[Bug] Incremental strategy `ignore` does not ignore missing columns, throws error instead

[bruno-szdl]

Is this a new bug?

• [X] I believe this is a new bug
• [X] I have searched the existing issues, and I could not find an existing issue for this bug

Current Behavior

In the docs it says

This is the behavior if on_schema_change: ignore, which is set by default, and on older versions of dbt.

If you add a column to your incremental model, and execute a dbt run, this column will not appear in your target table.

Similarly, if you remove a column from your incremental model, and execute a dbt run, this column will not be removed from your target table.

When I add a column it ignores it, but when I remove a column it raises an error. I tested it in 3 different adapters (snowflake, bigquery, duckdb) and it throws the same error. It says it can't find the missing column.

Expected Behavior

I expected it not to throw an error and just ignore it (add nulls maybe).

Steps To Reproduce

Run this model

{{ config(
    materialized='incremental'
    , on_schema_change='ignore'
) }}

select 
    1 as id
    , 100 as value
    , 'created' as status
    , cast('2024-01-01' as date) as date
union all
select 
    2 as id
    , 200 as value
    , 'processed' as status
    , cast('2024-01-02' as date) as date

Then remove a column and try rerunning it. In this case, I am also adding a new column

{{ config(
    materialized='incremental'
    , on_schema_change='ignore'
) }}

select 
    3 as id
    , 300 as value
    , True as is_processed
    , cast('2024-01-03' as date) as date

Relevant log output

Snowflake:
SQL compilation error: error line 3 at position 34
invalid identifier 'STATUS'

BigQuery:
Unrecognized name: status at [38:25]

DuckDB:
Binder Error: Referenced column "status" not found in FROM clause!

Environment

- OS: MacOS
- Python: 3.9.6
- dbt-adapters: dbt-snowflake, dbt-bigquery, dbt-duckdb

Additional Context

version 1.8

I also tested on 1.3 to test an older version, and same problem