Session¶

class lsst.daf.butler.registry.interfaces.Session(db: lsst.daf.butler.registry.interfaces._database.Database)¶

Bases: object

Class representing a persistent connection to a database.

Parameters:	db : `Database` Database instance.

Notes

Instances of Session class should not be created by client code; Database.session should be used to create context for a session:

with db.session() as session:
    session.method()
    db.method()

In the current implementation sessions can be nested and transactions can be nested within a session. All nested sessions and transaction share the same database connection.

Session class represents a limited subset of database API that requires persistent connection to a database (e.g. temporary tables which have lifetime of a session). Potentially most of the database API could be associated with a Session class.

Methods Summary

`dropTemporaryTable`(table)	Drop a temporary table.
`makeTemporaryTable`(spec, name, None] = None)	Create a temporary table.
`temporary_table`(spec, name, None] = None)	Return a context manager that creates and then drops a context manager.

Methods Documentation

dropTemporaryTable(table: sqlalchemy.sql.schema.Table) → None¶

Drop a temporary table.

Parameters:	table : `sqlalchemy.schema.Table` A SQLAlchemy object returned by a previous call to `makeTemporaryTable`.

makeTemporaryTable(spec: lsst.daf.butler.core.ddl.TableSpec, name: Optional[str, None] = None) → sqlalchemy.sql.schema.Table¶

Create a temporary table.

Parameters:	spec : `TableSpec` Specification for the table. name : `str`, optional A unique (within this session/connetion) name for the table. Subclasses may override to modify the actual name used. If not provided, a unique name will be generated.
Returns:	table : `sqlalchemy.schema.Table` SQLAlchemy representation of the table.

Notes

Temporary tables may be created, dropped, and written to even in read-only databases - at least according to the Python-level protections in the Database classes. Server permissions may say otherwise, but in that case they probably need to be modified to support the full range of expected read-only butler behavior.

Temporary table rows are guaranteed to be dropped when a connection is closed. Database implementations are permitted to allow the table to remain as long as this is transparent to the user (i.e. “creating” the temporary table in a new session should not be an error, even if it does nothing).

It may not be possible to use temporary tables within transactions with some database engines (or configurations thereof).

temporary_table(spec: lsst.daf.butler.core.ddl.TableSpec, name: Optional[str, None] = None) → Iterator[sqlalchemy.sql.schema.Table]¶

Return a context manager that creates and then drops a context manager.

Parameters:	spec : `ddl.TableSpec` Specification for the columns. Unique and foreign key constraints may be ignored. name : `str`, optional If provided, the name of the SQL construct. If not provided, an opaque but unique identifier is generated.
Returns:	table : `sqlalchemy.schema.Table` SQLAlchemy representation of the table.

Navigation

Session¶