DatabricksAdapter::list_relations_schemas_by_patterns

Error message

DatabricksAdapter::list_relations_schemas_by_patterns

What it means

Databricks' `list_relations_schemas_by_patterns_inner` is a `todo!()` placeholder; pattern-based relation/schema listing has not been implemented for DatabricksAdapter. Calling the corresponding metadata API panics with 'DatabricksAdapter::list_relations_schemas_by_patterns'.

Source

Thrown at crates/dbt-adapter/src/metadata/databricks/mod.rs:1393

            Ok(schema)
        };
        let reduce_f = |acc: &mut Acc,
                        relation: Arc<dyn BaseRelation>,
                        schema: AdapterResult<Arc<Schema>>|
         -> Result<(), Cancellable<AdapterError>> {
            acc.insert(relation.semantic_fqn(), schema);
            Ok(())
        };
        let map_reduce = MapReduce::new(factory, Box::new(map_f), Box::new(reduce_f), unique_id);
        map_reduce.run(Arc::new(relations.to_vec()), token)
    }

    fn list_relations_schemas_by_patterns_inner(
        &self,
        _patterns: &[RelationPattern],
        _token: CancellationToken,
    ) -> AsyncAdapterResult<'_, Vec<(String, AdapterResult<RelationSchemaPair>)>> {
        todo!("DatabricksAdapter::list_relations_schemas_by_patterns")
    }

    fn freshness_inner(
        &self,
        relations: &[Arc<dyn BaseRelation>],
        token: CancellationToken,
    ) -> AsyncAdapterResult<'_, BTreeMap<String, MetadataFreshness>> {
        if relations.is_empty() {
            return Box::pin(async { Ok(BTreeMap::new()) });
        }

        type Acc = BTreeMap<String, MetadataFreshness>;

        let factory = Box::new(AdapterConnectionFactory::new(
            self.adapter.engine().clone(),
            self.adapter.engine().threads(),
        ));

View on GitHub (pinned to 0267ce9170)

Solutions

Implement the method for DatabricksAdapter, e.g., querying `information_schema.tables`/`schemata` filtered by the RelationPattern set
Use implemented per-relation metadata calls instead of the pattern-listing API until this lands
Track the missing feature upstream

Example fix

// before
fn list_relations_schemas_by_patterns_inner(
    &self,
    _patterns: &[RelationPattern],
    _token: CancellationToken,
) -> AsyncAdapterResult<'_, Vec<(String, AdapterResult<RelationSchemaPair>)>> {
    todo!("DatabricksAdapter::list_relations_schemas_by_patterns")
}
// after
fn list_relations_schemas_by_patterns_inner(
    &self,
    patterns: &[RelationPattern],
    token: CancellationToken,
) -> AsyncAdapterResult<'_, Vec<(String, AdapterResult<RelationSchemaPair>)>> {
    self.query_information_schema_by_patterns(patterns, token)
}

Defensive patterns

Strategy: validation

Validate before calling

if adapter.name() == "databricks" {
    return Err(anyhow!("Pattern-based relation listing is not implemented for Databricks"));
}

Type guard

fn supports_pattern_listing(adapter: &dyn Adapter) -> bool {
    adapter.capabilities().list_relations_schemas_by_patterns
}

Try / catch

match std::panic::catch_unwind(AssertUnwindSafe(|| adapter.list_relations_schemas_by_patterns(patterns))) {
    Ok(res) => res,
    Err(_) => per_relation_listing_fallback(patterns),
}

Prevention

Confirm Databricks pattern-listing support before bulk metadata calls
Use information_schema queries directly for Databricks schema discovery as a fallback
Gate bulk cache warm-up flows on adapter capabilities
Track upstream Databricks metadata implementation

When it happens

Trigger: Invoking `list_relations_schemas_by_patterns` against a Databricks adapter, dispatching into this unimplemented inner method.

Common situations: A Databricks project performs a bulk relations/schema lookup by patterns (catalog refresh, relation cache warm-up) and hits the panic on first use of that API path.

Related errors

AI-assisted analysis of dbt-labs/dbt-core@0267ce9170 (2026-09-07). Data as JSON: /api/errors/d18c020a6854129b. Report an issue: GitHub.