expect_column_proportion_of_unique_values_to_be_between throwing a compilation error

[manuel-niederl]

Hello! Not sure if this issue is widespread or just due to my installation but the change made to this test in 10.3 is breaking our existing implementation of the test. I'm running dbt 1.2.6.

This line was changed from

count(distinct {{ column_name }})*1.0/count({{ column_name }})

to

cast(count(distinct {{ column_name }}) as {{ dbt.type_float() }})/count({{ column_name }})

which is throwing this error on our end:

dbt.exceptions.CompilationException: Compilation Error in test dbt_expectations_expect_column_proportion_of_unique_values_to_be_between_dim_invite_experiments_guest_id__1__0_99__guest_id_is_not_null (models/invite/core/invite-experimentation/_invite_experimentation.yml)
  'dict object' has no attribute 'type_float'

Please advise if this is a bug or an issue with my installation.

Update: Updating to DBT 1.3.7 fixed this issue but I still believe that I should not have encountered the bug because expectations supports lower DBT versions.

[clausherther]

I don't remember which version of dbt brought the jinja type macros into the dbt namespace but I assume the version of dbt you're using is too old. The current stable release is 1.7.x.

https://docs.getdbt.com/reference/dbt-jinja-functions/cross-database-macros#type_float

We only test and therefore only really support the latest stable release minor version, so currently 1.7.x.

[clausherther]

FYI, I updated the README to reflect the supported version.

[manuel-niederl]

Thanks a lot for the quick answer and for updating the Readme!