All but one of the PyTuple_SetItem() calls are executed without checking the return value.

Callers:
$ grep -r PyTuple_SetItem Modules/_sqlite 
Modules/_sqlite/connection.c:        PyTuple_SetItem(args, i, cur_py_value);
Modules/_sqlite/cursor.c:        PyTuple_SetItem(row, i, converted);
Modules/_sqlite/cursor.c:    if (PyTuple_SetItem(func_args, 0, Py_NewRef(operation)) != 0) {
Modules/_sqlite/cursor.c:                PyTuple_SetItem(descriptor, 0, column_name);
Modules/_sqlite/cursor.c:                PyTuple_SetItem(descriptor, 1, Py_NewRef(Py_None));
Modules/_sqlite/cursor.c:                PyTuple_SetItem(descriptor, 2, Py_NewRef(Py_None));
Modules/_sqlite/cursor.c:                PyTuple_SetItem(descriptor, 3, Py_NewRef(Py_None));
Modules/_sqlite/cursor.c:                PyTuple_SetItem(descriptor, 4, Py_NewRef(Py_None));
Modules/_sqlite/cursor.c:                PyTuple_SetItem(descriptor, 5, Py_NewRef(Py_None));
Modules/_sqlite/cursor.c:                PyTuple_SetItem(descriptor, 6, Py_NewRef(Py_None));
Modules/_sqlite/cursor.c:                PyTuple_SetItem(self->description, i, descriptor);


All of these are operating on newly created tuples, so I suggest replacing them with PyTuple_SET_ITEM() instead of adding error handling.


For the users in _pysqlite_query_execute() I also suggest to move the tuple creation closer to the code that fills it, in order to minimise the number of decref's needed in case of error.

Would the code be cleaner if it used PyTuple_Pack?

Yes, the _pysqlite_query_execute() case might be cleaner. But I was under the impression that using PyTuple_Pack() was significantly slower. Correct me if I'm wrong.

I've updated my PR to use PyTuple_Pack. Nice reduction in code size. Thanks, Mark!

I considered rewriting this code when touched that files last time (in issue43083). But PyTuple_SetItem() newer fails in these cases, so I left it as is to minimize change.

So it is a pure cosmetic change. I am not opposing it.

New changeset e9194ea6eaa18299d6ccbd3555ce150fab0c6184 by Erlend Egeberg Aasland in branch 'master':
bpo-43852: Improve tuple creation in sqlite3 (GH-25421)
https://github.com/python/cpython/commit/e9194ea6eaa18299d6ccbd3555ce150fab0c6184