On Wed, 25 Apr 2012 04:26:10 -0400, Deniz Turgut <dtur...@gmail.com> wrote: > On Wed, Apr 25, 2012 at 3:26 AM, Phil Thompson > <p...@riverbankcomputing.com> wrote: >> On Tue, 24 Apr 2012 14:28:02 -0400, Deniz Turgut <dtur...@gmail.com> >> wrote: >>> Would it be logical to add a __nonzero__ method to QPyNullVariant >>> which always returns False? >>> >>> I was going through a code of mine which broke with a new version of >>> PyQt. I realized that it was because there is no __nonzero__ method in >>> QPyNullVariant so bool(QPyNullVariant) is always True. >>> >>> Right now, if you use QVariant api 2, you can't check nicely (without >>> typechecking) whether you get a nonzero value or Null/Falsy value. >> >> From a purist point of view I don't associate Null with False (or with >> True either but the current behaviour reflects the Python default). >> >> From a portability point of view it's a change in the API. >> >> From a pragmatic point of view I need to be persuaded - I've not been >> affected by the issue myself. > > Actually, pragmatic part is not much of a big deal. Although it looks > a bit ugly to me, I can live with code like: > > if value and not isinstance(value, QtCore.QPyNullVariant): > do_something_with(value) > > Or I could deal with it much before, and eliminate a Null possibility. > > I guess my issue is closer to the purist view. I do agree that Null is > neither True, nor False, but I'd expect it to behave like None.
That sounds reasonable. > It is > not False but rather Falsy. That's why, I spent some time figuring out > why a test like 'if value:' was passing when a Null was returned. Happy to implement it on the None precedent. Phil _______________________________________________ PyQt mailing list PyQt@riverbankcomputing.com http://www.riverbankcomputing.com/mailman/listinfo/pyqt
