtime-event-emitter-disposition/</id>
  <title>ADR: RuntimeEventEmitter Seam Disposition — Rewire-Ready Consolidation, Not Retirement</title>
  <updated>2026-09-06T00:00:00.000Z</updated>
  <author><name>Spec Kitty</name></author>
  <link rel="alternate" type="text/html" href="https://docs.spec-kitty.ai/adr/3.x/2026-09-06-2-runtime-event-emitter-disposition/"/>
  <summary>Keeps the reserved E3 seam behind a factory, deletes the duplicate runtime.next.event_emitter class, fixes the flush-target bug, and bounds Mission B scope.</summary>
  <content type="html">&lt;h2 id=&quot;context-and-problem-statement&quot;&gt;Context and Problem Statement&lt;/h2&gt;
&lt;p&gt;&lt;code dir=&quot;auto&quot;&gt;src/runtime/next/event_emitter.py&lt;/code&gt; defines a concrete class &lt;code dir=&quot;auto&quot;&gt;RuntimeEventEmitter&lt;/code&gt;
whose eight &lt;code dir=&quot;auto&quot;&gt;emit_*&lt;/code&gt; callbacks are permanently no-op (&lt;code dir=&quot;auto&quot;&gt;pass&lt;/code&gt; / &lt;code dir=&quot;auto&quot;&gt;del payload&lt;/code&gt;,
&lt;code dir=&quot;auto&quot;&gt;event_emitter.py:65-88&lt;/code&gt;) and whose &lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt; is an explicit no-op
(&lt;code dir=&quot;auto&quot;&gt;:60-61&lt;/code&gt;). It is the residue of the retired sync &lt;code dir=&quot;auto&quot;&gt;EventEmitter&lt;/code&gt; (issue #5): the
journal/outbox producer died with the sync transport, and what remained is the
&lt;em&gt;seam&lt;/em&gt; — a call surface the &lt;code dir=&quot;auto&quot;&gt;next&lt;/code&gt; bridge already threads so a future producer
can register here without reshaping the bridge.&lt;/p&gt;
&lt;p&gt;The dead-code census (report 25 §1a) flagged this class as &lt;strong&gt;DORMANT with a stale
justification&lt;/strong&gt; and leaned toward retiring it — replacing the construction with
&lt;code dir=&quot;auto&quot;&gt;_internal_runtime.events.NullEmitter&lt;/code&gt;, retyping the signatures against the
Protocol of the same name, and deleting the file (88 LOC, plus the bridge threading).
The design red-team (report 24 §4) challenged that lean as &lt;strong&gt;wrong-headed&lt;/strong&gt; and
recommended REWIRE-or-explicit-defer instead.&lt;/p&gt;
&lt;p&gt;This decision matters because the disposition is a prerequisite for the &lt;strong&gt;Mission B
&lt;code dir=&quot;auto&quot;&gt;dead-port-disposition&lt;/code&gt;&lt;/strong&gt; work: WP01 of that mission acts on this seam, and it must
know whether it is deleting a dead port or consolidating a reserved one. Getting it
wrong in either direction is costly — a naive deletion throws away a designed last
mile and silently changes rollback behavior; leaving it untouched preserves a
name collision, a stale docstring, and a latent correctness bug.&lt;/p&gt;
&lt;p&gt;The following facts were verified against &lt;code dir=&quot;auto&quot;&gt;src/&lt;/code&gt; at HEAD &lt;code dir=&quot;auto&quot;&gt;e721763759&lt;/code&gt; (3.2.7rc1)
and against the installed &lt;code dir=&quot;auto&quot;&gt;spec_kitty_events&lt;/code&gt; 9.1.6 package:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;The hosted contract already exists and is waiting for a producer.&lt;/strong&gt; Installed
&lt;code dir=&quot;auto&quot;&gt;spec_kitty_events&lt;/code&gt; 9.1.6 carries all six &lt;code dir=&quot;auto&quot;&gt;mission_next&lt;/code&gt; runtime moments —
&lt;code dir=&quot;auto&quot;&gt;MissionRunStarted&lt;/code&gt;, &lt;code dir=&quot;auto&quot;&gt;NextStepIssued&lt;/code&gt;, &lt;code dir=&quot;auto&quot;&gt;NextStepAutoCompleted&lt;/code&gt;,
&lt;code dir=&quot;auto&quot;&gt;DecisionInputRequested&lt;/code&gt;, &lt;code dir=&quot;auto&quot;&gt;DecisionInputAnswered&lt;/code&gt;, &lt;code dir=&quot;auto&quot;&gt;MissionRunCompleted&lt;/code&gt; — in
&lt;code dir=&quot;auto&quot;&gt;zeitgeist_attrs.VOLATILE_EVENT_TYPES&lt;/code&gt; (verified: 6/6 present), with codec and
per-type redaction entries. The seam is the designed last mile of that contract,
not speculative 4.0.0 vocabulary.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;The seam is explicitly reserved in-tree.&lt;/strong&gt; Two conformance tests name
&lt;code dir=&quot;auto&quot;&gt;runtime.next._internal_runtime.events&lt;/code&gt; (post-decision: this ADR’s own (d)
deleted &lt;code dir=&quot;auto&quot;&gt;runtime.next.event_emitter&lt;/code&gt;, and both tests were repointed at the
