Qstack Separate Before Serializing Shared State

Separate before serializing shared state

Instructions and conventions are not concurrency control. Start by removing the shared write target. Add serialization only when one canonical writer is a real domain invariant.

Decide from ownership

1. Inventory every mutable object and every actor that can write it. Include files, branches, database rows, keys, caches, queues, and in-memory state.
2. Ask whether the actors publish independent facts. If they do, give each actor its own file, key, branch, or state directory and merge only at a read-only reporting boundary.
3. Confirm that the split removes shared mutation. Two workers writing separate fields of one JSON document still share a write target; separate documents with one owner each do not.
4. If one shared object is required, choose structural serialization: a single-writer actor, sequential phases, a lockfile with defined stale-lock behavior, an atomic transaction, or compare-and-swap.
5. Prove the design under overlap, retry, crash, and stale-state scenarios.

Report the ownership map, the sharing that can be removed, any invariant that requires serialization, and the concrete mechanism enforcing it. A request for analysis does not authorize changing the system. Assess compatibility and migration before changing a persisted format, public interface, or branch workflow. It also does not authorize commits, pushes, publishing, deployment, or external messages.

Adapted from Lauren Tan's PStack principle-separate-before-serializing-shared-state at commit 60c641e4fad674784b30abcf9f8915dea39df38d under the MIT License. See third-party notices.