Class serial
source code
object --+
|
property --+
|
datatypes.datatype --+
|
datatypes.integer --+
|
serial
Datatype for automatically gernerated ids.
From The FireBird FAQ
How can i make an auto-incrementing primary key column? and is the NOT NULL constraint mandatory for a primary key?
In Firebird, you achieve an auto-incrementing PK using a generator
and a BEFORE INSERT trigger:
CREATE GENERATOR GEN_PK_ATABLE;
COMMIT;
Define the column, e.g., ATABLE_ID, as BIGINT or INTEGER and yes,
the NOT NULL constraint is mandatory for a primary key.
The trigger for automatic population of the key is this:
CREATE TRIGGER BI_ATABLE FOR ATABLE
ACTIVE BEFORE INSERT
AS
BEGIN
IF(NEW.ATABLE_ID IS NULL) THEN
NEW.ATABLE_ID = GEN_ID(GEN_PK_ATABLE, 1);
END
This is NOT what orm does! It does...
How can I get the value of my generated primary key after an insert ?
Firebird doesn't currently return values from insert statements and,
in multi-user, it isn't safe to query the generator afterwards
(GEN_ID(Generator_name, 0) because you have no way to know whether the
current 'latest' value was yours or someone else's. The trick is to get
the generator value into your application as a variable before you post
your insert. This way, you make it available not just for your insert
but for any dependent rows you need to create for it inside the same
transaction:
SELECT GEN_ID(Generator_name, 1) AS MyVar FROM RDB$DATABASE;
Some data access interfaces and drivers provide ways to do this
automatically for you. For example, IB Objects implements the
GeneratorLinks property for statement classes.
How it is handeld by orm
orm is the kind of interface that does this automatically for you.
For any serial (and in fact common_serial, see above) column it will query the
corresponding sequence and store the resulting value in the right
attribute to be used in the INSERT statement. If you use a serial as
your primary key, the select_after_insert() mechanism (see
orm2.datasource.datasource) will work correctly.
|
__init__(self,
column=None,
sequence=None,
title=None,
validators=(),
widget_specs=())
|
|
__init_dbclass__(self,
dbclass,
attribute_name)
This methods gets called by dbobject's metaclass.
|
|
__set__(self,
dbobj,
value)
Set the attribute managed by this datatype class on instance to
value.
|
|
Inherited from datatypes.datatype:
__convert__,
__delete__,
__get__,
__select_after_insert__,
__select_this_column__,
__set_from_result__,
__setattr__,
add_widget,
check_dbobj,
data_attribute_name,
isset,
sql_literal,
widget_specs
Inherited from property:
__getattribute__,
__new__
Inherited from object:
__delattr__,
__hash__,
__reduce__,
__reduce_ex__,
__repr__,
__str__
|
|
Inherited from property:
fdel,
fget,
fset
|
|
Inherited from object:
__class__
|
__init__(self,
column=None,
sequence=None,
title=None,
validators=(),
widget_specs=())
(Constructor)
| source code |
-
- Parameters:
sequence - Either a string or a sql.identifyer instance, naming the
sequence to be used by this serial column. Defaults to
GEN_PK_<relation name>
- Overrides:
datatypes.datatype.__init__
|
__init_dbclass__(self,
dbclass,
attribute_name)
| source code |
This methods gets called by dbobject's metaclass. It supplies the db
property with info about the class it belongs to and its attribute
name.
-
- Overrides:
datatypes.datatype.__init_dbclass__
- (inherited documentation)
|
Set the attribute managed by this datatype class on instance to value.
This will be called by Python on attribute assignment. The
__set_from_result__ method does the same thing for data retrieved from
the RDBMS. See below.
-
- Overrides:
datatypes.datatype.__set__
- (inherited documentation)
|