@fabricorg/ports 0.5.1 → 0.5.2

package/CHANGELOG.md

@@ -1,35 +1,19 @@
 # @fabricorg/ports
 
+## 0.5.2
+
+### Patch Changes
+
+- e41c1f5: The 0.5.1 changelog entry narrated the whole seventh-review fix batch, nearly all of which landed in other packages. The only change to this package was a removed comment. The entry is corrected in place; no code changes.
+
 ## 0.5.1
 
 ### Patch Changes
 
-- 8436c64: Close what a seventh review found in the previous follow-up, two of them regressions it introduced.
-
-  **An idempotency conflict is settled, but which way depends on why.** The previous release treated
-  every conflict as "never applied" and reported it refused. Only a `parameters` or `actor` conflict
-  means that. An `authority_binding` or `action_version` conflict means the _original_ invocation
-  exists and may well have applied — the capture-authority adapter mints a fresh binding on every
-  attempt, so a first attempt whose response was lost conflicts with itself on the second. Deleting that
-  capture and calling it refused was the worst available answer; it is now surfaced as `unknown` under
-  status `conflict_prior_invocation`.
-
-  **A policy block or validation failure is now `unknown`, not `refused`.** Within one attempt both are
-  decided before any effect. On a recovery re-execution the worker skips adapter steps that already
-  succeeded, then re-evaluates policy with a fresh clock and re-parses parameters against a possibly
-  newer action, and either can flip after effects committed. Unlike the authority path, the host
-  persists no execution reason for these, so the client cannot tell first delivery from recovery; it
-  errs toward a person checking. The precise fix is for the host to persist `executionReason` on failed
-  invocations, which is a store change and is recorded rather than attempted here.
-
-  **`failed` means "never delivered" again.** A store fault after a successful delivery was being counted
-  in it; it now lives only in `storeFailures`. The 0.3.0 changelog's disclosure misdescribed what 0.2.0
-  had changed — `parked` never shipped as a count; only `expired` did — and it is corrected in place.
-
-  Every validation pass now sees a duplicated contract once, so one root cause produces one finding
-  rather than one per copy. `deriveAuthorizationGrants` no longer emits a reference for an intent whose
-  action the contract does not declare, matching what promotion resolves. A superseded comment above
-  `PORT_ID` that asserted the claim the fix retracted is removed.
+- 8436c64: Removed a superseded comment above `PORT_ID` that asserted the claim the previous release
+  had retracted. Nothing else in this package changed. The entry originally published here narrated
+  the whole seventh-review fix batch, most of which landed in `@fabricorg/experience-runtime` and
+  `@fabricorg/sdui-release`; it is corrected in place.
 
 ## 0.5.0
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@fabricorg/ports",
-  "version": "0.5.1",
+  "version": "0.5.2",
   "description": "Vendor-neutral port interfaces, a DTCG token resolver, and the contract-test kit every adapter must pass.",
   "license": "MIT",
   "type": "module",