Integer type mishandled in DDL statements

Athena uses different names for the integer type depending on whether the statement is a DML or DDL one. In the former it uses the regular INTEGER name, but in DDL it uses the name INT.

Currently PyAthena does not distinguish DDL from DML statements, which produces syntax errors when emitting CREATE TABLE statements.

Note that in other places (e.g. pyathena.pandas.utils) this has been correctly handled.

[cansjt]

I am sorry but we did have syntax errors with the INTEGER type at some point.

Maybe it also depends on the storage format from some weird reason, or something else. I am looking into it with a colleague to provide an example. If we find out that we had it wrong and cannot reproduce the issue, we’ll close the issue obviously. Just give us the day, please. Thanks.

[cansjt]

For memory’s sake: some DDL statements do not handle INTEGER, like:

ALTER TABLE default.some_table ADD COLUMNS (some_data INT);

This is not something that is yet implemented in PyAthena, but we have started implementing some Alembic support for it.