Ordering in relational databases sparked a classic debate in 2015. The question—how to store ordered information in a Relational Database—put SQL modeling under the microscope [1].
2015 Relational Dilemmas The 2015 thread framed the tension between native SQL ordering and how to encode sequence in tables. That debate foreshadowed a shift toward CRDT-driven local-first strategies [2].
CRDT-Driven Local-First Shift Today, CRDTs push ordering concerns into local-first designs. In SQLite, projects like crsql and vlcn aim for deterministic merges and conflict resolution [2].
Conflict Handling and Architecture CRDT approaches often pair deterministic merges with a hash-based tamperproof ledger of changes, supporting eventual convergence [2]. They also grapple with concurrency clocks such as Lamport clocks and vector clocks to reason about edits. Fireproof offers a non-SQL path—deterministic defaults and a ledger, in a world more like CouchDB or MongoDB but with cryptographic integrity [2].
Takeaway for 2025 From 2015’s SQL-ordering puzzles to 2025’s local-first SQLite, the arc is clear: convergence beats chaos when data flows across devices. Keep an eye on how crsql and vlcn evolve in SQLite [2].
References
How to store ordered information in a Relational Database (2015)
Question on methods to preserve order within relational databases; seeks approaches to model ordered data
View sourceCRDT and SQLite: Local-First Value Synchronization
Explores CRDTs for local-first SQLite, conflict resolution, deterministic merges, clock issues, and comparisons to Fireproof, DuckDB, CouchDB, MongoDB, crsql, vlcn.
View source