W16.2: mesh-config's whole test says the catalog-revision event on the branch's part, as the bus says it

Builds: design 47 G §532, §614; lead answer A-101

Claude-Session: https://claude.ai/code/session_01NcbgfGKaxw984XyAPfLm4h
This commit is contained in:
2026-11-11 23:34:51 +01:00
parent da5bbe0a17
commit e0428bd874
@@ -609,9 +609,8 @@ func (p *player) manifestIn(module, raw string) {
rev := p.catRev
p.mu.Unlock()
p.hearsCatalog()
// The key is the subscribed pattern itself: mesh-sdk's TopicMatches does not read `>` (reported to the lead), so only a key equal
// to it reaches the module.
p.event(catalog.Seat+"."+catalog.EventCatalogRevision+".>", catalog.CatalogRevisionEvent{Revision: rev, Kind: catalogclient.KindTrunk, Branch: "main", Changed: []string{module}})
// The seat's event on the branch's part, as the bus says it (design 47 G §532): the module's subscription is `<seat>.<event>.>`.
p.event(catalog.Seat+"."+catalog.EventCatalogRevision+".main", catalog.CatalogRevisionEvent{Revision: rev, Kind: catalogclient.KindTrunk, Branch: "main", Changed: []string{module}})
}
// buildIn makes the seat mesh-catalog hold the entry of a build's fingerprint, and the want that names it on a node.