Back to Plate

Slate Issues Gitcrawl Clusters

docs/slate-issues/gitcrawl-clusters.md

53.0.612.3 KB
Original Source

Slate Issues Gitcrawl Clusters

Summary

gitcrawl refresh produced 620 total clusters for 664 open threads. Only 28 clusters have more than one member. That is useful for duplicate discovery but too conservative for the Slate v2 architecture taxonomy.

Cluster IDHuman familyDecisionMembersRepresentativeMembersNotes
1dom-point-resolution-crasheskeep7#4564#4564 1.00, #3723 0.83, #4789 0.83, #3836 0.81, #5711 0.80, #3834, #4984Strong real cluster. Maps directly to DOM bridge / selection import-export failure class.
2dependency-bump-pr-noisepr-only7#6032#6032 1.00, #6049 0.86, #6062 0.85, #6025 0.84, #5869 0.84, #6026 0.81, #6060PR-only dependency/update cluster; exclude from issue architecture.
3mixed-react-input-focus-crashesweak4#3478#3478 1.00, #4001 0.83, #3497 0.81, #3777Weak/mixed: Redux crash, keyboard input, focus loss, Firefox accented input. Needs manual split.
4docs-jsdoc-and-docs-uxdocs3#6045#6045 1.00, #5350 0.94, #5520 0.94Docs/JSDoc/docs-site family; not core architecture.
5inline-boundary-cursor-movementkeep3#4074#4074 1.00, #4618 0.94, #3429 0.83Inline cursor movement and padded inline boundary family.
6history-set-selection-errorskeep3#3705#3705 1.00, #3756 0.94, #3921 0.81History set_selection and undo selection grouping family.
7react-editor-focus-after-programmatic-changekeep3#3634#3634 1.00, #5537 0.81, #4961 0.81ReactEditor.focus/input after programmatic focus or insert family.
8duplicate-rendertextprops-prspr-only2#5861#5861 1.00, #6054 0.94Duplicate PR cluster, not issue triage.
9android-mark-toggle-keyboard-dismissalkeep2#6022#6022 1.00, #6027 0.94Issue plus matching PR; good direct fix-chain candidate.
10async-decoration-caret-jumpkeep2#5987#5987 1.00, #6033 0.94Issue plus matching PR; direct decoration/runtime cluster.
11android-empty-node-voice-input-duplicationkeep2#5983#5983 1.00, #6020 0.94Issue plus matching PR; direct Android IME cluster.
12mobile-inline-void-selection-keyboardkeep2#5183#5183 1.00, #5391 0.81Android/iOS inline void keyboard and cursor placement family.
13android-ime-empty-node-compositionkeep2#4400#4400 1.00, #5883 0.94AndroidEditable / empty text node composition family.
14refocus-autoscrollkeep2#5826#5826 1.00, #5882 0.94Issue plus PR for unexpected auto-scroll on refocus.
15example-packaging-improvementsdocs2#4956#4956 1.00, #5172 0.94Examples/product packaging family; docs/example lane.
16input-event-missing-at-boundarieskeep2#5603#5603 1.00, #5669 0.81Input event missing at contenteditable boundary/deletion/number input.
17void-delete-selection-regressionskeep2#3991#3991 1.00, #4301 0.94Void delete and selection regression family.
18android-readonly-change-input-opskeep2#4994#4994 1.00, #5026 0.85Android readOnly state change and operation emission family.
19scroll-selection-into-view-regressionskeep2#5088#5088 1.00, #5473 0.94scrollSelectionIntoView not firing after update/delete family.
20blur-invalid-selection-updateskeep2#4376#4376 1.00, #5171 0.81Safari/Firefox blur or unfocused invalid selection update family.
21inline-void-copy-pastekeep2#4802#4802 1.00, #4806 0.85Inline void copy/paste family.
22safari-cyrillic-spellcheckkeep2#5095#5095 1.00, #5096 0.91iOS/macOS Safari Cyrillic spellcheck family.
23triple-click-block-selectionkeep2#3871#3871 1.00, #5847 0.94Triple-click block selection/deletion behavior family.
24placeholder-data-attribute-typesdocs2#3780#3780 1.00, #3781 0.94Placeholder prop/type API ergonomics; docs/API lane.
25inline-dom-end-selectionkeep2#3148#3148 1.00, #3150 0.94Chrome/Safari inline DOM end selection known upstream family.
26ie11-selection-onchangestale2#3800#3800 1.00, #4111 0.94IE11-specific selection/onChange family; stale/non-v2 unless explicitly supported.
27undo-selection-state-corruptionkeep2#3534#3534 1.00, #3551 0.94Undo selection and move_node wrong-state family.
28android-legacy-supportstale2#3112#3112 1.00, #3313 0.94Legacy Android support family; stale unless reproduced on current supported Android.

Decision Legend

DecisionMeaning
keepReal issue family. Use as evidence in the live ledger and v2 cluster map.
weakMachine cluster is mixed. Split manually before making architecture claims.
docsDocs/examples/API ergonomics lane. Useful, but not core runtime architecture.
staleLegacy/browser support lane. Needs current repro before v2 work.
pr-onlyOpen PR/dependency/update cluster. Keep for linkage, exclude from issue corpus counts.

Strongest Keep Clusters

  • dom-point-resolution-crashes: repeated Cannot resolve ... point failures around DOM bridge, nested editors, outside selections, iOS deletion, and stale DOM mappings.
  • android-mark-toggle-keyboard-dismissal, android-empty-node-voice-input-duplication, android-ime-empty-node-composition, android-readonly-change-input-ops, and android-legacy-support: Android remains a single large runtime pressure area even when gitcrawl splits it into small clusters.
  • inline-void-copy-paste, mobile-inline-void-selection-keyboard, inline-dom-end-selection, and void-delete-selection-regressions: inline/void boundary behavior is still one of the central DOM bridge families.
  • history-set-selection-errors and undo-selection-state-corruption: history and selection state still need transaction-level proof.
  • async-decoration-caret-jump: current open PR linkage makes this a high-signal runtime/projection cluster.

Weak Or Excluded Clusters

  • Cluster 2 and 8 are PR-only; exclude from open issue counts.
  • Cluster 3 is too mixed to trust directly. First ClawSweeper split: #3478 belongs to external-store controlled-value feedback, #3497 belongs to parent-rerender focus loss, and #4001 belongs to placeholder/IME DOM point desync. #3777 still needs the same treatment.
  • Cluster 26 and 28 are legacy-support rows. They should not drive v2 architecture unless reproduced on supported browsers/devices.

V2 Sync Notes

  • 2026-05-07 Mobile/IME macro Ralplan keeps Android/input clusters 9, 11, 12, 13, 16, and 18 as real runtime pressure, with adjacent DOM/focus clusters 1, 7, and 19 accounted as related proof-route pressure.
  • No cluster is promoted to an exact fixed/improved claim by the live refresh. Synthetic mobile proof remains synthetic; exact mobile/IME closure requires matching raw-device or browser artifacts for the reported environment.