consolidated module in the same change) as the reserved E3 producer seam
and say its payloads “belong back in this file” when a real emitter is
wired (&lt;code dir=&quot;auto&quot;&gt;tests/status/test_producer_conformance.py&lt;/code&gt;,
&lt;code dir=&quot;auto&quot;&gt;tests/contract/test_identity_contract_matrix.py&lt;/code&gt;).&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;The retention docstring is stale.&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;event_emitter.py:1-10&lt;/code&gt; claims E3 will
“register a real handler (the zeitgeist moment fan-out) at this seam.” The
zeitgeist moment fan-out already landed at a &lt;em&gt;different&lt;/em&gt; seam —
&lt;code dir=&quot;auto&quot;&gt;src/specify_cli/status/adapters.py:364-365&lt;/code&gt;
(&lt;code dir=&quot;auto&quot;&gt;ensure_zeitgeist_moment_handlers()&lt;/code&gt;, gated by
&lt;code dir=&quot;auto&quot;&gt;SPEC_KITTY_SYNC_MINIMAL_IMPORT&lt;/code&gt;). The docstring points E3 at a seam E3 already
bypassed.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;There is a name collision.&lt;/strong&gt; A &lt;em&gt;second, live&lt;/em&gt; class named
&lt;code dir=&quot;auto&quot;&gt;RuntimeEventEmitter&lt;/code&gt; — a &lt;code dir=&quot;auto&quot;&gt;Protocol&lt;/code&gt; — exists at
&lt;code dir=&quot;auto&quot;&gt;src/runtime/next/_internal_runtime/events.py:67-88&lt;/code&gt;, alongside a concrete
&lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt; (&lt;code dir=&quot;auto&quot;&gt;:95-123&lt;/code&gt;) and real structural adapters (&lt;code dir=&quot;auto&quot;&gt;DecisionGitLog&lt;/code&gt;,
&lt;code dir=&quot;auto&quot;&gt;_BufferingRuntimeEmitter&lt;/code&gt;). Two classes with one name in the same subtree is
the duplication smell the census (report 25 §1f) named.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;The concrete class carries surface the Protocol/NullEmitter lack.&lt;/strong&gt; The bridge
binds the &lt;em&gt;concrete&lt;/em&gt; class: it constructs it via &lt;code dir=&quot;auto&quot;&gt;RuntimeEventEmitter.for_feature(...)&lt;/code&gt;
at &lt;code dir=&quot;auto&quot;&gt;runtime_bridge.py:1552&lt;/code&gt; and &lt;code dir=&quot;auto&quot;&gt;:2739&lt;/code&gt;, and calls &lt;code dir=&quot;auto&quot;&gt;sync_emitter.seed_from_snapshot(...)&lt;/code&gt;
at &lt;code dir=&quot;auto&quot;&gt;:1614&lt;/code&gt; and &lt;code dir=&quot;auto&quot;&gt;:2745&lt;/code&gt;. Neither &lt;code dir=&quot;auto&quot;&gt;for_feature&lt;/code&gt; nor &lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt; exists on
the &lt;code dir=&quot;auto&quot;&gt;_internal_runtime&lt;/code&gt; Protocol &lt;em&gt;or&lt;/em&gt; on &lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt;. A “just swap in NullEmitter”
retirement is therefore &lt;strong&gt;not a drop-in&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;A latent correctness bug lives on this chain.&lt;/strong&gt; Under the strict retrospective
gate, &lt;code dir=&quot;auto&quot;&gt;engine_emitter&lt;/code&gt; is replaced by a &lt;code dir=&quot;auto&quot;&gt;_BufferingRuntimeEmitter&lt;/code&gt;
(&lt;code dir=&quot;auto&quot;&gt;runtime_bridge.py:2149-2150&lt;/code&gt;) so a rolled-back terminal advance cannot fire an
unretractable &lt;code dir=&quot;auto&quot;&gt;MissionRunCompleted&lt;/code&gt;. On gate-pass the buffer is flushed with
&lt;code dir=&quot;auto&quot;&gt;buffer.flush(ctx.sync_emitter)&lt;/code&gt; (&lt;code dir=&quot;auto&quot;&gt;:2187&lt;/code&gt;) — into the &lt;strong&gt;plain no-op&lt;/strong&gt;
&lt;code dir=&quot;auto&quot;&gt;sync_emitter&lt;/code&gt;, &lt;strong&gt;not&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;ctx.emitter_for_engine&lt;/code&gt;, the &lt;code dir=&quot;auto&quot;&gt;DecisionGitLog&lt;/code&gt;-wrapped
emitter the non-gated path uses (&lt;code dir=&quot;auto&quot;&gt;:2124-2126&lt;/code&gt;, wrapped at &lt;code dir=&quot;auto&quot;&gt;:1560&lt;/code&gt;). The buffer is
installed for &lt;strong&gt;every&lt;/strong&gt; advance under strict policy (&lt;code dir=&quot;auto&quot;&gt;:2129&lt;/code&gt;), not only terminal
ones, and &lt;code dir=&quot;auto&quot;&gt;DecisionGitLog&lt;/code&gt; persists only &lt;code dir=&quot;auto&quot;&gt;DecisionInputRequested&lt;/code&gt; /
&lt;code dir=&quot;auto&quot;&gt;DecisionInputAnswered&lt;/code&gt; (&lt;code dir=&quot;auto&quot;&gt;src/specify_cli/events/decision_log.py:147-160&lt;/code&gt;; every
other emit is pass-through). In the engine &lt;code dir=&quot;auto&quot;&gt;decision_required&lt;/code&gt; and &lt;code dir=&quot;auto&quot;&gt;terminal&lt;/code&gt; are
mutually exclusive branches (&lt;code dir=&quot;auto&quot;&gt;_internal_runtime/engine.py:448,476&lt;/code&gt;), so a terminal
advance buffers only &lt;code dir=&quot;auto&quot;&gt;MissionRunCompleted&lt;/code&gt;, which the git log ignores anyway.
Consequence: under strict policy, every &lt;code dir=&quot;auto&quot;&gt;decision_required&lt;/code&gt; advance through the
legacy &lt;code dir=&quot;auto&quot;&gt;runtime_next_step&lt;/code&gt; path buffers its &lt;code dir=&quot;auto&quot;&gt;DecisionInputRequested&lt;/code&gt; and flushes it
into the no-op — &lt;code dir=&quot;auto&quot;&gt;decisions.events.jsonl&lt;/code&gt; is never appended — whereas on the
non-gated path the same event reaches the coordination-branch decision git log
(via &lt;code dir=&quot;auto&quot;&gt;emitter_for_engine&lt;/code&gt;). Today this is masked because &lt;code dir=&quot;auto&quot;&gt;sync_emitter&lt;/code&gt; is no-op
and &lt;code dir=&quot;auto&quot;&gt;DecisionGitLog&lt;/code&gt; is the only real channel — but any disposition that touches
this chain will either silently freeze or silently fix it, so it must be
adjudicated explicitly.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;A second bypass sits on the composition dispatch path.&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;runtime_bridge.py:1976&lt;/code&gt;
passes the plain &lt;code dir=&quot;auto&quot;&gt;ctx.sync_emitter&lt;/code&gt; (not &lt;code dir=&quot;auto&quot;&gt;emitter_for_engine&lt;/code&gt;) into
&lt;code dir=&quot;auto&quot;&gt;_advance_run_state_after_composition&lt;/code&gt;, whose &lt;code dir=&quot;auto&quot;&gt;_emit_decision_required&lt;/code&gt; calls
&lt;code dir=&quot;auto&quot;&gt;sync_emitter.emit_decision_input_requested(payload)&lt;/code&gt;
(&lt;code dir=&quot;auto&quot;&gt;runtime_bridge_engine.py:226&lt;/code&gt;). Decision events on that path never reach
&lt;code dir=&quot;auto&quot;&gt;DecisionGitLog&lt;/code&gt; regardless of retrospective policy. Same defect class; named here
so the Mission B implementer does not fix it silently under cover of the retype.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id=&quot;decision-drivers&quot;&gt;Decision Drivers&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Do not discard a reserved, contract-backed seam.&lt;/strong&gt; The hosted vocabulary is
provisioned and the seam is explicitly reserved by tests; retiring it means
re-adding ~29 &lt;code dir=&quot;auto&quot;&gt;sync_emitter&lt;/code&gt; threading sites plus two construction sites when E3
lands.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Do not overreach into feature work.&lt;/strong&gt; Wiring a &lt;em&gt;live&lt;/em&gt; zeitgeist producer is E3
work with hosted-egress and redaction implications; a disposition ADR (and a
dead-port-disposition mission) must not smuggle that in.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Resolve the honest defects the census surfaced&lt;/strong&gt; — the name collision, the
stale docstring, and the flush-target bug — rather than freezing them under a
“leave it alone” decision.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Preserve behavior for every current caller.&lt;/strong&gt; The only live behavior on this
chain is &lt;code dir=&quot;auto&quot;&gt;DecisionGitLog&lt;/code&gt; wrapping the no-op; a consolidation must keep decision
events durably committed, and must not add commit semantics to plain-door callers.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Give Mission B a fail-closed boundary&lt;/strong&gt; so WP01 knows exactly what is safe to
change and what is ADR-blocked.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id=&quot;considered-options&quot;&gt;Considered Options&lt;/h2&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Rewire-ready consolidation (chosen)&lt;/strong&gt; — keep the seam; merge the duplicate
concrete class into the &lt;code dir=&quot;auto&quot;&gt;_internal_runtime&lt;/code&gt; Protocol/NullEmitter world (promote
&lt;code dir=&quot;auto&quot;&gt;for_feature&lt;/code&gt; + &lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt; onto the seam surface, delete
