OptionaldefaultFallback ordering columns — raw, unquoted identifier names from MJ metadata (the entity's primary key) — the driver applies ONLY when orderBy is absent, to make offset pagination deterministic. Each driver applies them per its own dialect: SQL drivers quote each identifier (so mixed-case / reserved-word PK columns resolve on case-sensitive dialects); MongoDB uses them as raw field names. Trusted (PK names from introspected metadata), so NOT clause-screened.
OptionalfieldsColumns to project. Empty/undefined = all columns.
OptionalfilterFilter expression in the data source's FilterDialect (a SQL WHERE body, or a driver-translated AST for non-SQL sources).
OptionalincrementalStructured incremental lower-bound: return only rows where Field >= Value (INCLUSIVE). The driver
renders this to its own dialect — SQL drivers to a quoted WHERE <Field> >= <literal> using their own
identifier quoting, MongoDB to { <Field>: { $gte: <coerced> } } — so an incremental-sync caller
NEVER writes dialect SQL. Combined (ANDed) with filter when both are present. Inclusive by
design so a caller can re-read the watermark boundary and rely on downstream dedup rather than risk
skipping a row whose sub-second timestamp truncates to the saved value.
OptionalmaxMaximum rows to return (page size).
Remote table/view/collection name; resolved against the data source DefaultSchema/DefaultDatabase when unqualified.
OptionaloffsetZero-based row offset for pagination.
OptionalorderOrder-by expression in the data source's dialect (caller-supplied; screened for read-only safety).
RunView-equivalent request against a single remote object.