Fix: Set dbt execute variable for parse time and run time accordingly

[themisvaltinos]

This update sets the dbt variable execute to False during parse-time and true for run time. Dbt models and macros have conditional logic which is guarded with this flag for runtime execution:

{% if execute %}
  {% set result = run_query("SELECT COUNT(*) FROM my_table") %}
  SELECT {{ result.columns[0][0] }} as count
{% endif %}

These jinja expressions might depend on the data warehouse or to objects (like graph) not available at parse time, so the execute ensure these only should run at the runtime stage.

dbt docs: https://docs.getdbt.com/reference/dbt-jinja-functions/execute

[izeigerman]

No, we should not do this. execute is set unconditionally to True for a reason. Since we used the rendered query for change detection and categorization we want to follow the execution path during rendering. If we happen to encounter the adapter call we fail and return None which indicates that the query cannot be rendered and categorized.

[themisvaltinos]

Since we used the rendered query for change detection and categorization we want to follow the execution path during rendering. If we happen to encounter the adapter call we fail and return None which indicates that the query cannot be rendered and categorized.

I see was wondering about this, thanks that makes sense. and I guess now that we can fail and still proceed later to runtime this can't lead to any issues

[izeigerman]

Why isn't graph available at load time?

[themisvaltinos]

Why isn't graph available at load time?

I was mirroring dbt, but since we already have the manifest loaded you're right it will be available at "our parse time", it's a different distinction. I will proceed to add it