Static Typing in Python

[Donn Cave]

In article <tyfn06f87xj.fsf at lxplus059.cern.ch>,
 Jacek Generowicz <jacek.generowicz at cern.ch> wrote:
> JCM <joshway_without_spam at myway.com> writes:
...
> > The FOLDOC definition
> > 
> >   http://wombat.doc.ic.ac.uk/foldoc/foldoc.cgi?query=weak+typing
> > 
> > which mentions things like
> > 
> >   int a = 5;
> >   float b = a;
> > 
> > to me doesn't have anything to do with weak typing, since the int is
> > correctly converted to a float.  I guess thay're calling it "weak
> > typing" because the cast is implicit.
> 
> Yes, and the more implicit type conversion a language does, the more
> likely it is to get it wrong (because what is correct depends on the
> programmers intention), therefore there is a greater likelyhood of
> getting the wrong result. Which is why I consider implicit type
> conversions to be an aspect of weak typing. (However, I have no
> problems with you not seeing it that way.)
> 
> If you don't call it "strong/weak typing", do you have an alternative
> name to describe differences in how "easy going" a type system is?
> 
> Consider the following table.
> 
> 
>             ML    Python   PERL
> 
> 1.5 + 1     No     Yes     Yes
> 
> "3" + 3     No     No      Yes
> 
> 
> ML is clearly stricter than Python, which is stricter than PERL. Do
> you have a name for this sort of strictness ?

I say "strong" as in "strong static typing".  Don't say "strict",
because "ML is a strict language" doesn't say anything about
its type system, it's about order of evaluation.  Haskell is
non-strict, but has strong static typing.

Speaking of which, Haskell's type system supports type classes
like Num (number), where Int and Float are instances of Num
that implement its "+" function.  I'm probably leaving out a
paragraph or two of the interesting parts, but the end result
is that 1.5 + 1 works.  Haskell is not weakly typed.

   Donn Cave, donn at u.washington.edu