dCursorMixin.setFieldVal() can improperly set value if current value is None

isinstance(rec[fld], cls) is equivalent to issubclass(fldType, cls), no?

So the only substantial change is on line 961, correct?

I'm not against changing these isinstance() calls to issubclass() calls, but I just wanted to make sure I understood the fixes first.

Replying to paul:

isinstance(rec[fld], cls) is equivalent to issubclass(fldType, cls), no?

Not exactly.
While isinstance(0, int) returns True, isinstance(None, int) returns False.
isinstance() function could be right here, if you could prevent situation where rec[fld] = None. But you can't.
But issubclass(int, int) always return True - this is the point of changes.
This is perfectly visible in my MsSQL 'bit' type mapping example.
While it's 'int' type data with range [0, 1], UI layer returns boolean type value instead for bound field.
In such situation conversion fails if isinstance() is used, dataset is incorrectly updated and backend returns exception, like 'Column True not found'.
By using issubclass() function, fields are always updated to value of correct type, regardless previous value is None or not.
I think, there should be yet another change in line 957, to avoid unneeded conversion (e.g. None to 0), from

		if fldType is not None:

to

		if fldType is not None and val is not None: