On 9/10/07, Neil Cerutti <[EMAIL PROTECTED]> wrote:
> On 2007-09-10, Chris Mellon <[EMAIL PROTECTED]> wrote:
> > On 9/10/07, Neil Cerutti <[EMAIL PROTECTED]> wrote:
> >> On 2007-09-08, Wildemar Wildenburger <[EMAIL PROTECTED]> wrote:
> >> > Lawrence D'Oliveiro wrote:
> >> >>>>> if y[0]:
> >> >>>> Not a good idea.
> >> >>> Why not?
> >> >>
> >> >> Because there is a situation where your version of the test
> >> >> will fail even if the first element of y is non-null.
> >> >
> >> > Such as? Seriously people, a little more verbosity wouldn't
> >> > hurt here. This isn't a mystery game.
> >>
> >> >>> if "": True
> >> ...
> >> >>> if 0: True
> >> ...
> >> >>> if []: True
> >> ...
> >> >>> if {}: True
> >> ...
> >> >>> if None: True
> >> ...
> >>
> >> That's may not be all the possibilities. Lots of Python objects
> >> can evaluate to false in a boolean context.
> >
> > All of these except 0 and None will also return False if you
> > check the length. In fact, I am not aware of any builtin where
> > len() will return 0, but it will be true in a boolean context.
> >
> > If you're advocating being more specific in the hope of
> > catching errors, this is a pretty terrible example. The OP
> > wasn't checking for "null", whatever that means - he was
> > checking for the empty string, in a list known (or intended) to
> > contain only strings. The simple boolean check is totally
> > appropriate.
>
> Agreed; but I prefer 'if y[0] == ""', absent more context and
> better names.
>Probably should use u"" if you're going to take that route, as this will fail spuriously if y[0] contains a unicode string that can't be implicitly converted to ascii. Personally, I prefer the boolean check and I'll let operations fail elsewhere if there's a type mismatch. -- http://mail.python.org/mailman/listinfo/python-list
