Roadmap

Roadmap

The goal: any SQL database to any SQL database, from a local app you run yourself.

SQL Server → PostgreSQL is the first pair because it was needed first, not because the tool is about those two products. Everything provider-specific is isolated behind a visible seam; everything about migrating safely is shared and stays shared.

Order of work

1. Promote the provider seam to interfaces — ISourceReader / ITargetWriter, extracted just before the second provider lands, so the abstraction is informed by what actually varies.
2. A second source: MySQL / MariaDB — chosen because it stresses the seam differently (different information_schema dialect, type names, identity semantics).
3. A second target: SQL Server — bidirectionality forces PostgreSQL specifics (COPY, session_replication_role, setval) to become capabilities a provider either has or emulates.
4. Self-contained builds — shipped: every tagged release publishes single-file executables per platform; see Getting Started.
5. Table selection — migrate a subset, with the safety analysis that refuses a truncate whose cascade would reach outside the subset.
6. Checksum verification — optional user-defined aggregate comparisons (sums over numeric columns) so a migration can prove money totals match, not merely that rows arrived.

Deliberately out of scope

• Schema creation/translation — a separate problem with its own failure modes. Bring your own schema.
• Continuous replication — this is a cutover tool; CDC pipelines are a different product.
• Data transformation — reshaping during migration makes a failure indistinguishable from a transformation bug. Move faithfully, transform afterwards.

Full reasoning: docs/ROADMAP.md.