Design 47 H: H-A-343, the runner's events are waited on by name

This commit is contained in:
2026-11-21 22:11:53 +01:00
parent f2dfc50f8d
commit 47515b8730
@@ -2117,3 +2117,10 @@ Where two preparers disagreed, the chosen one is named.
before it starts the new one; a bundle's call that outlives the bound is cut and said. No repository fix for this
instance; nothing to rerun (main.155 carries it). P-NATS-native, P-failure-loud, fix-a-bug-met-twice (the restart
loop of row 179 killed calls the same way).
- **H-A-343 (the runner's events are waited on by name, 22:11)** A session waiting on a pipeline's end could not: the
mesh MCP server's wait refuses "mesh-pipeline-runner.pipeline-ended" because the catalog's half names neither a
seat nor a module of that name for the event, so the watcher read the journal on a timer. H2 line: every event the
runner says (pipeline-started, step-ended, pipeline-ended, the re-pin) is declared in its seat's definition under
the seat's events family, so a wait by seat and event name is answered and nobody polls; the same for the builder's
and the indexer's events (ask-taken, version-reserved, H-A-338 (4)). Its check: the seat list carries the family;
mesh_wait on each name is answered in the whole-module test. P-NATS-native, map-onto-how-the-mesh-is-built.