Problem
A SQL query can succeed via Atelier REST (iris_query) but fail when executed inside an embedded Python method. There's no tool to predict this before wiring up code, leading to hard-to-diagnose runtime failures deep in production code.
Proposed Tool: iris_validate_embedded_sql
iris_validate_embedded_sql("SELECT node_id FROM Graph_KG.nodes")Returns:
{
"sql": "SELECT node_id FROM Graph_KG.nodes",
"works_external": true,
"works_embedded": false,
"reason": "Column resolution on class-projected table triggers ddtab bug in build 161",
"workaround": "Use iris.sql.exec() with %NOINDEX hint, or query the backing global directly"
}Implementation Notes
- Run the query in both contexts (external via Atelier REST, embedded via
execute_via_generator)
- Compare results / errors
- Cross-reference with
iris_build_constraints (see related issue) for known build-specific issues
- Could be a test mode of
iris_query rather than a separate tool: iris_query(query, validate_embedded=true)
Value
Saves hours of debugging when queries mysteriously fail only in embedded context. Especially valuable for IRIS 2026.x where embedded Python SQL behavior changed significantly between builds.
Problem
A SQL query can succeed via Atelier REST (
iris_query) but fail when executed inside an embedded Python method. There's no tool to predict this before wiring up code, leading to hard-to-diagnose runtime failures deep in production code.Proposed Tool: iris_validate_embedded_sql
iris_validate_embedded_sql("SELECT node_id FROM Graph_KG.nodes")Returns:
{ "sql": "SELECT node_id FROM Graph_KG.nodes", "works_external": true, "works_embedded": false, "reason": "Column resolution on class-projected table triggers ddtab bug in build 161", "workaround": "Use iris.sql.exec() with %NOINDEX hint, or query the backing global directly" }Implementation Notes
execute_via_generator)iris_build_constraints(see related issue) for known build-specific issuesiris_queryrather than a separate tool:iris_query(query, validate_embedded=true)Value
Saves hours of debugging when queries mysteriously fail only in embedded context. Especially valuable for IRIS 2026.x where embedded Python SQL behavior changed significantly between builds.