Why not 3.__class__ ?

[Tim Peters]

[James_Althoff at i2.com]
> ...
> At a higher level it seems that if you want to do an "isinstance" test,
> it is nicer to use the class object that you already have "in hand" (or
> can easily get to with a literal constant) rather than having to know
> about an external module (for which you need to know the name and package
> path) with class names (that you also need to know) -- especially given
> that the general case, isinstance(x,y.__class__), is something one needs
> to use in some circumstances in any case.
>
> In the end,
> (0).__class__ works -- it justs adds unfortunate "line noise" and
> momentarily invokes the "is it an expression or is it a tuple"
> confusion
>
> 0 .__class__ works, too -- but is somewhat fragile because of the
> required white space
>
> Still, these are better than nothing.  :-)

I like 2.2, where you can skip all that and just say

    isinstance(whatever, int)

If you're tempted to write that

    isinstance(whatever, (0).__class__)

instead in 2.2 (which also "works"), you're likely confused about things
more basic than the consequences of maximal-munch lexing <0.9 wink>.