Why use self.type in function "convert_tbl_column_to_sqla_col"

[Mine-Echo]

function convert_tbl_column_to_sqla_col is at superset/models/helpers.py
https://github.com/apache/superset/blob/master/superset/models/helpers.py#L1403

    def convert_tbl_column_to_sqla_col(
        self,
        tbl_column: "TableColumn",
        label: Optional[str] = None,
        template_processor: Optional[BaseTemplateProcessor] = None,
    ) -> Column:
        label = label or tbl_column.column_name
        db_engine_spec = self.db_engine_spec
        column_spec = db_engine_spec.get_column_spec(self.type, db_extra=self.db_extra)
        type_ = column_spec.sqla_type if column_spec else None
        if expression := tbl_column.expression:
            if template_processor:
                expression = template_processor.process_template(expression)
            col = literal_column(expression, type_=type_)
        else:
            col = sa.column(tbl_column.column_name, type_=type_)
        col = self.make_sqla_column_compatible(col, label)
        return col

why here use self.type but not tbl.column? I think the latter is right.
column_spec = db_engine_spec.get_column_spec(self.type, db_extra=self.db_extra)
When debugging, I found self,type is "table", then column_spec will be None. I think self.type is not native type of the column, here we should pass tal_column. type, which is "STRING" in my test.
Is this really a bug in superset? Or did I just understand it wrong somewhere?

[dosubot[bot]]

Hey @Mine-Echo! How's it going? I'm here to help you with any questions or bugs you have in the Superset codebase. Let's tackle this together!

The use of self.type in the convert_tbl_column_to_sqla_col function is intentional and relates to the broader context of how the class is designed to interact with the database engine specifications. The self.type attribute is meant to represent the type of the datasource or the context in which the conversion is happening, rather than the specific type of the individual column being converted. This is why it is used to retrieve the column specification from the database engine.

In contrast, tbl_column.type represents the specific type of the column itself, such as "STRING". The design choice to use self.type instead of tbl_column.type might be to ensure that the conversion process aligns with the overall datasource type or context, which could affect how columns are interpreted or processed.

If self.type results in a None column specification, it might indicate that the datasource type does not have a direct mapping in the db_engine_spec.get_column_spec method, which could be a limitation or an area for improvement in handling specific cases. However, this does not necessarily mean it's a bug; it could be an intentional design choice that requires understanding the broader application context.

If you believe that using tbl_column.type would be more appropriate for your specific use case, it might be worth exploring how this change affects the behavior and whether it aligns with the intended design of the application.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}