Re: Consistently using __throw_* (was: Re: PATCH: Eliminate uses of "sleep"...)

[Jonathan Wakely <cow at compsoc dot man dot ac dot uk>]

On Wed, Jan 05, 2005 at 03:52:41PM +0100, Paolo Carlini wrote:

> >"__throw_exception_again;" should be used in place of "throw;" to rethrow
> >an active exception.  With -fno-exceptions __throw_exception_again is
> >replaced with an empty string, which is not what's wanted.
> 
> You are right, in my opinion, but let's not involve our release manager
> with our (v3) bugs... ;)

Yes, sorry, Mark - I didn't mean to hold up your patches with my picky
observation.

> I mean, some time ago I did an audit for this issue, and for a consistent
> fix we need to add a few functions to the set in functexcept.h.
> At least:
> 
> 	__throw (which boils down to either throw or abort)

If that is to be used in place of "throw;" then isn't that what
__throw_exception_again() is intended for?  It doesn't abort() but does
it need to?  Is it not enough to just let a "throw;" statement compile
with -fno-exceptions ?

If -fno-exceptions was given then we will already have aborted when the
original __throw_xxx function was run, unless the exception we've caught
was thrown by user code.

> 	__throw_symbol_error
> 	__throw_facet_error
> 	__throw_positioning_error
> 	__throw_fuzzy_logic

There will also need to be a std::tr1::__throw_bad_weak_ptr() for TR1.
I've currently defined that in <tr1/memory> to avoid adding a new
tr1/functexcept.h file.

> and consistently use some already present, like
> 
> 	__throw_overflow_error	
> 	__throw_underflow_error
> 
> If we agree, I can prepare a patch. Benjamin?
> 
> Thanks for your observation,
> Paolo.
> 
> 
> 

-- 
VOTE, v.  The instrument and symbol of a free man's power to make
a fool of himself and a wreck of his country.
	- Ambrose Bierce, 'The Devil's Dictionary'