[boost] Re: i/o state savers exception-safety

From: Daryle Walker (darylew_at_[hidden])
Date: 2002-08-04 01:44:19

on 8/2/02 3:14 PM, Thomas Witt at witt_at_[hidden] wrote:

> On Thursday 01 August 2002 15:52, David Abrahams wrote:
>> From: "Daryle Walker" <darylew_at_[hidden]>
[response left out]
>> 
>> 
>> IMO this response is much too vague to be appropriate for Boost. You should
>> be sure about what the standard says on stream state re-setting, and you
>> should understand what the standard says about uncaught_exception() to see
>> if its use is appropriate here rather than appealing to what "many don't
>> think" about its usefulness.
> 
> I have to agree with dave here. Though I see the point that support for
> std::uncaught_exception is patchy.
> 
> That said I do think that io-state saver needs a serious rework regarding
> exception safety. From first analysis there are actually three saver types
> where the destructor can throw, state, exception and rdbuf. Only one of them
> is documented as such. With rdbuf being the most esoteric in my opinion (It
> may throws if rdbuf restores a null pointer). On second thought even locale
> may throw if registered callbacks throw, but I am not sure about this one.
[TRUNCATE]

I adding a change to the docs explaining that both the failure-state and
exception-state savers can throw on destruction (due to the way the
accessors they use interact).  I looked at the Standard, and I don't think
passing "rdbuf" a NULL pointer doesn't cause a throw.  However, the
failure-state method "clear" adds the "bad" state to whatever input state it
gets if rdbuf is NULL.  If you have a NULL rdbuf, then I think you didn't
initialize your stream correctly, and I'm not taking responsibility for that
bad (pardon the pun) behavior.

-- 
Daryle Walker
Mac, Internet, and Video Game Junkie
darylew AT mac DOT com