[BLCCoreStudio]

Summary

Upgrades database/sqlite from a compact schema-v1 overview into a schema-v3 deep technical reference centered on SQLite's actual embedded architecture: SQL-to-bytecode execution, B-tree storage, pager/VFS boundaries, rollback journals, WAL, single-writer concurrency, query planning, local-file safety, backup, and security.

Changes

• explain the embedded-library boundary and why SQLite is architecturally different from client/server databases
• document tokenizer/parser/code-generation flow, VDBE bytecode, B-tree storage, pager/page cache, and VFS responsibilities
• cover rowid and WITHOUT ROWID tables, dynamic typing/type affinity, STRICT tables, indexes, EXPLAIN QUERY PLAN, and planner statistics
• add transaction modes, savepoints, SQLITE_BUSY, rollback-journal recovery, WAL snapshots, checkpointing, and the one-writer-per-database-file rule
• document WAL/network-filesystem constraints, live -wal/-shm/journal file handling, corruption failure modes, and filesystem safety boundaries
• add backup API, VACUUM INTO, integrity/recovery guidance, migration discipline, threading, mobile/desktop/server-local deployment patterns, and operational triage
• add security guidance for untrusted database files/SQL, parameter binding, extension loading, and defensive configuration
• cover JSON, FTS5, virtual tables, capacity/performance trade-offs, common mistakes, and beginner-to-advanced learning paths
• upgrade metadata to schema v3 with public-domain status, embedded/library/local deployment facets, Technology Universe coverage, and typed alternatives to PostgreSQL and MySQL
• expand the curated source set using SQLite's official documentation and canonical Fossil repository

Quality bar

The module treats SQLite as a pager/B-tree/VFS database library with file-backed transactional state, not as a miniature network RDBMS. It makes the single-writer model, rollback-vs-WAL mechanics, checkpoint behavior, rowid storage, planner/index trade-offs, live auxiliary files, filesystem assumptions, backup discipline, and corruption/security boundaries explicit enough to reason about real embedded and local-first systems.