When using a buffered cursor, we need to store the field_count inside
Mrdb_Cursor, since db.commit/rollback will overwrite/clear mysql->field_count
inside Connector/C.
When converting time or datetime values with microseconds, the
calculation was wrong, e.g. a millisecond value of .123 was converted
to .000123 instead of .123000. This was already fixed in C/C but not
in C/Python.
- added mutexes for thread safety
- when calling get_connection thread pool will now not return the next
free connection, but the connection that was not used the longest time.
While the default paramstyle is still 'qmark', MariaDB Connector/Python
now also supports the paramstyles 'format' and 'pyformat'. The paramstyles
can't be mixed.
description attribute is now implemented as getter function().
Changes:
- for strings length will now report the number of codepoints (instead) of bytes
- variable length datatypes will report -1 (packed len < 1).
- For floating point values (decimal, float, double) the values for precision and scale will be zero, in case the server sends decimals=31 (which means no precision/scale) was specified for the column.
directly in BLOB. They will be identified in function mariadb_get_pickled()
under the following conditions:
- First two bytes must be 0x8003
- Last byte must be 0x2E
- the depickling call must return a valid (not NULL) object.
mariadb.indicator_null,
mariadb.indicator_default,
mariadb.indictor.ignore
Added support for Tuple and list objects:
List and Tuple will be stored as blob in a dynamic column
