--- In boost@[...].. , Jens Maurer <Jens.Maurer@[...]..>  wrote:
>  williamkempf@[...].. wrote:
>  > I've been thinking a bit about the desire for catch_exceptions()
to
>  > be able to handle registered exception types
> 
>  [ Lots of explanation removed. ]
> 
>  > Any thoughts or comments on the idea?
> 
>  This looks interesting. However, I believe we should consider
>  this only after having initially released the boost test library
>  and after someone actually writing tests comes up with a need for
>  it. Usually, I'd expect the ex.what() string to be sufficiently
>  descriptive. Except for some standard library implementations where
>  std::bad_alloc::what() is always the empty string. But we do handle
>  all standard exceptions explicitly already, so that point is moot.

I definately agree that this should not be considered for the initial
release of the Boost Test Library (at least not if it will slow or
hinder it's final inclusion). However, there are several reasons
that this idea might be considered beyond the obvious.

A) An exception type may not be derived from a std::exception type.
B) The user might want to register special error handling beyond
simple output of the cause for failure (execution_tools or even
catch_exceptions() may be used outside of test harnesses).
C) The user may want to override previously registered handlers to
do specific actions instead of or in addition to the previous
behavior.

Bill Kempf