>When I set a breakpoint, immediately when I press F9, the VC IDE crashes
>with a memory exception (NULL-ptr). This happens only under certain
>circumstances however. Also, stepping through the code with the debuggger
>shows some strange behaviour. All this happens only in the following
>situation:

>I have a class (B) derived from a template class (A). The derived class (B)
>does not contain any members or methods of its own. The template class
>contains two methods: foo_inner() and foo_outer(). Both may be called
>independently, foo_outer() calls foo_inner().

>When B::foo_inner() is called directly, stepping from the calling code into
>foo_inner() using F11 the next stmt shown is not the first stmt of
>foo_inner() but is inside the first stmt of foo_inner(). If I call
>b::foo_outer() and step into it, the de{*filter*} takes me right into
>foo_inner() - this time showing the actual statement. Looks like the
>de{*filter*} skips one level of the call stack (where the skipped levels aren't
>shown either).

>This strange behaviour made me wish to set the breakpoint directly in the
>method. But - bang! I'd suspect that these symptoms are related.

>I have no idea where to start looking. Maybe some of you out there know of
>similar problems.

>I'm using VC6 with the Dinkumware STL (2.33). The latter might not have to
>do anything with the problem but because templates are involved I thought I
>better mention it.

> When I set a breakpoint, immediately when I press F9, the VC IDE crashes
> with a memory exception (NULL-ptr). This happens only under certain
> circumstances however. Also, stepping through the code with the debuggger
> shows some strange behaviour. All this happens only in the following
> situation:

> I have a class (B) derived from a template class (A). The derived class
(B)
> does not contain any members or methods of its own. The template class
> contains two methods: foo_inner() and foo_outer(). Both may be called
> independently, foo_outer() calls foo_inner().

> When B::foo_inner() is called directly, stepping from the calling code
into
> foo_inner() using F11 the next stmt shown is not the first stmt of
> foo_inner() but is inside the first stmt of foo_inner(). If I call
> b::foo_outer() and step into it, the de{*filter*} takes me right into
> foo_inner() - this time showing the actual statement. Looks like the
> de{*filter*} skips one level of the call stack (where the skipped levels
aren't
> shown either).

> This strange behaviour made me wish to set the breakpoint directly in the
> method. But - bang! I'd suspect that these symptoms are related.

> I have no idea where to start looking. Maybe some of you out there know of
> similar problems.

> I'm using VC6 with the Dinkumware STL (2.33). The latter might not have to
> do anything with the problem but because templates are involved I thought
I
> better mention it.

> Regards
> Rudi