Fix swimlane Kanban column grouping when Group By is a formula

[bailob]

Description

This PR fixes a glitch in the Bases Kanban swimlane mode when the view is grouped by a formula.* property and swimlanes are enabled (e.g. swimlane = Status).

Bug

After editing a task’s metadata (e.g. changing scheduled so the formula result moves from “Overdue” to “Today”), the card would immediately jump into the None column in swimlane mode, and it would only correct itself after an Obsidian force reload or after changing the Bases sort (which forces a re-run).

Flat (non-swimlane) Kanban mode behaved correctly.

Root Cause

In swimlane mode, the column key was re-derived from pathToProps (including cached formula.* outputs). After a frontmatter edit, Bases’ groupedData can update promptly, while cached formula outputs can lag, causing a temporary mismatch and a fallback to None.

Fix

In renderWithSwimLanes(), column assignment now always uses the already-built groups map (which is derived from Bases groupedData). Swimlanes are then applied on top of that using the swimlane property value. This keeps swimlane mode consistent with flat mode and with Bases’ computed grouping.

How I tested

• Obsidian desktop
• Bases Kanban view
• Group By: formula property (e.g. scheduled bucket formula)
• Swimlane: status
• Edited task scheduled so it changes buckets
• Verified the card stays in the correct column immediately (no reload / sort toggle needed)

Note

This change was produced as a pure “vibe coding” patch and submitted via OpenAI Codex.

[callumalpass]

Thank you for this PR, @bailob . Very much appreciated!

[callumalpass]

Merged manually via local merge and push to main. See commit 4201a52.