&lt;code dir=&quot;auto&quot;&gt;event_emitter.py&lt;/code&gt;, retype the bridge against the single Protocol); fix the
flush target; correct the docstring. Do &lt;strong&gt;not&lt;/strong&gt; wire a live producer — defer
that to E3.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Retire now&lt;/strong&gt; — delete the seam, collapse construction onto &lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt;,
drop &lt;code dir=&quot;auto&quot;&gt;for_feature&lt;/code&gt;/&lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt;, remove the threading.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Status quo&lt;/strong&gt; — keep both classes, the stale docstring, and the latent bug
as-is.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Rewire fully now&lt;/strong&gt; — wire a live zeitgeist-moment producer at the seam
immediately. &lt;em&gt;Out of scope&lt;/em&gt; (this is E3 feature work); listed only to name it.&lt;/li&gt;
&lt;/ol&gt;
&lt;h2 id=&quot;decision-outcome&quot;&gt;Decision Outcome&lt;/h2&gt;
&lt;p&gt;&lt;strong&gt;Chosen option: “Rewire-ready consolidation” (Option 1)&lt;/strong&gt;, because it is the only
option that (a) honors the reserved, contract-backed seam, (b) removes the name
collision and stale docstring the census flagged, (c) forces the latent flush-target
bug to be adjudicated rather than silently frozen, and (d) keeps live-producer
feature work (E3) cleanly deferred. It is a &lt;strong&gt;rename-and-merge plus a bug fix&lt;/strong&gt;, not
a port deletion and not feature work.&lt;/p&gt;
&lt;p&gt;Concretely, the ADR adjudicates the five questions Mission B WP01 must answer:&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;(a) Rewire target + registry mechanism.&lt;/strong&gt; The canonical interface is the
&lt;code dir=&quot;auto&quot;&gt;_internal_runtime/events.py&lt;/code&gt; &lt;code dir=&quot;auto&quot;&gt;RuntimeEventEmitter&lt;/code&gt; &lt;strong&gt;Protocol&lt;/strong&gt;; the seam’s
null-object is &lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt;. The construction sites (&lt;code dir=&quot;auto&quot;&gt;runtime_bridge.py:1552,2739&lt;/code&gt;)
become the &lt;em&gt;registry seam&lt;/em&gt;: a factory that returns &lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt; by default (and
under &lt;code dir=&quot;auto&quot;&gt;SPEC_KITTY_SYNC_MINIMAL_IMPORT&lt;/code&gt;) and, when E3 registers one, a real
Protocol-conforming producer — mirroring the env-gated import-tail registration
pattern proven at &lt;code dir=&quot;auto&quot;&gt;status/adapters.py:364-365&lt;/code&gt;
(&lt;code dir=&quot;auto&quot;&gt;ensure_zeitgeist_moment_handlers()&lt;/code&gt;). The bridge depends on the Protocol and a
factory, never on a concrete class import.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;(b) The Protocol’s missing surface.&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;for_feature&lt;/code&gt; (mission-identity resolution,
the one thing a future adapter needs — &lt;code dir=&quot;auto&quot;&gt;runtime_bridge.py:1552,2739&lt;/code&gt;) and
&lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt; (&lt;code dir=&quot;auto&quot;&gt;:1614,2745&lt;/code&gt;) must be added to the seam before/as the
concrete class is deleted. Preferred shape: a &lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt; classmethod (or a
module-level factory) and &lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt; as a no-op/​pass-through on
&lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt;, so the bridge’s two construction sites and two seed sites keep
compiling against the consolidated seam. &lt;strong&gt;Name the promoted constructor
&lt;code dir=&quot;auto&quot;&gt;for_mission&lt;/code&gt;&lt;/strong&gt;, not &lt;code dir=&quot;auto&quot;&gt;for_feature&lt;/code&gt;: the Terminology Canon forbids minting new
&lt;code dir=&quot;auto&quot;&gt;feature*&lt;/code&gt; aliases for the Mission domain object, and the consolidated seam is a
new public surface. Keep &lt;code dir=&quot;auto&quot;&gt;for_feature&lt;/code&gt; only as a transitional alias with a removal
note, if the two construction sites cannot be renamed in the same change.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;(c) The buffer flush-target bug.&lt;/strong&gt; It is real and must be &lt;strong&gt;fixed, not frozen&lt;/strong&gt;.
