Repository navigation
Collapse the duplicated rejected-op regression block #151
Description
At reviewed SHA 811d6a267b34f1eba2ed0969dd24ccdda565371b, the same rejected-connect fixture, helper, and cases are maintained in two regression files.
- The copied block spans
test/ka4-storable-values.regression.test.ts:27-225andtest/reject-no-mutation.regression.test.ts:35-238. - Keep the distinct oracles: convergence/order coverage in the latter, and cloneable-but-Yjs-unstorable/property coverage later in the former.
- Collapse only the copied block and shared fixture/helper; do not delete either whole suite.
Verification and scope
Exact reviewed locations:
- duplicated helper and basic rejection cases:
test/ka4-storable-values.regression.test.ts:27-225 - corresponding block:
test/reject-no-mutation.regression.test.ts:35-238 - a third comprehensive table overlaps four historical Rejected ops mutate the document — the "byte-identical on reject" (D4) guarantee is false #10 cases but owns all-kind/all-code completeness and abort/retry semantics:
test/ka4-rejection-byte-identity.test.ts:113-715
An independent source-semantic pass confirmed the exact duplicated cases and narrowed the recommendation. It also confirmed that the unique later sections of all three suites must remain.
Freshness: open heads #123 (38e6ec6), #138 (ca2a3ec), #142 (ba591284), and #143 (32c3dcc) do not modify these test blocks. Existing #149 and #150 concern malformed stamps and contradictory node identity; merged #140 concerns canonical grow ordering. None overlaps this maintenance finding.
Suggested outcome: single-source the shared workflow/helper and one table of the basic rejection cases, while preserving the convergence, exhaustive rejection-code, and generated storability-domain oracles.
Reviewed SHA: 811d6a267b34f1eba2ed0969dd24ccdda565371b.
Activity