#3689 fix(targets): honor 'schema' setting as alias for default_target_schema Pull request opened by Rasaec5 What When a SQL target sets a schema setting but does not set default_target_schema, the SDK now uses the configured schema as the destination schema instead of deriving one from the incoming stream ID. Why Closes #2766. Many SQL targets expose a schema connection setting (for example the MeltanoLabs variant of target-snowflake). Today, if default_target_schema is unset, SQLSink.schema_name derives the schema from the stream ID — e.g. a record on stream TAP_SCHEMA-users is written to a schema named tap_schema. When the connecting account has no privileges on that derived schema, schema/table/file-format creation fails. Users reasonably expect the schema they configured to be honored. How SQLSink.schema_name now resolves in this order: default_target_schema (unchanged — keeps precedence so existing setups are unaffected) schema (new fallback) stream-derived (- / --) None Implementation: Added a SQLSink.schema_config cached property that reads the schema setting and returns it verbatim (no name conforming), mirroring the existing default_target_schema property. schema_name falls back to schema_config before the stream-derived branch. This is intentionally non-breaking: default_target_schema still wins when both are set, and behavior is unchanged when neither is set. Files changed singer_sdk/sql/sink.py — add schema_config; update schema_name. tests/sql/test_sink.py — add test_schema_name_schema_setting parametrized cases. tests/sql/test_schema_alias.py — new dedicated test module for the alias behavior. Testing Unit tests run without external credentials using a dummy SQL connector/sink/target. Resolution-order matrix: schema overrides stream-derived for 1-, 2-, and 3-part stream names; default_target_schema takes precedence when both are set; only default_target_schema set; neither set (stream-derived fallback, including None for 1- and 4-part stream names). Edge cases: empty-string schema falls through; schema returned verbatim; table_name derivation unaffected. Regression: existing default_target_schema / stream-derived cases still pass. python -m pytest tests/sql/test_schema_alias.py tests/sql/test_sink.py -v tests/sql/test_sink.py -> 13 passed tests/sql/test_schema_alias.py -> 13 passed Open question for reviewers Should schema also be registered as a documented builtin setting on SQLTarget (an explicit alias of default_target_schema), or is reading it from config enough? This PR keeps it read-from-config to avoid colliding with targets that already define their own schema setting. Notes / checklist X I have read the Contributing Guide. X pre-commit run --all, nox -s tests, and nox -t typing pass locally. X This contribution was assisted by an AI coding agent (per Meltano's Agentic Coding guidelines). Summary by Sourcery Honor the

schema

configuration as a fallback destination schema for SQL sinks when

default_target_schema

is not set, while preserving existing precedence and stream-derived behavior. New Features: • Support the

schema

config setting as an alias/fallback for

default_target_schema

in SQLSink schema resolution. Enhancements: • Clarify and expand SQLSink docstrings for connector, connection, schema, table, and key properties accessors. Documentation: • Add a contributor README documenting the chosen issue, reproduction steps, implementation plan, and testing strategy for the schema alias change. Tests: • Add dedicated tests validating schema resolution precedence between

default_target_schema

,

schema

, and stream-derived names, including edge cases and regression coverage. meltano/sdk