WBML-0004 — Event Bus Runtime Suite

✅ Verified WBML-0004 68 PASS / 0 FAIL / 0 SKIP

Research question

Does the installed PoliticalWorldAPI Event Bus actually behave like the older creator source suggests, and where has runtime 1.14.0 drifted from that source?

Environment

WorldBox:          0.51.2 build 719
Git:               build-719@5dec
NeoModLoader:      1.2.0.1
PoliticalWorldAPI: 1.14.0
Lab:               0.0.4-fix2

Phase A — discovery, payload and unsubscribe

Runtime discovery returned 23 event IDs.

The suite verified:

non-null event list
known target IDs present
defensive copy behavior
custom/unknown event subscription accepted and removable
null handler rejected
duplicate exact subscription rejected/ignored
exact and wildcard delivery
party.renamed identity/kingdom/party fields
rename old/new values through OldValue/NewValue
per-subscriber payload mutation isolation
exact/wildcard unsubscribe
no callbacks after removal
subscriptions count returned to zero

The custom/unknown event result differs from the inspected API 1.9 source assumption and is preserved as versioned runtime evidence.

Phase B — callback containment

One callback intentionally threw an exception. A healthy callback subscribed to the same event still executed.

Diagnostics recorded the callback error.

The suite also verified self-unsubscribe during dispatch and UnsubscribeAll cleanup.

Phase C — recursion and stress

The recursive callback requested a chain of 64.

Runtime stopped at:

recursive=16
tail=16

The tail subscriber receiving the same 16 accepted events is important: the recursion guard did not turn the dispatch into a completely broken stream.

Then 100 sequential test writes produced:

100/100 accepted
100 callbacks
33 ms observed

Only the counts are part of the behavioral verification. The timing is not a universal benchmark.

Harness corrections

Initial harness

It required unknown-event rejection and required rename values in OldName/NewName. Runtime disagreed.

Those failures became evidence that the harness assumptions were stale.

fix1

The payload mapping was corrected, but recursive/stress handlers still filtered only NewName. Runtime 1.14.0 placed the tested value in NewValue, so the Lab itself dropped every Phase C event.

Zero callbacks from that run were rejected as harness evidence.

fix2

The filter resolved rename new value using runtime-aware mapping. The full suite then passed.

Final result

PASS=68 FAIL=0 SKIP=0
A=PASS B=PASS C=PASS
SUITE RESULT: PASS

Original party name restored and final subscriptions reached zero.

Evidence boundary

This verifies the tested dispatch paths, not every payload field of all 23 event IDs.

Sanitized runtime excerpt