Files
mysqlclient/MySQLdb/CHANGELOG
2001-09-07 01:08:17 +00:00

60 lines
2.4 KiB
Plaintext

0.9.1c1
* Added ER_PARSE_ERROR as a ProgrammingError, because
ER_SYNTAX_ERROR doesn't get the job done.
* In Python < 2.0.1, PyArgs_ParseTupleAndKeywords() returns a new
reference when the O format is used. In 2.0.1 and later, it
returns a borrowed reference. Since it's not actually documented
either way that I can tell, I must assume this is some sort of
bugfix. However, it does mean that you need to test against the
Python version, and Py_INCREF() if it is 2.0.1 or later. If you
Py_INCREF() all the time, you get a memory leak in earlier
version; and if you never Py_INCREF(), it works fine in earlier
versions but crashes later versions. If you're an end-user, don't
worry about this...
* In Python 2.2, the internal _PyTuple_Resize() dropped the (unused)
third argument, so we have to test against the Python version here
as well to maintain backwards compatibility.
* Some deprecated MySQL API functions were removed.
* A literal() method was added to the database connection. It takes
a single argument and converts it into an SQL literal. If the
argument is a non-string sequence, the sequence items are
converted to SQL literals, and the sequence of converted items is
returned as a list. This was previously done internally by Cursor
objects. You don't normally need to use it, but it might be useful
for debugging.
* Return DECIMAL/NUMERIC columns as floats.
* A number of documentation updates.
0.9.0
Too many changes from 0.3.5 to go into much detail.
* MySQLdb was turned into a true Python package.
* _mysql didn't change all that much. Constants were moved out into
the MySQLdb package structure. Exceptions were moved to
_mysql_exceptions.py, but these are imported into _mysql. This
makes the binary (and source) significantly smaller.
* Reduced threadsafety to 1 (from 2), and ripped out a lot of the
crap that attempts to let you safely share a connection between
threads. If you are using transactions, you can only hand off the
connection from one thread to another after a commit or
rollback. Sharing a connection will hurt your performance anyway,
so give each thread it's own connection.
* A Set class was added for use with SET columns.
* Loads of docstrings were added to take advantage of pydoc.
* Operationally, there's not much difference.