feat: editor settings, remove readonly from single query run, smarter grid column widths by emrberk · Pull Request #584 · questdb/ui

emrberk · 2026-07-06T09:06:06Z

Summary

Adds an Editor settings modal and reworks how the result grid sizes its columns, while letting you keep editing the SQL editor without interruption during a running query.

Changes

Editor settings modal (editor tabs menu → Editor settings):
- Run with selection: when on, editor run actions execute your highlighted text; when off, they always run the statement under the cursor. Shared query links still run their selected fragment regardless of this setting.
- Maximum column width: cap how wide result columns can auto-size (or leave empty for fully automatic). You can still drag to resize any column.
Keep editing while a query runs: the editor is no longer locked during execution. You can type, add lines above a running query, or start a new run; the run indicator follows the query as it moves, and the result still lands on the right statement (or is cleanly dropped if you rewrite it mid-flight).
Smarter result-grid column widths: columns are measured from the actual rendered font, wide columns share the available space more sensibly instead of hard-capping, and widths re-settle once web fonts finish loading. Long array values are truncated to fit with an ellipsis.

Notes

New settings persist in local storage (editor.runWithSelection, grid.maxColumnWidth); no schema/migration changes.
Both settings apply to the main Result grid and to Notebook result cells.

Testing

Unit tests added for width sampling/clamping, the run-with-selection gating, inflight-query offset tracking, the notification-key reducer, and settings parsing/validation.
E2E coverage added for the settings modal, run-with-selection behavior, editing while a query runs, and column-width allocation.
yarn typecheck, yarn lint, yarn test:unit, and yarn build all pass.

… grid column widths

emrberk · 2026-07-06T09:06:32Z

Code Review — staged diff (level 3, full mission-critical pass)

Reviewed at level 3 — all 13 review agents, per-finding verification. Scope of the change: editing the editor while a query runs (Monaco inflight-offset tracking), a "Run with selection" setting, a max-column-width setting, canvas-based result-grid column measurement, and a notification-key collision reducer.

Quality gates: yarn typecheck ✅ · yarn lint ✅ · yarn build ✅ · yarn test:unit ✅ (1305 tests). No quality-gate failures.

The core high-risk machinery — supersede/runSeq guarding, inflight offset math, the notification-key collision reducer, the width memo chain, and every changed cross-context callsite — was code-traced and is correct. No data-loss or data-misrepresentation bug survived verification. The remaining findings are a few minor UX, accessibility, and performance items.

Issues

#	Issue	Category	Severity	Location	Description	Repro	Suggested fix
1	Focus may not return to trigger on modal close	Accessibility & UX	Minor (NEEDS VERIFICATION)	in-diff (`src/components/EditorSettingsModal/index.tsx:229`, `src/scenes/Editor/Monaco/tabs.tsx:672`)	The modal opens from a `DropdownMenu.Item`. On close, Radix `FocusScope` restores focus to the element that was active when the dialog mounted — potentially the now-unmounted dropdown item — so focus can fall to `<body>` instead of the tabs "…" button, stranding keyboard/AT users. Could not reproduce live (the running dev instance did not reflect the new menu item), so this is static-analysis only; the common case often works.	Keyboard: open tabs "…" → Editor settings → Escape; inspect `document.activeElement`.	If confirmed live, handle `onCloseAutoFocus` on `Dialog.Content` and refocus the tabs trigger from the `tabs.tsx` callsite.
2	Per-cell O(L) upper-bound scan in sampling	Performance & rendering	Minor	in-diff (`src/components/ResultGrid/inlineGridUtils.ts:269`)	`sampleColumnWidths` runs an unconditional O(L) `cellWidthUpperBound` scan (up to `MAX_MEASURE_CHARS`=2000 chars) on every sampled cell, even when the running `width` already dominates. Not a complexity regression (both old and new are O(C·R·L) because formatting is O(L)), but a constant-factor cost on wide string columns (up to 1000 sampled rows × long cells). A cheap `length × maxAsciiAdvance` O(1) upper bound could gate the scan.	Wide `string_agg`/varchar result, 1000 sampled rows.	Add an O(1) length-based upper-bound early-continue before the scan (`len * maxAdvance * margin + chrome ≤ width`); a lower-bound gate would under-measure.
3	Toggle draft not saved when width is invalid	Accessibility & UX	Minor	in-diff (`src/components/EditorSettingsModal/index.tsx:124`)	`handleSave` early-returns on an invalid max-column-width before persisting `runWithSelection`. If the user flips the switch and enters an invalid width, Save only surfaces the width error; the toggle is not persisted that click. Mitigated: the modal stays open and the switch draft is retained, so fixing the width and re-saving persists both — nothing is permanently lost. Defensible atomic-validation, flagged for UX transparency.	Flip the switch, type `10` in the width field, click Save.	Optional: persist the valid setting independently, or make the coupling explicit.
4	Array cell renders full (clipped) at extreme min width	Performance & rendering	Minor	in-diff (`src/components/ResultGrid/inlineGridUtils.ts:195`)	When the column is so narrow that ≤3 array content chars fit, `fitArrayToWidth` returns the full array string (CSS-clips it) instead of a `...]` ellipsis. Reachable via max-column-width = 60 on an `ARRAY` column. Pre-existing design (old code's `maxContentLength > 3` guard did the same); data still available via copy. Cosmetic only.	Set max column width to 60, run a query with a long `ARRAY` column.	Optional: return `wrapArray("...", dim)` instead of `full` so the truncation intent still reads.

