Semantic prototype scope
The executable artifact covers the finite semantic model through Slice 5. Operational runtime and server behavior sit outside this evidence boundary.
Limitations
The boundary of the CatDB thesis
The narrow thesis is the useful one. CatDB puts integration meaning into typed mappings; it does not make the physical or distributed work disappear.
Storage and execution
The papers specify SQLite, PostgreSQL, connectors, federation, materialization, and distributed execution as separate systems layers.
Performance evidence
Performance claims require the published protocol, versioned harnesses, datasets, raw runs, environments, and statistical analysis.
Formal coverage
Lean checks selected path and mapping invariants. Migrations, query rewrites, storage behavior, and engine correctness require their own formal targets.
Conditional integration reuse
The N plus M reuse argument depends on adequate shared-domain mappings. Exceptions, ambiguity, loss, and domain maintenance remain.
Distributed tradeoffs
Coordination, availability, latency, and convergence remain explicit design choices. Some schema changes do not commute.
Identity needs evidence and authority
Similarity can produce candidates. It does not establish semantic identity or decide which source may authorize a merge.
MDReader case study
Paper P18 provides the application model and implementation-entry contract. Product behavior requires its own accessibility, security, user, and performance studies.
Next systems study
The systems evaluation joins storage adapters, query planning, materialization, and distributed execution to the preregistered benchmark protocol, preserving raw runs and failed trials.