[research] UX walls + building-elements pain-points investigation#151
Open
rcasto123 wants to merge 1 commit into
Open
[research] UX walls + building-elements pain-points investigation#151rcasto123 wants to merge 1 commit into
rcasto123 wants to merge 1 commit into
Conversation
Research-only artifact (no code changes). Documents wall, door, window, common-area, and conference-room pain points reported by users, with prioritized P0/P1/P2 recommendations and open questions for product/eng. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
✅ Deploy Preview for floorcraft-app ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
8 tasks
rcasto123
added a commit
that referenced
this pull request
Apr 27, 2026
…cade-delete (#155) Four targeted fixes for the most-cited frustrations from the walls-pain-points research (PR #151) and walls-implementation audit (PR #153). Each lands as a self-contained change inside the existing data model — no schema work, no geometry refactor, no door/window placement changes. Fix 1 — Walls actually drag. Walls render their wrapping <Group> at (0, 0) because their geometry lives in `points: number[]` (the `ownsPosition` contract in ElementRenderer.tsx). The previous drag-end path wrote `{x, y}` which the renderer ignored, so dragging a wall snapped it back. New `translateWall(id, dx, dy)` helper in wallEditing.ts shifts every entry of `points[]` by the drag delta. Doors and windows attached via `parentWallId` follow automatically because they resolve their world position from the parent wall's points each render. Fix 2 — Endpoint snap during wall drawing AND vertex drag. New `findNearestWallVertex()` in wallAttachment.ts. Wired into `useWallDrawing.snapPoint` ahead of grid snap, and into `applyVertexMove` in wallEditing.ts, so a click or vertex drop within `ENDPOINT_SNAP_PX (10) / stageScale` canvas units of any existing wall vertex lands exactly on that vertex. Two walls drawn corner-to-corner now share an endpoint cleanly with no 1-3px gap. Fix 3 — Shift cardinal angle lock. New `lockToCardinal(ax, ay, px, py)` in a new wallSnap.ts module that projects a candidate point onto the nearest 0°/45°/90°/135° ray from the previous committed vertex. Active during the preview phase (so the user sees what they'll commit) and at the commit itself. Also flows through `applyVertexMove` so vertex drag with Shift held constrains to a cardinal direction relative to the adjacent vertex. Fix 4 — Cascade-delete toast with Undo. The cascade itself already lived in `deleteElements` (seatAssignment.ts); this adds a single info toast ("Wall and N attached elements deleted") with an Undo action that re-merges the captured pre-delete element + employee snapshots back into the live stores. We deliberately keep the cascade behaviour rather than silently re-anchoring children to the nearest other wall — silent mutation is the worse failure mode. Tests: src/__tests__/wallsP0Fixes.test.ts adds 16 integration-level assertions, covering all four fixes plus regression cases (Shift NOT held, no-attachment delete doesn't toast). Existing wall test suites remain green (250 files, 1705 tests). Source-of-truth research: - docs/ux-research/walls-pain-points-2026-04-27.md (PR #151) - docs/engineering-research/walls-implementation-2026-04-27.md (PR #153) Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
8 tasks
rcasto123
added a commit
that referenced
this pull request
Apr 27, 2026
… retire connectedWallIds (#156) Adds the four P1 wall-drawing wins identified by the parallel UX (#151) and engineering (#153) audits, building on top of the P0 fixes that just landed in #155. Changes: 1. Live dimension readout pill (Fix 1). Each in-flight wall segment — committed-but-pre-finalisation segments AND the live rubber-band preview — now renders a small rounded-rect pill at its midpoint showing the segment's real-world length and angle. The pill uses `formatLength` + `toRealLength` from `src/lib/units.ts` so a 12-ft wall reads "12.0 ft" everywhere; angles within 5° of a cardinal gain a "→ N/S/E/W" indicator. Inverse-zoom-scaled the same way the alignment-guide labels are. When Shift is held, the pill reflects the LOCKED angle/length (because `currentPoint` already runs through the cardinal-lock projection at preview time), so the user sees what they will commit, not what they're hovering at. 2. Rectangle / room tool (Fix 2). New `'room'` tool registered in `ToolSelector` (lucide `Square` icon, Shift+O hotkey). Click-drag from corner A to corner B and on release we commit FOUR connected `WallElement`s in a single zundo batch — one undo rolls back the whole rectangle. Endpoint snap from #155 applies to both corners during the drag; cardinal-lock (Shift) constrains the rectangle to a square (shorter side wins, sign preserved). The rectangle preview renders width+height dimension pills using the same helper as the wall-tool pills. 