Upgrade to 4.0. This is a major release and may risk some backwards

incompatibilities.  Here are the major changes.

- Dropped support for Python below 2.3 and PostgreSQL below 7.4.
- Improved performance of fetchall() for large result sets
  by speeding up the type casts (as suggested by Peter Schuller).
- Exposed exceptions as attributes of the connection object.
- Exposed connection as attribute of the cursor object.
- Cursors now support the iteration protocol.
- Added new method to get parameter settings.
- Added customizable row_factory as suggested by Simon Pamies.
- Separated between mandatory and additional type objects.
- Added keyword args to insert, update and delete methods.
- Added exception handling for direct copy.
- Release the GIL while making a connection
  (as suggested by Peter Schuller).
- If available, use decimal.Decimal for numeric types.
- Allow DB wrapper to be used with DB-API 2 connections
  (as suggested by Chris Hilton).
- Made private attributes of DB wrapper accessible.
- Dropped dependence on mx.DateTime module.
- Support for PQescapeStringConn() and PQescapeByteaConn();
  these are now also used by the internal _quote() functions.
- Added 'int8' to INTEGER types. New SMALLINT type.
- Added a way to find the number of rows affected by a query()
  with the classic pg module by returning it as a string.
  For single inserts, query() still returns the oid as an integer.
  The pgdb module already provides the "rowcount" cursor attribute
  for the same purpose.
- Improved getnotify() by calling PQconsumeInput() instead of
  submitting an empty command.
- Removed compatibility code for old OID munging style.
- The insert() and update() methods now use the "returning" clause
  if possible to get all changed values, and they also check in advance
  whether a subsequent select is possible, so that ongoing transactions
  won't break if there is no select privilege.
- Added "protocol_version" and "server_version" attributes.
- Revived the "user" attribute.
- The pg module now works correctly with composite primary keys;
  these are represented as frozensets.
- Removed the undocumented and actually unnecessary "view" parameter
  from the get() method.
- get() raises a nicer ProgrammingError instead of a KeyError
  if no primary key was found.
- delete() now also works based on the primary key if no oid available
  and returns whether the row existed or not.

0 files changed, 0 insertions, 0 deletions