Reputation: 713
There was a change in the type of Column of one of the tables from Integer to String.
Logic(PBase):
__tablename__ = "logic"
Id(Integer, primary_key=True)
this column changed to string
Logic(PBase):
__tablename__ = "logic"
Id(String, primary_key=True)
Now I am using alembic to auto-generate the migration script. To detect type change, I have supplied compare_type=True in env.py
with connectable.connect() as connection:
context.configure(
connection=connection,
target_metadata=target_metadata,
compare_type=True
)
Doing so, the migration script is generated fine. Here is the content of the generated migration script:
from alembic import op # noqa
import sqlalchemy as sa # noqa
def upgrade():
### commands auto generated by Alembic - please adjust! ###
op.alter_column('logics', 'id',
existing_type=sa.INTEGER(),
type_=sa.String())
### end Alembic commands ###
def downgrade():
### commands auto generated by Alembic - please adjust! ###
op.alter_column('logics', 'id',
existing_type=sa.String(),
type_=sa.INTEGER())
### end Alembic commands ###
But when I run the upgrade command I get a syntax error (here is the end part of traceback):
File "/Users/amit/.virtualenvs/be_new/lib/python2.7/site-packages/sqlalchemy/engine/default.py", line 436, in do_execute cursor.execute(statement, parameters) sqlalchemy.exc.OperationalError: (OperationalError) near "ALTER": syntax error u'ALTER TABLE logics ALTER COLUMN id TYPE VARCHAR' ()
What maybe the issue here?
Upvotes: 7
Views: 8672
Reputation: 23134
I had a similar problem which I solved with the following syntax:
def upgrade():
### commands auto generated by Alembic - please adjust! ###
op.alter_column('logics', 'id', type_=sa.String, existing_type=sa.INTEGER)
### end Alembic commands ###
def downgrade():
### commands auto generated by Alembic - please adjust! ###
op.alter_column('logics', 'id', type_=sa.INTEGER, existing_type=sa.VARCHAR)
### end Alembic commands ###
Essentially I omitted the parentheses after the type declarations and on the downgrade
I used the type VARCHAR
because String
is not an SQL type.
Upvotes: 7