[chuck-users] bug on virtuality

[Kassen]

eduard;


What do you think?
>

I tried a few things but I'm a bit stuck here as well. Right now, unless I'm
very mistaken, this looks to me like it's indeed wrong and another
expression of something being wrong in the type-system where arrays are
concerned.

Here in particular array types don't seem respect extending of classes,
type-wise. I Hacked up your class structure for a bit, having Point1D extend
Point and Point2D extend Point1D, etc but that didn't help.

I ended up writing a small trivial test case. I feel this should work, it
should print a series of 3 zeros.... yet it doesn't;

//==================-8<======================-
class bar
    {
    int value;
    }

class baz extends bar
    {
    }

baz foo[3];

fun void print( bar p[])
    {
    for(0 => int n; n<p.cap(); n++)
        {
        <<<p[n].value>>>;
        }
    }

print(foo);
//==========================

To prove my case, consider this trivial case;
//========================
class bar
    {
    int value;
    }

class baz extends bar
    {
    }

baz foo;

fun void print( bar p)    <<<p.value>>>;

print(foo);
//===================

...which does fly; foo is a "baz" which is in turn a "bar". Elements of
arrays of this type should be the same.

I suspect this is a clearer example of a case I ran into some months ago but
it wasn't clear to me back then what aspect caused the issue (reported back
then straight to Ge for reasons I can't quite recall). I'd say the issue
isn't passing arrays as arguments as such (this is fine) but the type of
array elements that are instances of classes of that extend another class.
Their type will only become a real issue when they are passed to functions
as variables of the mother type, hence your issue here.

I'm a bit surprised this doesn't seg-fault, I think mine did back then.

Anyway; I say it's a type-system bug in arrays.

Nice catch, I'm fairly sure this one has been in there for a while but was
hard to pinpoint. It seems to be a good season for hunting!

Yours,
Kas.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/chuck-users/attachments/20081019/0d8438c1/attachment.htm>