[Bug] --empty doesn't work when using get_columns_in_relation

[georgea-tk]

Is this a new bug in dbt-snowflake?

• I believe this is a new bug in dbt-snowflake
• I have searched the existing issues, and I could not find an existing issue for this bug

Current Behavior

The --empty flag changes the relation name to be something like this:
(select * from DATABASE_NAME.SCHEMA_NAME.TABLE_NAME where false limit 0)
This becomes an issue when the adapter get_columns_in_relation function is called as it tries to do something like this, which is invalid snowflake sql:
describe table (select * from DATABASE_NAME.SCHEMA_NAME.TABLE_NAME where false limit 0)

This means we get the following sql errors when using functions like dbt_utils.star:
Syntax error: unexpected '('. (line 1)

Expected Behavior

anywhere where the {{ relation }} is used after a describe statement, this should be handled differently for the --empty flag to not be the subquery, but instead use the default original relation.
Like here:
https://github.com/dbt-labs/dbt-snowflake/blob/main/dbt/include/snowflake/macros/adapters.sql#L30

Steps To Reproduce

1. dbt-snowflake 1.8
2. running with the --empty command
3. on any model which includes the {{ dbt_utils.star() }} macro

Relevant log output

No response

Environment

- OS: Linux amd64
- Python: 3.10
- dbt-core: 1.8.0
- dbt-snowflake: 1.8.0

Additional Context

No response

[jtcohen6]

Thanks for opening @georgea-tk. I agree this is something we need to fix to fully support --empty runs.

An immediate fix I can think of is updating snowflake__get_columns_in_relation from {{ relation }} to {{ relation.render() }}, which will bypass the conditional logic here.

{% macro snowflake__get_columns_in_relation(relation) -%}
  {%- set sql -%}
    describe table {{ relation.render() }}
  {%- endset -%}
  ...
{% endmacro %}

My concern with such a fix is that we'll need to make it in a variety of places, anywhere that a ref() could conceivably be passed. I think get_columns_in_relation is one of the most common, more than drop or create_view_as, but we should aim for consistency.

[kokorin]

Not sure if it will be useful, but we have the same bug with relation created using DBT's source function

[FishtownBuildBot]

Opened a new issue in dbt-labs/docs.getdbt.com: dbt-labs/docs.getdbt.com#5732

[justin-fundrise]

When will this bug fix be released? Thanks!