Playground: Evaluate PGlite

[flyingrobots]

Problem Statement

Instead of using pg-mem, we'll be using PGlite. Why? PGlite is Postgres compiled to WASM, pg-mem is emulated.

Factor	PGlite	pg-mem
Fidelity to real Postgres (DDL, extensions, behaviours)	High	Lower
In-browser + persistence support	✔ (IndexedDB etc)	✔ (in-memory + browser)
Production-style or complex schema support	Good	Probably risky
Complexity of setup/integration	Moderate (you’ll still test carefully)	Low (lighter)
Use case fit (your GraphQL→DDL→instance scenario)	Strong	Weak to moderate
Risk of divergence when you “promote” to real Postgres	Lower	Higher

In-Scope

• Lean how to integrate with Wesley Website deployments
• Confirm that it works
• Confirm that it supports common pg extensions
• Try standing up a moderately complex DDL
• Try migration
• Try some queries

Out-of-Scope

• Integrate deployment with Wesley Wesbite (will be its own task)

Proposed Solution

1. Install PGlite
2. Evaluate suitability for Wesley Website integration/Wesley in the browser demo needs

Alternatives Considered

• pg-mem but chosen over this because it's more fully featured

Impact Areas

• CLI
• Docs
• API
• Infra/CI
• Other

Priority

None