Short answer: this is expected behavior. Elementary normalizes Snowflake/Postgres text types, so VARCHAR/CHARACTER VARYING (and NVARCHAR/STRING on Snowflake) are treated as the same “TEXT” family rather than comparing length or the exact warehouse-specific spelling. That’s why you see CHARACTER VARYING in the test output even if your contract shows VARCHAR(1024). On Snowflake, Elementary explicitly maps STRING, VARCHAR, and NVARCHAR to TEXT, so a change like VARCHAR(MAX) to VARCHAR(x) doesn’t flag by default. If you need stricter length/type enforcement, the knowledge sources don’t document support for that today (and Databricks/Spark aren’t supported for these schema tests). ([Schema changes from baseline](https://docs.elementary-data.com/data-tests/schema-tests/schema-changes-from-baseline)). Related notes from prior discussions: - In Snowflake, VARCHAR vs TEXT are treated synonymously by Elementary; datatype changes like VARCHAR(MAX) → VARCHAR(x) aren’t flagged by default (support confirmed this behavior) ([Slack thread](https://slack.com/archives/C02CTC89LAX/p1689814382480289)). - Historical issues where VARCHAR vs CHARACTER VARYING caused false “type changed” messages were addressed in later versions; ensure you’re on an updated Elementary package if you still see mismatches purely due to naming differences ([Slack thread](https://slack.com/archives/C02CTC89LAX/p1679319966726349)). If your warehouse is Databricks/Spark, schema changes tests aren’t supported (except Databricks with Unity Catalog, per docs), which can also explain unexpected results ([Docs](https://docs.elementary-data.com/data-tests/schema-tests/schema-changes-from-baseline), [Slack note](https://slack.com/archives/C02CTC89LAX/p1695800200373959)).

I’m sorry, but the knowledge sources don’t state the exact version where the VARCHAR vs CHARACTER VARYING normalization/mismatch was fixed. They mention that: - Elementary treats text-type synonyms as equal (e.g., VARCHAR/TEXT/NVARCHAR/STRING) and does not flag length differences like VARCHAR(MAX) vs VARCHAR(x) by default, but no specific version is given for when this behavior was introduced or fixed ([Schema changes](https://docs.elementary-data.com/data-tests/schema-tests/schema-changes)). - A related mismatch issue was discussed and later addressed “in one of the latest versions,” but without a precise version number; users were advised to run dbt deps to update and retest ([Slack thread](https://slack.com/archives/C02CTC89LAX/p1679319966726349)). - Another dashboard counting bug was fixed in Python 0.7.5 / dbt package 0.7.2, but that’s unrelated to type synonym handling ([Slack thread](https://slack.com/archives/C02CTC89LAX/p1676946903124339)). If you can share your current Elementary dbt package version, I can suggest upgrading to the latest available in your project and re-running dbt deps, as prior Slack guidance resolved similar schema type display issues by upgrading ([Slack thread](https://slack.com/archives/C02CTC89LAX/p1679319966726349)).