3. Auto-close on snap-back to start (Fix 3). When drawing a polyline wall with 3+ committed vertices, if the next click lands within `ENDPOINT_SNAP_PX / stageScale` of the FIRST committed vertex, the click commits a closing-back-to-start vertex AND auto-finalises the wall — no double-click required. Visual cue: a filled green circle at the start vertex when the cursor is in close-snap range, mirroring the endpoint-snap hit-marker idiom. 4. Retire `connectedWallIds` (Fix 4). The dead metadata slot on `WallElement` — declared but never written or read outside test fixtures — is removed from the type, every construction site (`useWallDrawing`, demo seeder, three template seeders), and the live test fixtures that referenced it. The legacy-payload loader defensively strips the field on load so older autosave payloads round-trip cleanly through the new shape. Tests: `src/__tests__/wallsP1Improvements.test.ts` covers all four features (11 tests). The pill helpers, the rectangle-builder, and the square-constrain helper are pure functions tested directly; the auto-close path is tested through the `useWallDrawing` hook. Verification: tsc clean, vitest 1716 passed (1705 original + 11 new), build green, eslint clean on touched files. Refs: #151 (UX research), #153 (engineering audit), #155 (P0 predecessor). Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
8 tasks
rcasto123
added a commit
that referenced
this pull request
Apr 27, 2026
…handle attached elements (#158) Walls are now fully editable post-placement. Two new affordances close the "delete-and-redraw" gap that the research docs (PR #151) flagged as the top remaining wall pain point: 1. **Add vertex** — hovering near a wall edge (in the wall tool, or with a single wall selected) shows an open-green-circle "+ vertex" indicator at the projected point on the polyline. Clicking commits a new vertex in `points[]` between the two boundary vertices. Curved segments split into two halves with halved bulges so the visible arc continues. The newly-inserted vertex is selected as the active vertex so a follow-up drag fine-tunes it (or Backspace removes it). 2. **Remove vertex** — clicking a vertex sets it active (a slightly-larger filled circle with a dark outline + a "Backspace to remove" tooltip). Backspace/Delete prunes that single vertex instead of nuking the whole wall. Doors/windows whose `positionOnWall` falls on the removed segment(s) are cascade-deleted with the same Undo-toast pattern as PR #155's wall-delete cascade. Removing a vertex from a 2-vertex wall falls back to the existing wall-delete path. Double-click on a vertex is an equivalent mouse-only affordance. New helpers: - `addVertexAt(wall, segmentIndex, point)` — pure, returns updated wall + new vertex index (callers wire the store update + active selection) - `removeVertex(wall, vertexIndex)` — pure, returns null when the wall would become degenerate - `findNearestPointOnWallEdge(segs, px, py)` — chord projection across a wall's polyline, distance returned for caller-side tolerance - `removeWallVertex(wallId, vertexIndex)` — orchestrates the cascade, snapshot, store mutation, and Undo toast in one zundo step State: - `useUIStore.activeVertex: { wallId, vertexIndex } | null` clears on any selection change so a stale index can never refer to a different wall's points array No renderer changes — geometry edits are enough. Refs PRs #155 (P0), #156 (P1), research docs #151 / #153. Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Status
Research-only artifact. Not for merge. This PR exists to make the document discoverable alongside the parallel Engineering investigator's report. The synthesis pass after both land turns these into an MVP plan.
TLDR of the document
The user's frustration ("hard to design a proper office floor plan") traces to two stacked conceptual problems:
Top three things that make the editor feel broken
x=0, y=0on the Group resets them on the next render. Users believe the editor is broken.P0 recommendations (fix this week)
pointson drag end). Effort: S.findNearestStraightWallHit. Effort: M.P1 / P2 highlights
Conference Room/Common Areafilled boxes with wall-bounded rooms.Open questions for product / eng
connectedWallIdsthe seed of a future graph model, or dead code to delete?conference-roomfilled-box autosaves?Files
docs/ux-research/walls-pain-points-2026-04-27.md(new)Testing
N/A — documentation only, no code changes.