&lt;code dir=&quot;auto&quot;&gt;buffer.flush(...)&lt;/code&gt; (&lt;code dir=&quot;auto&quot;&gt;runtime_bridge.py:2187&lt;/code&gt;) must target the durable decision
sink the non-gated path uses — &lt;code dir=&quot;auto&quot;&gt;ctx.emitter_for_engine&lt;/code&gt; (the &lt;code dir=&quot;auto&quot;&gt;DecisionGitLog&lt;/code&gt;
wrap, &lt;code dir=&quot;auto&quot;&gt;:2124-2126&lt;/code&gt;) — not the plain no-op &lt;code dir=&quot;auto&quot;&gt;ctx.sync_emitter&lt;/code&gt;. The fix is a coupled
correctness change and must land with a red-first test whose fixture is a
&lt;strong&gt;strict-policy &lt;code dir=&quot;auto&quot;&gt;decision_required&lt;/code&gt; advance&lt;/strong&gt; (the only kind that buffers a
git-loggable event): it must prove the buffered &lt;code dir=&quot;auto&quot;&gt;DecisionInputRequested&lt;/code&gt; is appended
to &lt;code dir=&quot;auto&quot;&gt;decisions.events.jsonl&lt;/code&gt; after the flush. A terminal-advance fixture cannot go red
because the git log ignores &lt;code dir=&quot;auto&quot;&gt;MissionRunCompleted&lt;/code&gt;; keep the terminal/rollback scenario
as a separate assertion that a gate refusal discards the buffer and writes nothing.
(There is no double-emit risk: on the gated path the engine wrote only into the
buffer, so the flush is the first and only &lt;code dir=&quot;auto&quot;&gt;DecisionGitLog&lt;/code&gt; write. Rollback is
preserved: the gate at &lt;code dir=&quot;auto&quot;&gt;:2178-2181&lt;/code&gt; runs before the flush and the buffer is one-shot,
&lt;code dir=&quot;auto&quot;&gt;runtime_bridge_retrospective.py:138-149&lt;/code&gt;.) The composition-path bypass
(&lt;code dir=&quot;auto&quot;&gt;:1976&lt;/code&gt; → &lt;code dir=&quot;auto&quot;&gt;runtime_bridge_engine.py:226&lt;/code&gt;) is the same defect class and is &lt;strong&gt;in
scope for WP01 with its own red-first test&lt;/strong&gt;; fixing one and freezing the other is
prohibited.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;(d) The name collision.&lt;/strong&gt; Resolve it by &lt;strong&gt;removing the duplicate&lt;/strong&gt;: delete the
concrete &lt;code dir=&quot;auto&quot;&gt;event_emitter.py::RuntimeEventEmitter&lt;/code&gt; and keep the single canonical
&lt;code dir=&quot;auto&quot;&gt;RuntimeEventEmitter&lt;/code&gt; Protocol in &lt;code dir=&quot;auto&quot;&gt;_internal_runtime/events.py&lt;/code&gt;. (If a transitional
step must keep both classes momentarily, the concrete one is renamed
&lt;code dir=&quot;auto&quot;&gt;NullRuntimeEventEmitter&lt;/code&gt; — but the target state has one name.)&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;(e) Mission B scoping.&lt;/strong&gt; See Consequences.&lt;/p&gt;
&lt;h3 id=&quot;consequences&quot;&gt;Consequences&lt;/h3&gt;
&lt;h4 id=&quot;positive&quot;&gt;Positive&lt;/h4&gt;
&lt;ul&gt;
&lt;li&gt;The reserved E3 seam and its constructor (shipped as &lt;code dir=&quot;auto&quot;&gt;for_mission&lt;/code&gt;, per (b)) and
&lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt; capability
survive; a future producer registers an adapter without re-threading the bridge.&lt;/li&gt;
&lt;li&gt;One class named &lt;code dir=&quot;auto&quot;&gt;RuntimeEventEmitter&lt;/code&gt; remains; the duplication smell is gone.&lt;/li&gt;
&lt;li&gt;The stale docstring is corrected to point at &lt;code dir=&quot;auto&quot;&gt;status/adapters.py&lt;/code&gt; (the real
zeitgeist seam), so the next agent is not misdirected.&lt;/li&gt;
&lt;li&gt;A latent correctness bug (strict-gated decision events skipping the git log) is
fixed with a regression test instead of being silently frozen.&lt;/li&gt;
&lt;li&gt;&lt;del&gt;Net LOC still drops (~88 LOC of concrete class + collapsed threading), so the
