Coverage Priority Map Testing Review Non React Post Bun Check

Inputs

• Coverage source: lcov.info
• Constraints:
  • temporary exclude /react
  • temporarily exclude obvious component shells outside /react
  • no browser or e2e coverage work
  • no coverage vanity
  • file-first ranking, not package theater

Coverage Run

• Command: scoped bun test --coverage --coverage-reporter=lcov --coverage-dir=.coverage-repo-2026-03-24p --reporter=dots over the temporary non-React file list
• Result: Fresh non-React coverage: 2745 pass, 0 fail, 558 files, 2.77s.
• Caveat: full-repo bun test --coverage is still polluted by React-side suite poison, so this pass uses the stable /react-excluded run instead of lying to itself.

Suite Health

• pnpm test:profile -- --top 25: green
• pnpm test:slowest -- --top 25: green
• fast suite result: 2983 pass, 0 fail, 606 files, 4.30s
• average: 0.552ms
• median: 0.234ms
• slow-bucket thresholds: 75ms/test, 150ms/file
• threshold breaches: none
• top outliers are still mostly React-heavy fast-lane files, so they are suite-health data, not ranking input for this temporary cut
• skipped-test debt scan: none worth fixing
• commented-out spec scan: none worth fixing
• cross-spec import scan: none

Scoring Rules

• Scope is packages/**/src/**.
• Exclude test files, barrels, declaration files, obvious type-only files, generated junk, the temporary /react slice, and obvious component shells outside /react.
• Penalize wrappers, crumbs, giant sludge, DOM-heavy leftovers, and low-ROI utility dust.
• Reward transforms, queries, merge helpers, parser or serializer seams, plugin resolution, plugin contracts, and non-React static runtime seams.
• Missing-from-lcov runtime files are treated as fully uncovered. Pretending they are covered is clown math.
• package_score is the sum of the top 5 remaining file scores in that package.

Strong Take

Do not do another non-React package sweep.

There is one last real non-React cleanup phase left. It is mostly parser or serializer seams plus a few bounded transform or plugin contract gaps. After the score >= 5 queue is burned down, stop. The remaining misses are mostly score-4 dust, DOM-heavy leftovers, wrapper clusters, or giant serializer sludge.

Threshold Counts

• score >= 6: 2 files
• score >= 5: 21 files
• score >= 4: 48 files
• score >= 3: 86 files
• score >= 2: 134 files
• score >= 1: 168 files

Strict Next Batch

1. link: upsertLink.ts — score 6
2. markdown: convertNodesSerialize.ts — score 6

Wider Final Batch

1. core: DOMPlugin.ts — score 5
2. autoformat: AutoformatPlugin.ts — score 5
3. basic-nodes: BaseHeadingPlugin.ts — score 5
4. core: inferWhiteSpaceRule.ts — score 5
5. core: pipeRenderElementStatic.tsx — score 5
6. core: cleanHtmlFontElements.ts — score 5
7. core: pluginDeserializeHtml.ts — score 5
8. table: getTableCellBorders.ts — score 5
9. core: cleanHtmlTextNodes.ts — score 5
10. markdown: splitIncompleteMdx.ts — score 5
11. core: isLastNonEmptyTextOfInlineFormattingContext.ts — score 5
12. core: traverseHtmlNode.ts — score 5
13. list-classic: moveListItemSublistItemsToListItemSublist.ts — score 5
14. markdown: mdastToSlate.ts — score 5
15. markdown: markdownToSlateNodesSafely.ts — score 5
16. suggestion: addMarkSuggestion.ts — score 5
17. suggestion: removeMarkSuggestion.ts — score 5
18. table: deleteRowWhenExpanded.ts — score 5

Packages By Value

1. markdown — package score 25
2. core — package score 25
3. table — package score 22
4. basic-nodes — package score 21
5. suggestion — package score 20
6. list-classic — package score 19
7. code-block — package score 14
8. docx-io — package score 14
9. list — package score 11
10. slate — package score 11

Best Files By Value

1. link — upsertLink.ts — score 6, coverage 93.9%, uncovered 6
2. markdown — convertNodesSerialize.ts — score 6, coverage 95.6%, uncovered 6
3. core — DOMPlugin.ts — score 5, coverage 68.8%, uncovered 20
4. autoformat — AutoformatPlugin.ts — score 5, coverage 88.9%, uncovered 8
5. basic-nodes — BaseHeadingPlugin.ts — score 5, coverage 93.5%, uncovered 6
6. core — inferWhiteSpaceRule.ts — score 5, coverage 79.2%, uncovered 5
7. core — pipeRenderElementStatic.tsx — score 5, coverage 83.3%, uncovered 5
8. core — cleanHtmlFontElements.ts — score 5, coverage 66.7%, uncovered 4
9. core — pluginDeserializeHtml.ts — score 5, coverage 97.5%, uncovered 4
10. table — getTableCellBorders.ts — score 5, coverage 90.2%, uncovered 4

Stop Condition

Stop non-React coverage after the phase-3 roadmap below. After that, the remaining misses are mostly wrapper dust, tiny crumbs, DOM-heavy leftovers, and giant low-ROI serializer slabs.

That is where more non-React coverage turns into percentage cosplay. Switch to React or architecture-safety work instead.

Caveats

• Package totals are noisier than file totals. Trust the file ranking more.
• core and markdown still look inflated because they have several medium-value parser or static crumbs left.
• docx-io still looks bigger than it deserves because html-to-docx.ts is a giant low-ROI serializer slab.

Full Data

• 2026-03-24-coverage-priority-packages-testing-review-non-react-post-bun-check.tsv
• 2026-03-24-coverage-priority-files-testing-review-non-react-post-bun-check.tsv