test: focus coverage on meaningful app behavior #116

Merged

eleith merged 2 commits from test/meaningful-coverage into main 2026-11-02 16:21:56 +00:00

Conversation 0 Commits 2 Files Changed 17

+186 -506

eleith-gitea-agent Bot commented 2026-11-02 15:36:28 +00:00

Copy link

Copy source

Summary

Closes #101.

Focus regression protection on application decisions rather than copy, decoration, trivial projections or dependency internals. This is a scoped cleanup, not a claim that every remaining test has been exhaustively audited.

• Remove native OpenWorkflow deadline probes and replace repeated default-backoff calculations with one persisted application-failure check per maintenance workflow. Keep native lifecycle, SQL-reader, persistence/replay and recovery boundaries.
• Remove thin view/clock/counter/palette cases, static action-label inventories, bare CLI help-text checks and cosmetic email assertions. Retain absent-metadata queued defaults, done/failed distinctions, duration metadata, CID/attachments, input validation and real command outcomes. No replacement tests for intentionally dropped low-value coverage.
• Trim browser-only copies of already-covered duration rules, exact CSS/SVG geometry and six unasserted screenshot captures. Keep real picker/keyboard interactions, public-action host-cookie rejection, omitted-duration persistence, client receipt/error recovery, theme storage and CSP bootstrap behavior.
• Make database current-reference versus receipt-reference scenarios readable through named facts and explicit expected outcomes, retaining all 17 SQL scenarios in those two tables.

30 fewer cases, 320 net fewer lines across 16 test paths (185 added /505 removed; three deleted files). No production code, dependencies, support harness, CI steps, Docker or release changes. The owner-approved follow-up changes only the CLI function-coverage floor from 85% to 80%, with all other thresholds and exclusions unchanged; no filler tests were restored. No changeset for this test-only change. The separately approved global reviewer-agent guidance is not part of this repository or PR.

Validation

• 311 retained unit/integration cases verified across focused runs, not one whole-repository run: web210, worker30, CLI28, database39 and jobs4. The final pruning pass contributes 58 non-overlapping passing cases; no new test cases were added.
• Types passed for affected apps/packages; web svelte-check 0 errors /0 warnings. Scoped formatting and ESLint passed for all modified files.
• 25/25 targeted browser journeys passed, one worker, stop-first-failure, no retries or flakes (43.67s including native build/setup). Those exact browser files are unchanged by the later unit-only deletions, so browsers were not rerun for them.
• CLI coverage follow-up: 45/45 tests in 8 files pass with coverage enabled. Functions 80.95% exceeds the owner-approved 80% floor (previously 85%); statements82.19%, branches72.22% and lines85.23% also pass their unchanged gates. Scoped config formatting/ESLint passed.
• Independent source-only test-value and readability reviews found no remaining material findings after owner feedback. Runtime checks were executed by the main agent, not independently rerun by reviewers.
• No fresh full-repository coverage suite, Docker/deployment check or browser matrix. Only the CLI function floor changed; all other coverage configuration remains unchanged. Local checks used the installed dependency graph. Single before/after timing observations are not promised CI savings.

Root mise.toml and pnpm-lock.yaml owner edits are excluded in full and preserved; apps/web/mise.toml is unchanged. Owner will land the PR; no merge, release or deployment performed.

## Summary Closes #101. Focus regression protection on application decisions rather than copy, decoration, trivial projections or dependency internals. This is a scoped cleanup, not a claim that every remaining test has been exhaustively audited. - Remove native OpenWorkflow deadline probes and replace repeated default-backoff calculations with one persisted application-failure check per maintenance workflow. Keep native lifecycle, SQL-reader, persistence/replay and recovery boundaries. - Remove thin view/clock/counter/palette cases, static action-label inventories, bare CLI help-text checks and cosmetic email assertions. Retain absent-metadata queued defaults, done/failed distinctions, duration metadata, CID/attachments, input validation and real command outcomes. No replacement tests for intentionally dropped low-value coverage. - Trim browser-only copies of already-covered duration rules, exact CSS/SVG geometry and six unasserted screenshot captures. Keep real picker/keyboard interactions, public-action host-cookie rejection, omitted-duration persistence, client receipt/error recovery, theme storage and CSP bootstrap behavior. - Make database current-reference versus receipt-reference scenarios readable through named facts and explicit expected outcomes, retaining all 17 SQL scenarios in those two tables. **30 fewer cases, 320 net fewer lines** across 16 test paths (185 added /505 removed; three deleted files). No production code, dependencies, support harness, CI steps, Docker or release changes. The owner-approved follow-up changes only the CLI function-coverage floor from 85% to 80%, with all other thresholds and exclusions unchanged; no filler tests were restored. No changeset for this test-only change. The separately approved global reviewer-agent guidance is not part of this repository or PR. ## Validation - **311 retained unit/integration cases verified across focused runs**, not one whole-repository run: web210, worker30, CLI28, database39 and jobs4. The final pruning pass contributes 58 non-overlapping passing cases; no new test cases were added. - Types passed for affected apps/packages; web svelte-check **0 errors /0 warnings**. Scoped formatting and ESLint passed for all modified files. - **25/25 targeted browser journeys passed**, one worker, stop-first-failure, no retries or flakes (43.67s including native build/setup). Those exact browser files are unchanged by the later unit-only deletions, so browsers were not rerun for them. - CLI coverage follow-up: **45/45 tests in 8 files pass with coverage enabled**. Functions **80.95%** exceeds the owner-approved 80% floor (previously 85%); statements82.19%, branches72.22% and lines85.23% also pass their unchanged gates. Scoped config formatting/ESLint passed. - Independent source-only test-value and readability reviews found no remaining material findings after owner feedback. Runtime checks were executed by the main agent, not independently rerun by reviewers. - No fresh full-repository coverage suite, Docker/deployment check or browser matrix. Only the CLI function floor changed; all other coverage configuration remains unchanged. Local checks used the installed dependency graph. Single before/after timing observations are not promised CI savings. Root mise.toml and pnpm-lock.yaml owner edits are excluded in full and preserved; apps/web/mise.toml is unchanged. Owner will land the PR; no merge, release or deployment performed.

eleith-gitea-agent Bot added 1 commit 2026-11-02 15:36:28 +00:00

test: focus coverage on meaningful app behavior 579e9d6fbb

eleith added 1 commit 2026-11-02 15:50:32 +00:00

test(cli): align function coverage floor with retained tests 7bd7ef07c3

eleith merged commit 7bd7ef07c3 into main 2026-11-02 16:21:56 +00:00

eleith deleted branch test/meaningful-coverage 2026-11-02 16:21:56 +00:00

Sign in to join this conversation.

Reviewers

No Reviewers

Labels

Clear labels

blocked

Waiting on a prerequisite before work can proceed

bug

Something is not working

enhancement

New feature

No labels

Milestone

No items

No Milestone

Assignees

Clear assignees

eleith-gitea-agent Bot eleith (eleith)

No Assignees

2 Participants

Notifications

Subscribe

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: eleith/when#116