census’s simplification goal is largely met — just via merge, not deletion.&lt;/del&gt; &lt;strong&gt;Amended 2026-09-06 after merge (PR #3921):&lt;/strong&gt; this estimate was wrong. The factory, registry, &lt;code dir=&quot;auto&quot;&gt;for_mission&lt;/code&gt;, and seam docstring the decision itself requires add +120 lines to &lt;code dir=&quot;auto&quot;&gt;_internal_runtime/events.py&lt;/code&gt;; the mission’s net under &lt;code dir=&quot;auto&quot;&gt;src/runtime/next/&lt;/code&gt; is +161/−105 = +56. The structural goals (one seam class, no collision, no dead duplicate) are met; the LOC reduction is not, and E3 should not expect one from wiring (#3929).&lt;/li&gt;
&lt;/ul&gt;
&lt;h4 id=&quot;negative&quot;&gt;Negative&lt;/h4&gt;
&lt;ul&gt;
&lt;li&gt;More work than a straight &lt;code dir=&quot;auto&quot;&gt;git rm&lt;/code&gt;: the merge touches &lt;code dir=&quot;auto&quot;&gt;runtime_bridge.py&lt;/code&gt;,
&lt;code dir=&quot;auto&quot;&gt;runtime_bridge_engine.py&lt;/code&gt;, &lt;code dir=&quot;auto&quot;&gt;_internal_runtime/events.py&lt;/code&gt;, the two
conformance tests’ comments, and the one live importer of the concrete class
(&lt;code dir=&quot;auto&quot;&gt;tests/specify_cli/events/test_decision_log_coord.py:17&lt;/code&gt;), and adds red-first
tests for the two flush-target fixes.&lt;/li&gt;
&lt;li&gt;The flush-target fix is a behavior change (decision events now durably commit on
the strict-gated path). It is a &lt;em&gt;correctness&lt;/em&gt; change, but it must be called out
and tested, not slipped in.&lt;/li&gt;
&lt;li&gt;The seam remains no-op end-to-end until E3; this ADR does not deliver a live
producer, so the “waiting for a producer” state persists (by design).&lt;/li&gt;
&lt;/ul&gt;
&lt;h4 id=&quot;neutral&quot;&gt;Neutral&lt;/h4&gt;
&lt;ul&gt;
&lt;li&gt;The zeitgeist moment fan-out at &lt;code dir=&quot;auto&quot;&gt;status/adapters.py:364&lt;/code&gt; is unaffected — it is a
separate, already-live seam; this decision only stops the emitter docstring from
claiming to be it.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id=&quot;mission-b-dead-port-disposition-scope-boundary&quot;&gt;Mission B (&lt;code dir=&quot;auto&quot;&gt;dead-port-disposition&lt;/code&gt;) scope boundary&lt;/h3&gt;
&lt;p&gt;&lt;strong&gt;ADR-BLOCKED — Mission B must NOT:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Delete &lt;code dir=&quot;auto&quot;&gt;event_emitter.py&lt;/code&gt; as a &lt;em&gt;pure port deletion&lt;/em&gt; / “collapse onto NullEmitter
and drop the seam.”&lt;/li&gt;
&lt;li&gt;Remove the constructor (&lt;code dir=&quot;auto&quot;&gt;for_mission&lt;/code&gt;) / &lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt; capability.&lt;/li&gt;
&lt;li&gt;Change the buffer flush semantics without fixing the flush target (freezing the
bug is prohibited).&lt;/li&gt;
&lt;li&gt;Wire a live zeitgeist producer (that is E3, out of scope).&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;strong&gt;SAFE under this ADR — Mission B WP01 may:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Perform the rename-and-merge consolidation (behavior-preserving apart from the
flush fix): promote the constructor (as &lt;code dir=&quot;auto&quot;&gt;for_mission&lt;/code&gt;) and &lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt;
onto the seam, delete the duplicate concrete class, retype the bridge against the
Protocol + factory.&lt;/li&gt;
&lt;li&gt;Fix the flush target at &lt;code dir=&quot;auto&quot;&gt;:2187&lt;/code&gt; to &lt;code dir=&quot;auto&quot;&gt;ctx.emitter_for_engine&lt;/code&gt;, and the composition
dispatch emitter at &lt;code dir=&quot;auto&quot;&gt;:1976&lt;/code&gt; likewise, each with a red-first regression test whose
fixture is a strict-policy &lt;code dir=&quot;auto&quot;&gt;decision_required&lt;/code&gt; advance.&lt;/li&gt;
&lt;li&gt;Correct the &lt;code dir=&quot;auto&quot;&gt;event_emitter.py&lt;/code&gt; docstring (or its successor’s) and update the two
conformance tests’ comments to point at the consolidated seam.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id=&quot;confirmation&quot;&gt;Confirmation&lt;/h3&gt;
&lt;p&gt;This decision is validated when Mission B WP01 lands a change in which: (1) exactly
