The samples below are extractions from 3 failed cases among some thousand or so working test cases for the type_traits lib. Note that I am concerned with compilation of the tests, and not execution of the tests.

ADD_CONST

This following is a sample code from one of the add_const tests. It compiles for each of the builtin types, and enums, but chokes on user defined types.. I have included add_const<int const> to show how builtin compiles fine.

struct A {
A(int i) { }
};

template <class T> struct add_const {
typedef T const type;
};

int main() {
add_const<int const>::type c = 0;
add_const<A const>::type a = 0;
//Error: illegal combination of types
}

ADD_VOLATILE

Similiar to const above, this one handles everything but the user defined type.

struct A {
A(int i) { }
};

template <class T> struct add_volatile {
typedef T volatile type;
};

int main() {
add_volatile<int volatile>::type c = 0;
add_volatile<A volatile>::type a = 0;
//Error: illegal combination of types
}

IS_BASE_AND_DERIVED

This is a somewhat new approach to perfoming templatized RTTI without the C++ RTTI mechanism.. It returns true if the derived class is derived from the base class and false otherwise. Virtually all of the 200 or so tests from this section work (including the "tricky" partial specialization compiler torture tests), but when supplied with two references, some problems arise..

template <class T, class U> struct A;
template <class T, class U> struct A<T,U&> { };
template <class T, class U> struct A<T&,U> { };
template <class T, class U> struct A<T&,U&> { };

int main() {
A<int&,int&> a;
//Error: ambiguous match of class template partial specialization 'A'
}

Richard

Thanks, I'll add them to the bug list!

Richard Grant wrote:

> The samples below are extractions from 3 failed cases among some thousand or so working test cases for the type_traits lib.

Nothing to do with typetraits, just a simple boost question <g>

Do you have a dmc_tools.jam I can steal to build the boost libraries
with bjam?
Now I have got my head around the regression suite, I would quite like
to add DMC to my testing.  Unfortunately, I am not yet familiar enough
with either DMC or xxx-tools.jam files to write my own yet.

-- 
AlisdairM

>Do you have a dmc_tools.jam I can steal to build the boost libraries with bjam?

You'll have to go begging to Christof. I'm still working by hand.

Richard

On Thu, 27 Mar 2003 22:57:15 +0000, Alisdair Meredith wrote:
>> The samples below are extractions from 3 failed cases among some thousand or so working test cases for the type_traits lib.
> Do you have a dmc_tools.jam I can steal to build the boost libraries with bjam?

Have a look at http://cmeerw.org/prog/dm/boost.html - still not perfect, but mostly useable.


bye, Christof

-- 
http://cmeerw.org                                 JID: cmeerw@jabber.at mailto cmeerw at web.de

...and what have you contributed to the Net?

I fixed the first two, but the third has me concerned:

"Richard Grant" <fractal@clark.net> wrote in message news:b5vfln$23ut$1@digitaldaemon.com...
> This is a somewhat new approach to perfoming templatized RTTI without the
C++
> RTTI mechanism.. It returns true if the derived class is derived from the
base
> class and false otherwise. Virtually all of the 200 or so tests from this section work (including the "tricky" partial specialization compiler
torture
> tests), but when supplied with two references, some problems arise..
>
> template <class T, class U> struct A;
> template <class T, class U> struct A<T,U&> { };
> template <class T, class U> struct A<T&,U> { };
> template <class T, class U> struct A<T&,U&> { };
>
> int main() {
> A<int&,int&> a;
> //Error: ambiguous match of class template partial specialization 'A'
> }

I think this case *is* ambiguous. Class partial specialization matches do not go with the 'best' match like function overloading does, it goes with whichever is 'most specialized'. Most specialized is determined by if the arguments for one can be used as arguments to another, but not vice versa. In the case of references, each specialization here can be arguments to each other specialization. None is 'more specialized' than the other, hence it is ambiguous.

Never mind :-(

Christof Meerwald wrote:

> On Thu, 27 Mar 2003 22:57:15 +0000, Alisdair Meredith wrote:
> >> The samples below are extractions from 3 failed cases among some thousand or so working test cases for the type_traits lib.
> > Do you have a dmc_tools.jam I can steal to build the boost libraries with bjam?
>
> Have a look at http://cmeerw.org/prog/dm/boost.html - still not perfect, but mostly useable.
>
> bye, Christof
>
> --
> http://cmeerw.org                                 JID: cmeerw@jabber.at mailto cmeerw at web.de
>
> ...and what have you contributed to the Net?

Is this compatible with Boost 1_30_0, please?

John

On Wed, 16 Apr 2003 12:05:14 +0100, John Fletcher wrote:
> Christof Meerwald wrote:
>> On Thu, 27 Mar 2003 22:57:15 +0000, Alisdair Meredith wrote:
>> > Do you have a dmc_tools.jam I can steal to build the boost libraries with bjam?
>> Have a look at http://cmeerw.org/prog/dm/boost.html - still not perfect, but mostly useable.
> Is this compatible with Boost 1_30_0, please?

the patch for Boost 1.29.0 doesn't cleanly apply to Boost 1.30.0 (but the Jam related things should be OK).

BTW, I am currently working on updating my patch for Boost 1.30.0... (will hopefully have something useable in the next few days)


bye, Christof

-- 
http://cmeerw.org                                 JID: cmeerw@jabber.at mailto cmeerw at web.de

...and what have you contributed to the Net?