[Insight-developers] Borland vs Mesh

[Miller, James V (Research)]

Okay, next hurdle....

FixedArray has two internal classes: "Reference" and "ConstReference", that
are full fledged classes that "act" like reference symantics.

These internal classes seem to be tripping the compiler.  When I place a static
const int inside FixedArray, I get a ") expected" error.  If I remove the "use"
of the Reference and ConstReference classes from itkArrayTest.cxx then 
everything compiles fine.

Do we need these Reference and ConstReference internal classes?

Jim



> -----Original Message-----
> From: Brad King [mailto:brad.king@kitware.com]
> Sent: Thursday, September 05, 2002 8:04 PM
> To: Miller, James V (Research)
> Cc: Insight-developers (E-mail)
> Subject: Re: [Insight-developers] Borland vs Mesh
> 
> 
> > The code at the bottom of this email will not compiler 
> under the Borland
> > compiler.  It returns a
> [snip]
> > int main(int, char *[])
> > {
> >   B<float, 3> b1;                   // okay
> >   C<float, 3, A<float, 3, 5> > c1;  // okay
> >
> >   C<float, 3> c2;                   // error: Illegal base 
> class type
> >   // C<float, 3, A<float, 3> > c3;     // internal compiler error
> > }
> 
> I just tried it with the Kylix 3 Open Edition compiler:
> 
> Borland C++ 5.7 Open Edition Copyright (c) 1987, 2002 Borland
> 
> It doesn't support it at all.  Using your code, all four 
> lines in main()
> produce an error message.  One would think that support would 
> improve in a
> more recent version.
> 
> -Brad
>