[jameskraus]

The Canvas editor has a fixed 16-row height, so a tall side pane leaves empty space between the editor and the footer. This PR makes the Edit tab a column whose Markdown editor grows down to the footer.

Before	After

Behavior

• On the Edit tab, the editor fills the pane down to the footer, keeping the pane's 16px inset.
• When the pane is shorter than the editor's 16 rows, the editor keeps that height and the pane scrolls, as before.
• The editor has no resize grip, because an editor that fills the pane can't be dragged smaller.
• History is unchanged.

Changes

• channels/Canvas.module.css. The scroll body, the outer stack and the Edit tab panel are flex columns. The editor grows (flex: 1 0 auto) and has resize: none. The Edit panel sets display: none when hidden, because its display: flex would otherwise override the hidden attribute.
• channels/CanvasEditor.tsx. These two containers use the Canvas classes instead of the shared ChannelTemplates stack, and the Edit panel loses its inner wrapper div.

Design

• Shared owners used: Textarea (code variant), panel Tabs, and the existing Canvas panel and footer. No shared component or token changes.
• Gap: the shared Textarea has no fill mode, and the design guide's Textarea recipe includes manual vertical resizing. Canvas removes the grip in its own CSS, because a filling editor can't be resized. If other surfaces need an editor that fills its space, a Textarea fill option belongs in the design system.
• Rendered checks (browser fixture):
  • 1440×900 dark, in Chromium and WebKit
  • 820×900 light
  • 150% text, dark
  • 1280×560 light, where the editor keeps 16 rows and the pane scrolls:

Tests

• Browser. channel-tabs.spec.mjs › Canvas opens once beside Thread and retains its channel draft gains one assertion: the editor ends at the footer's 16px inset. Layout needs a real browser. Without the CSS change, the assertion fails in Chromium and WebKit (a 369px gap); with it, it passes. No browser cases are added or removed.
• Vitest. The Canvas editor tests pass unchanged.

Stack

Stacked on #1303, above #570.

Readiness

The existing review state is preserved. The restack keeps this PR's layout change; the complete channel-tabs and Canvas-history browser files pass in Chromium and WebKit on the integrated stack (14 checks).

Native desktop, live-relay checks, and human acceptance of the latest stack remain deferred.

[wesbillman]

🤖 Star Lord automated source review via Wes’s account, integrating independent Rocket (systems/lifecycle), Mantis (correctness/security/concurrency) and Groot (product/accessibility) reviews.

No actionable defects found in this incremental editor-fill change. Source-review assessment: 9/10 for minimalness, elegance and correctness.

Reviewed head 4e21a49d4d4410f81f646a3ea33ecf3f36f10439 against base 726d62246813d532345dd9be54d64371d90ebf8e. Checked the complete three-file diff, shared Textarea/Tabs and panel integration, short-pane overflow, hidden Edit/History handling, draft retention, and success/error/retry/cancel focus paths. The existing browser journey gains a layout assertion without adding cases. The unchanged underlying Canvas feature was not exhaustively re-audited.

Optional documentation alignment: src/bundled/canvas/Canvas.module.css:29 intentionally removes the resize grip, as the PR describes. Consider recording this Canvas-specific fill-mode exception in src/shared/design-system/DESIGN.md, whose Form composition section still specifies manual vertical resizing and a visible grip. This is not a request to restore the grip or block the change.

Public description, commit messages and all three attached screenshots inspected; no disclosure finding on those surfaces.

Validation limits: source inspection only; no tests, builds, app launches or PR code executed. CI was not assessed in this review. Rendered keyboard/focus behavior, both browser engines, narrow/short panes, enlarged text, native desktop/live-relay behavior, performance and human acceptance remain unverified by this review. The PR’s reported QA is author-provided evidence, not a run performed here. This COMMENT is not approval or merge authorization.

[wesbillman]

🤖 Carl, an automated reviewer, commenting via Wes's GitHub account.

Reviewed head 4e21a49d4 against base 726d62246. No findings. I checked:

• The resize override. Canvas.module.css is unlayered, so resize: none beats .buzz-textarea's resize: vertical in @layer components.
• Hidden panels. .edit[hidden] is needed and correctly specific, and the History panel keeps the UA hidden behavior.
• The grid-to-flex swap. Replacing grid .stack with a flex column keeps the same --space-4 gap, and cross-axis stretch still bounds width.
• Short panes. flex: 1 0 auto keeps the 16-row minimum, so they scroll as described.

The new 16px footer-inset assertion is a real layout check that belongs in a browser test. The earlier optional DESIGN.md note about the fill-mode exception still stands.

Non-blocking: the description says "Readiness: Draft", but the PR is marked ready. Align one or the other per the repo's PR checklist. Merge depends on #570 (see that review). This COMMENT is not an approval.

[wesbillman]

🤖 Star Lord automated source review via Wes’s account, integrating independent Rocket (systems/lifecycle), Mantis (correctness/security/concurrency), and Groot (product/accessibility) reviews.

No actionable findings in the restacked editor-fill change. Source-review assessment: 9/10 for minimalness, elegance and correctness.

Reviewed head 87dac25 against base 5a7b41b. Checked the complete three-file diff and current Canvas integration: bounded pane scrolling, hidden Edit/History handling, shared Textarea and Tabs, draft retention, async guards, and success/error/retry/cancel focus paths. The existing browser journey adds a layout assertion without adding cases. This is an integration review of this PR, not an exhaustive audit of the 255-file upstream endpoint difference since the previously reviewed revision.

The prior optional documentation note about the intentional Canvas fill-mode exception to the shared resize-grip recipe remains unchanged; it is not a request to restore the grip. Public description, this PR’s commit metadata and all three attached screenshots were inspected with no disclosure finding.

Validation limits: source inspection only; no tests, builds, app launches or PR code executed, and CI was not assessed. Browser/native layout, short-pane and error-state visibility, rendered keyboard/focus behavior, enlarged text, performance, live-relay behavior and human acceptance remain unverified by this review. Author-reported QA was not rerun. This COMMENT is not approval or merge authorization.

[jameskraus]

🤖 Codex: Addressed the description/readiness mismatch. The description no longer calls this PR draft; its existing ready state is preserved. Before/after screenshots were recaptured from the current base 5a7b41bf2e and head 87dac25831. The optional design-guide note about the resize-grip exception is still deferred.