I have an SQLAlchemy table with one primary key column and one column with a default value (sqlalchemy.sql.functions.now()).
Because both columns in the table have a default value (from a sequence or the PostgreSQL now() function), I should be able to insert a row without a values argument, or with an empty mapping, e.g. database.execute(table.insert()) or database.execute(table.insert(), values={}). But inserting a row either way raises a NotNullViolationError:
asyncpg.exceptions.NotNullViolationError: null value in column "id" of relation "post" violates not-null constraint
DETAIL: Failing row contains (null, null).
It seems that a default value isn't being generated for the ID column, which is violating its not-null constraint.
The exception isn't raised and the row is inserted if I pass a value, e.g. database.execute(table.insert(), values={"posted_at": datetime.datetime.now()}), but that defeats the purpose of defining a default value in the database schema.
mathause
commented
1 year ago
I have an SQLAlchemy table with one primary key column and one column with a default value (
sqlalchemy.sql.functions.now()).Because both columns in the table have a default value (from a sequence or the PostgreSQL
now()function), I should be able to insert a row without avaluesargument, or with an empty mapping, e.g.database.execute(table.insert())ordatabase.execute(table.insert(), values={}). But inserting a row either way raises aNotNullViolationError:It seems that a default value isn't being generated for the ID column, which is violating its not-null constraint.
The exception isn't raised and the row is inserted if I pass a value, e.g.
database.execute(table.insert(), values={"posted_at": datetime.datetime.now()}), but that defeats the purpose of defining a default value in the database schema.