We've noticed that the SUBSTRING function complies with BigQuery "standard SQL" semantics but not ISO standard SQL semantics. I am proposing to fix it, and that necessitates some breaking changes that I have outlined in https://issues.apache.org/jira/browse/CALCITE-4427.
If you care about this, please weigh in today. My time is limited and it is not efficient to change the specification during PR review. Julian
