Corrected the boolean parsing, but not sure how to handle TIMESTAMP WITH TIME ZONE:

SHOW COLUMNS FROM group_component;
+--------------+--------------------------+-------+----------------+-----------+
| Field | Type | Null | Default | Indices |
+--------------+--------------------------+-------+----------------+-----------+
| id | INT | false | unique_rowid() | {primary} |
| id_group | INT | false | NULL | {} |
| id_component | INT | false | NULL | {} |
| created_dt | TIMESTAMP WITH TIME ZONE | false | now() | {} |
| wall_sec | INT | true | NULL | {} |
+--------------+--------------------------+-------+----------------+-----------+

Thanks for the contribution! Can you add some tests that cover these cases?

As for TIMESTAMP WITH TIME ZONE, I think the time zone is handled at a different level so as far as SQLAlchemy's concerned it should be mapped to a plain TIMESTAMP. (This means changing _type_map from dict() with kwarg syntax to a dict literal so it can contain keys with spaces). This is what SQLAlchemy does in its postgres dialect.

Also, it might be a good idea to map both bool and boolean to sqltypes.BOOLEAN. I see [here][(https://github.com/zzzeek/sqlalchemy/blob/eaceeae0f721be72cc589a310e50f0de9f24d721/lib/sqlalchemy/dialects/postgresql/base.py#L1324) that in postgres it's apparently spelled boolean and so we might change CockroachDB to report boolean instead of bool for consistency.

Fixed in #82.