False-positives

Font-load double-sample: The useFontsReady → invalidateMeasuredFont → re-sample is required, not wasteful — the first (fallback-font) measurement is genuinely wrong and must be redone once the webfont loads. Bounded to once per result. Not a bug.
columnDefs rebuild / array re-truncate on resize drag: Both bounded (O(C); visible cells only) and only during active resize. Negligible; not worth complicating the code.
useContainerWidth initial null → empty grid forever: Requires a ResultGrid mounted in a permanently-zero-width container that never fires a nonzero ResizeObserver entry. Both current callsites mount in laid-out containers; unreachable today. Latent constraint, not a bug.
AI-conversation updateNotificationKey refusal breaks linkage: SAFE. meta.queryKey is persisted to IndexedDB before the dispatch (AIConversationProvider:923), and "Fix query" reads executionRefs, not the notification store. The refusal is display-only and only fires when a newer real editor notification legitimately occupies the key — keeping it is correct.
Reducer refusal leaves flat/keyed arrays divergent: The refusal is return state — a fully atomic referential no-op; both structures stay in their already-consistent state. Verified by reducers.test.ts:85 (next.toBe(state)).
document.fonts unguarded / canvas unavailable / security sinks: All document.fonts access is behind "fonts" in document; canvas-null degrades to the fallback char-width path; only TextMetrics.width (universal) is used. No XSS/injection/secret sink introduced — max-width flows as a numeric prop, and share-link params are pre-existing and target the user's own DB.

Summary

Verdict: Approve — all remaining findings are minor or optional; none blocking.

Draft findings: 4 verified / kept; ~15+ candidate concerns dropped as false-positives or by-design after per-finding source verification.
In-diff vs out-of-diff: 4 in-diff, 0 out-of-diff. The zero is genuine, not a cross-context underrun: the changed contracts are either new required params/props whose few callers were all updated in the same diff, or the reducer refusal (a graceful improvement at every callsite). The cross-context pass widened greps across tests/stories/dynamic construction and cleared the AI callsite, the Monaco shift callsites, the notebook execution path (correctly has no symmetric rekey need), and all useLocalStorage / Switch / ResultGrid consumers.
Regressions/tradeoffs: None found. FixQueryButton replacing its ! assertions with real null guards, and the LocalStorageProvider value memoization, are genuine quality/perf improvements. The edit-while-running machinery, supersede guarding, and offset math are correct and well-unit-tested.

feat: editor settings, remove readonly from single query run, smarter…

30f552d

… grid column widths

emrberk and others added 2 commits July 6, 2026 13:37

test fix

2f8a77d

Merge branch 'main' into feat/editor-settings

90eac2e

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

feat: editor settings, remove readonly from single query run, smarter grid column widths#584

feat: editor settings, remove readonly from single query run, smarter grid column widths#584
emrberk wants to merge 3 commits into
mainfrom
feat/editor-settings

emrberk commented Jul 6, 2026

Uh oh!

emrberk commented Jul 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

emrberk commented Jul 6, 2026

Summary

Changes

Notes

Testing

Uh oh!

emrberk commented Jul 6, 2026

Code Review — staged diff (level 3, full mission-critical pass)

Issues

False-positives

Summary

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant