Re: [Chicken-users] New immediate values (was: DBI)

[Thomas Chust]

On 29. Februar 2008 17:28:37 -0500 John Cowan <address@hidden> wrote:

> Thomas Chust scripsit:
>
> > this is really a question of taste, isn't it? You could just as well
> > argue  that a NULL value should be of a type that is a subtype of every
> > other  existing type but contains no other concrete instances, which
> > would imply  that all the type predicates should return #t when applied
> > to the NULL  value. This makes a lot of sense when the NULL value is
> > used to indicate  the absence of an object reference and is the way it
> > is handled in several  object oriented languages.
>
> As far as I know, no OO language does that.  They interpret (or IMHO
> misinterpret) a *static* declaration that a variable is of type Foo
> as meaning that it can contain a Foo or null, but nowhere is null
> *dynamically* a member of type Foo: tests analogous to number?, string?,
> etc. always return false on null.

Hello,

for example in Java there are some places where the dynamic type of null is 
effectively a subtype of another class:

 // this returns (String)null and doesn't throw a ClassCastException
 "foo".getClass().cast(null)

There are also places where the language behaves differently, though:

 // this returns false
 null instanceof String

So the static typing behaviour may differ from the dynamic one and it may 
be inconsistent. But a distinction between static and dynamic typing 
behaviour wouldn't apply to Scheme.

Of course I don't want to have the same typing mess as in Java in CHICKEN 
and I do think the whole practice of having nullable reference types (by 
default) is questionable. I just don't think it is completely canonical 
that the type of NULL should be disjoint from every other type, especially 
if you think of types as being sets (or classes) of objects.

> [...]
> OT:  One of the Java puzzlers is this:
>
>         Foo foo = null;
>         foo.bar();
>
> Under what circumstances does that *not* cause a NullPointerException?

If Foo.bar is a static method this will work just fine ;-)

cu,
Thomas

--
Murphy's Law is recursive.  Washing your car to make it rain doesn't work.