Describe the bug
When running sql that produces multiple result sets, the Cursor.description is not updated when calling Cursor.nextset(). In particular, when the first result set doesn't produce rows (like an update without output clause), it is not possible to fetch rows of the 2nd result set.
curs.execute("""
update ... set ...;
update ... set ... output inserted.*;
""")
curs.nextset() # no rows from the first set
print(rows.fetchall()) # this will raiseException message: TypeError: 'NoneType' object is not iterable
Stack trace:
File "/***/python3.12/site-packages/mssql_python/cursor.py", line 1821, in fetchall
raise e
File "/***/python3.12/site-packages/mssql_python/cursor.py", line 1818, in fetchall
return [Row(self, self.description, row_data, column_map) for row_data in rows_data]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/***/python3.12/site-packages/mssql_python/row.py", line 45, in __init__
for i, col_desc in enumerate(description):
^^^^^^^^^^^^^^^^^^^^^^
The provided code works using PyODBC. For that particular case, it is of course possible to split into two Cursor.execute calls. This is a reduced test-case where SQL-table variables are used, which requires the single execute-Call.
Further technical details
Python version: 3.12 using mssql-python==0.13.1
SQL Server version: Azure SQL
Operating system: Ubuntu 24.04 on WSL2
Describe the bug
When running sql that produces multiple result sets, the
Cursor.descriptionis not updated when callingCursor.nextset(). In particular, when the first result set doesn't produce rows (like an update without output clause), it is not possible to fetch rows of the 2nd result set.The provided code works using PyODBC. For that particular case, it is of course possible to split into two
Cursor.executecalls. This is a reduced test-case where SQL-table variables are used, which requires the singleexecute-Call.Further technical details
Python version: 3.12 using
mssql-python==0.13.1SQL Server version: Azure SQL
Operating system: Ubuntu 24.04 on WSL2