one class named &lt;code dir=&quot;auto&quot;&gt;RuntimeEventEmitter&lt;/code&gt; exists under &lt;code dir=&quot;auto&quot;&gt;src/runtime/next/&lt;/code&gt;; (2) the
bridge constructs the seam via a factory returning &lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt; by default with
the constructor/&lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt; capability intact; (3) new regression tests
prove that a strict-policy &lt;code dir=&quot;auto&quot;&gt;decision_required&lt;/code&gt; advance appends its buffered
&lt;code dir=&quot;auto&quot;&gt;DecisionInputRequested&lt;/code&gt; to the decision git log after the flush, that the same holds
on the composition dispatch path, and that a refused terminal gate discards the buffer
without a git write; and (4) the existing bridge-parity and producer-conformance
tests stay green. Confidence: &lt;strong&gt;high&lt;/strong&gt; on the facts (all verified at
&lt;code dir=&quot;auto&quot;&gt;e721763759&lt;/code&gt; against installed &lt;code dir=&quot;auto&quot;&gt;spec_kitty_events&lt;/code&gt; 9.1.6); &lt;strong&gt;medium&lt;/strong&gt; on the exact
consolidation mechanics, which are Mission B’s to finalize within this boundary.&lt;/p&gt;
&lt;h2 id=&quot;pros-and-cons-of-the-options&quot;&gt;Pros and Cons of the Options&lt;/h2&gt;
&lt;h3 id=&quot;option-1--rewire-ready-consolidation-chosen&quot;&gt;Option 1 — Rewire-ready consolidation (chosen)&lt;/h3&gt;
&lt;p&gt;Keep the seam, merge the duplicate class into the Protocol/NullEmitter, fix the
flush target, correct the docstring; defer the live producer.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Pros:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Preserves a reserved, contract-backed seam and its identity-resolution surface.&lt;/li&gt;
&lt;li&gt;Eliminates the name collision and the stale docstring.&lt;/li&gt;
&lt;li&gt;Fixes the latent flush-target bug under test.&lt;/li&gt;
&lt;li&gt;&lt;del&gt;Still achieves most of the census’s LOC reduction.&lt;/del&gt; (Amended 2026-09-06: it does not — see Consequences.)&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;strong&gt;Cons:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;More involved than a deletion; touches several bridge files and adds a test.&lt;/li&gt;
&lt;li&gt;Introduces a (correct, tested) behavior change on the strict-gated path.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id=&quot;option-2--retire-now&quot;&gt;Option 2 — Retire now&lt;/h3&gt;
&lt;p&gt;Delete &lt;code dir=&quot;auto&quot;&gt;event_emitter.py&lt;/code&gt;, collapse onto &lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt;, drop the threading.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Pros:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Maximal immediate simplification; smallest diff to &lt;em&gt;count&lt;/em&gt;.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;strong&gt;Cons:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Not a drop-in: &lt;code dir=&quot;auto&quot;&gt;for_feature&lt;/code&gt;/&lt;code dir=&quot;auto&quot;&gt;seed_from_snapshot&lt;/code&gt; are absent from the Protocol
and &lt;code dir=&quot;auto&quot;&gt;NullEmitter&lt;/code&gt;, so the bridge would need surgery anyway.&lt;/li&gt;
&lt;li&gt;Throws away a reserved seam the hosted &lt;code dir=&quot;auto&quot;&gt;spec_kitty_events&lt;/code&gt; 9.1.6 contract is
waiting on; E3 must re-thread ~29 sites + 2 construction sites later.&lt;/li&gt;
&lt;li&gt;Silently &lt;em&gt;freezes&lt;/em&gt; the flush-target bug (or, if it re-points construction,
silently changes it) without adjudication or a test.&lt;/li&gt;
&lt;li&gt;Contradicts two in-tree conformance tests that reserve the seam.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id=&quot;option-3--status-quo&quot;&gt;Option 3 — Status quo&lt;/h3&gt;
