DuckDB Schema: Columns beginning with _ cause binder error when missing from data

[ahiadnissimclutch]

When a field in the schema (e.g., _id) is defined but does not appear in the actual data, I receive the following error:

duckdb.duckdb.BinderException: Binder Error: Column "_id" referenced that exists in the SELECT clause - but this column cannot be referenced before it is defined

However, this error does not occur for other fields without an underscore; those fields seem to be handled properly even if they are missing from the data

[ahiadnissimclutch]

Follow Up:

When I tried to escape schema field names that start with an underscore (like "_id") using quotes, the dataframe processed without an exception, but the results were wrong.
The _id column just shows the field name (“_id”) in every row, instead of the actual data.
When ive tried to use the official DuckDB SDK, I get the correct results.

Schema:
{ '"_id"': "varchar", "name": "varchar"}
Data:

[
  { "_id": "1", "name": "xxx" },
  { "_id": "2", "name": "yyy" }
]

DF Result:
+-----+------+
| _id | name |
+-----+------+
| _id | xxx |
| _id | yyy |
+-----+------+