Evaluate and eliminate remaining 3 effect blocks in BaseNodeViewer

[malibio]

Overview

Evaluate whether the 3 remaining $effect blocks in base-node-viewer.svelte can be eliminated or must remain as legitimate side effects.

Background

Issue #628 successfully reduced effects from 9 → 4 → 3 by:

• Converting state to $derived (viewerPlaceholder, currentViewedNode)
• Moving async loading from $effect to onMount (leverages {#key} component recreation)
• Eliminating redundant watchers

Remaining Effects

1. Content Save Watcher (Line ~437) - $effect.pre

Purpose: Automatically saves node content to database when users edit

Why it might be legitimate:

• Must run BEFORE DOM updates ($effect.pre)
• Performs side effects (database writes)
• Coordinates ancestor persistence to prevent FOREIGN KEY errors
• Detects new vs existing nodes for immediate vs debounced saves

Potential elimination:

• Move save logic to event handlers (on:input, on:blur)
• Use store subscriptions instead of effect watching
• Evaluate if this duplicates work already done by updateNodeContent()

2. Node Deletion Detection (Line ~532) - $effect

Purpose: Detects when nodes are deleted and cleans them up from database

Why it might be legitimate:

• Compares previous vs current state to detect deletions
• Performs side effects (API calls to delete nodes)
• Coordinates with structural watcher to prevent CASCADE errors

Potential elimination:

• Move to explicit event handlers for delete operations
• Use store subscriptions for deletion detection
• Evaluate if deletion should be explicit action vs implicit detection

3. Structural Change Tracker (Line ~626) - $effect.pre

Purpose: Tracks node hierarchy for cleanup and persistence coordination

Why it might be legitimate:

• Must run BEFORE DOM updates ($effect.pre)
• Maintains tracking Sets for deletion detection
• Coordinates cleanup for removed nodes

Potential elimination:

• Move structural tracking to the store layer
• Use explicit hierarchy management instead of watching
• Evaluate if this duplicates work done elsewhere

Investigation Questions

1. Do these effects duplicate work?

   • Is content saving already handled by updateNodeContent()?
   • Is deletion already handled by explicit delete handlers?
   • Is structure tracking already done by the store?

2. Can we use event-driven patterns instead?

   • Can content saves trigger on explicit events (blur, debounced input)?
   • Can deletions be explicit actions rather than detected changes?
   • Can structure changes be tracked via store subscriptions?

3. Should this logic live in the component?

   • Could content coordination move to the store?
   • Could deletion detection move to a service?
   • Could structure tracking be centralized?

Acceptance Criteria

• Each effect analyzed for necessity
• Alternative patterns evaluated (events, subscriptions, derived state)
• Decision documented for each effect (keep vs eliminate)
• If eliminated: refactored with tests passing
• If kept: documented why it's legitimately necessary
• Architecture improved (effects in component vs store/service)

Success Metrics

Best case: 0 effects (all moved to better patterns)
Realistic: 1-2 effects (truly necessary side effect coordination)
Documentation: Clear explanation of why remaining effects can't be eliminated

Related

• Issue Audit and simplify base-node-viewer effects #628 - Parent issue (now completed with 3 effects remaining)
• Issue Epic: Eliminate $effect Anti-patterns - Embrace Svelte 5 Reactivity #623 - Parent epic for effect anti-patterns
• Commit e172e19 - onMount refactoring that eliminated async loading effect