&lt;p&gt;Leave both classes, the stale docstring, and the bug in place.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Pros:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Zero immediate risk; no diff.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;strong&gt;Cons:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Preserves the two-classes-one-name collision the census flagged.&lt;/li&gt;
&lt;li&gt;Keeps a docstring that actively misdirects the next agent about the E3 seam.&lt;/li&gt;
&lt;li&gt;Leaves a real correctness bug (strict-gated decision events skipping the git
log) latent and undocumented.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id=&quot;option-4--rewire-fully-now-out-of-scope&quot;&gt;Option 4 — Rewire fully now (out of scope)&lt;/h3&gt;
&lt;p&gt;Wire a live zeitgeist-moment producer at the seam immediately.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Pros:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Delivers the end-to-end runtime-moment fan-out.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;strong&gt;Cons:&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Feature work (E3) with hosted-egress, redaction, and ordering concerns — not a
disposition, and outside a dead-port-disposition mission. Named only to exclude
it.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id=&quot;more-information&quot;&gt;More Information&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;Evidence: report 24 (&lt;code dir=&quot;auto&quot;&gt;work/post-convergence/24-coreloop-design-redteam-alphonso.md&lt;/code&gt;,
§4 port dispositions) and report 25 (&lt;code dir=&quot;auto&quot;&gt;25-dead-code-census-randy.md&lt;/code&gt;, §1a/§1f/§4).
Both are local, untracked (&lt;code dir=&quot;auto&quot;&gt;work/&lt;/code&gt; is gitignored); every load-bearing fact above was
re-verified in-tree and cites its code anchor, so the decision does not rest on them.&lt;/li&gt;
&lt;li&gt;Related decision: &lt;a href=&quot;/adr/3.x/2026-04-25-1-shared-package-boundary/&quot;&gt;&lt;code dir=&quot;auto&quot;&gt;2026-04-25-1-shared-package-boundary.md&lt;/code&gt;&lt;/a&gt;
(the &lt;code dir=&quot;auto&quot;&gt;_internal_runtime&lt;/code&gt; internalization that produced the second seam).&lt;/li&gt;
&lt;li&gt;Code anchors (HEAD &lt;code dir=&quot;auto&quot;&gt;e721763759&lt;/code&gt;): &lt;code dir=&quot;auto&quot;&gt;src/runtime/next/event_emitter.py&lt;/code&gt;;
&lt;code dir=&quot;auto&quot;&gt;src/runtime/next/_internal_runtime/events.py:67-123&lt;/code&gt;;
&lt;code dir=&quot;auto&quot;&gt;src/runtime/next/runtime_bridge.py:195,1215,1472,1552,1560,1614,2124-2126,2149-2150,2187,2739,2745,2754&lt;/code&gt;;
&lt;code dir=&quot;auto&quot;&gt;src/runtime/next/runtime_bridge_retrospective.py:69-149&lt;/code&gt;;
&lt;code dir=&quot;auto&quot;&gt;src/specify_cli/status/adapters.py:364-365&lt;/code&gt;; installed
&lt;code dir=&quot;auto&quot;&gt;spec_kitty_events/zeitgeist_attrs.py&lt;/code&gt; (&lt;code dir=&quot;auto&quot;&gt;VOLATILE_EVENT_TYPES&lt;/code&gt;);
&lt;code dir=&quot;auto&quot;&gt;tests/status/test_producer_conformance.py&lt;/code&gt;,
&lt;code dir=&quot;auto&quot;&gt;tests/contract/test_identity_contract_matrix.py&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;Consumed by: the planned Mission B &lt;code dir=&quot;auto&quot;&gt;dead-port-disposition&lt;/code&gt; (WP01; not yet minted).
Sibling governance PR: #3888.&lt;/li&gt;
&lt;li&gt;Accepted by the operator on 2026-09-06 (PR #3898): Option 1, rewire-ready consolidation.
Codemap retirement in the same PR confirmed. Consequences amended post-merge
(PR #3921): the net-LOC claim was withdrawn. Amended the same day after a
pre-merge adversarial squad (architect / debugger / reviewer lenses): bug restated
onto &lt;code dir=&quot;auto&quot;&gt;decision_required&lt;/code&gt; advances, confirmation test made satisfiable, composition
path brought in scope, promoted constructor renamed &lt;code dir=&quot;auto&quot;&gt;for_mission&lt;/code&gt;.&lt;/li&gt;
&lt;/ul&gt;</content>
</entry>
