psf-memo/docs/reviews/mute-broadcast-result-summary.md

Mute Broadcast Result — Architect Review

By architect.

Task and commits reviewed

• Task: mute-broadcast-result (refactorer handoff merge_and_process refactorer 722c486ef6).
• Merged swarmforge-refactorer (fast-forward from a1e4a4f). Commits:
  • c0e18d8 Add mute broadcast result specification (specifier)
  • 24d7f0f Implement mute broadcast result (coder)
  • 722c486 Refactor mute broadcast result: share result handlers, add property tests (refactorer)
• Architect review commit: a7d9ca6299 (dead-store removal + error-getter test hardening + tool-written mutation/acceptance manifests).
• Verification record git_sha: a7d9ca6299 (client).
• Records/summary commit: this commit.

The feature shows a broadcast result modal on the profile page after a mute/unmute (Memor 0x6d16/0x6d17): on success the success message, the mute txid, and a link to that transaction on the block explorer (new tab); on failure the broadcast error message. A successful mute flips the button to Unmute and a successful unmute flips it back to Mute; a failed broadcast leaves the button in its previous state. The modal stays open until dismissed and dismissing closes it without navigating. Client-only, mirrors the like/tip and New Post result modals. Touches psf-memo-client only.

Architectural findings

• UI/Core separation (good). The mute result state machine lives in the pure controller src/services/profile-page.js (_broadcastMute, getMuteBroadcastMessage, getMuteResultError, dismissMuteResult) with the memoDb/memoFollow/memoMute/viewer-address dependencies injected by the React page. src/components/app-body/profile/mute-result.js and the extracted src/components/explorer-tx-link.js are presentational plain React.createElement components shared by the browser JSX build and the Node acceptance adapter (acceptance/lib/render-mute-result.js). The whole feature is testable without a DOM or network.
• Dependency rule (good). profile-page.js depends only on the pure block-explorer util. MemoMute/MemoDb (IO) and the wallet are constructed in the React adapter shell (profile/index.js) and injected inward. No core module imports a component, service, or IO module; dependency direction runs from the adapter shell inward to the controller.
• Information hiding (good, one simplification). _broadcastMute records the outcome in lastMuteResult and exposes only getters; the reflected button state is updated inside _setState after the injected handler succeeds, so a failed broadcast cannot change the button. The reset of showMuteResultModal at the start of _broadcastMute was a dead store — the flag is set true unconditionally after the awaited broadcast, so no observer can read the reset value. Removed (see Fixes).
• DRY / acceptance structure (good). The refactorer generalized the like/tip and mute broadcast-result acceptance assertions behind BROADCAST_RESULTS + renderBroadcastResult, the follow/unfollow/mute/unmute broadcast checks behind assertMemoBroadcastPrefix, and extracted the shared ExplorerTxLink used by both LikeResult and MuteResult. Behavior is unchanged; the broad DRY run now reports only the pre-existing handlers.js step-handler boilerplate (untouched lines), and the changed production/test set reports no duplicate candidates.
• Mutation surface. explorer-tx-link.js, mute-result.js, and like-result.js each scan as 0 mutation sites (structural, like block-explorer.js): they are built from !/ternary/template-literal markup with no arithmetic, comparison, or boolean operators the tool targets. Their behavior is pinned by unit and property tests instead. The only mutated module with sites is profile-page.js.
• Observation, not changed — merge included two earlier master commits. The fast-forward from a1e4a4f to 722c486 also brought in ea67979 (remove redundant author name from posts) and 8eab46d (notifications CSS + entry tweaks), both committed to master before the mute spec and outside the refactorer handoff. They were not the subject of this review; the full client suite (unit/property/acceptance/lint/build) passes with them.
• Observation, not changed — render-like-result/render-mute-result adapters are near-identical. They render different components and DRY reports no candidate at its thresholds, so they are left as separate test helpers (test helpers stay separate from the modules under test).

Fixes applied

• Removed the dead showMuteResultModal reset. ProfilePage._broadcastMute set the modal hidden, then set it visible unconditionally after the awaited broadcast. The reset was unobservable and left a surviving equivalent mutant (false -> true); deleting it removes the site and the dead state write.
• Pinned getMuteResultError's empty-string contract. Added two unit tests: no result (constructor default null) and a failure result with no message. These kill both || -> && mutants in the getter (the null case exercises the short-circuit guard, the no-message case exercises the trailing || '').
• No further production changes were needed. Module structure, dependency direction, and information hiding were accepted.

Verification results

Record (pinned to the review commit a7d9ca6299):

File	Component
docs/reviews/mute-broadcast-result-verification.json	psf-memo-client

• Language mutation (mutate4javascript, --max-workers 8, --mutate-all; wrapper confirmed all sites selected):
  • src/services/profile-page.js: 31 killed, 0 survived, 0 uncovered. First differential run selected all 32 sites (30 killed / 2 survived: the dead modal reset and the getMuteResultError guard) → after the fixes, 31 sites, all killed.
  • src/components/explorer-tx-link.js, src/components/app-body/profile/mute-result.js, src/components/post-feed/like-result.js: 0 mutation sites (structural).
  • Total: 31 killed, 0 survived, 0 uncovered.
  • src/services/memo-db.js was not mutated: the client HTTP adapter is a standing mutation exclusion (ESM + ../config directory import).
• DRY (dry4javascript, scoped to the changed production files, the acceptance adapters, and the new/changed tests): No duplicate candidates found. The broad run over acceptance/lib/handlers.js reports only pre-existing step-handler boilerplate on lines untouched by this task.
• Cyclomatic complexity / CRAP (crap4javascript): max CRAP 4.0 (ProfilePage._broadcastMute CC 4, getMuteBroadcastMessage CC 4, getMuteResultError CC 4, all 100% covered); ExplorerTxLink and MuteResult CC 2, LikeResult CC 1, every other changed function CC ≤ 3 at 100%. All functions are below the 8.0 threshold; no refactor was warranted.
• Soft Gherkin acceptance mutation (gherkin-mutator --level soft --workers 8), psf-memo-client/specs/mute-broadcast-result.feature: 5 total, 0 killed, 5 survived, 0 errors. All 5 are single-character case mutations of example values (addr in scenarios 1–4, broadcast_error in scenario 4) used consistently on both the Given setup and the Then assertion side, so the mutated value still matches — the documented intrinsic-equivalent class. None is an implementation gap and none was chased. gherkin-mutator wrote a manifest with "scenarios":[] because every scenario has an intrinsic survivor; committed as tool-written.

Suite status

verify.sh client result: pass at git_sha a7d9ca6299:

• unit: 484 pass / 0 fail
• property: 111 pass / 0 fail
• acceptance: all 36 generated suites passed (including all 4 mute-broadcast-result scenarios)
• lint: ok
• build: ok

Handoffs

• End-of-chain git_handoff to specifier (priority 50) to merge swarmforge-architect into master, using this records/summary commit.
• git_handoff to coder and refactorer (priority 00) with the review commit a7d9ca6299 for follow-up review: the only production change is the removal of the dead showMuteResultModal reset in ProfilePage._broadcastMute; the rest is two edge-case unit tests and tool-written manifests.