|
Boost : |
From: Don Waugaman (dpw_at_[hidden])
Date: 2000-03-18 02:36:08
> on 3/17/00 5:01 PM, Miki Jovanovic at miki_at_[hidden] wrote:
>
> > Hi all,
> >
> > It just occured to me that the
> >
> > noncopyable() {}
> > ~noncopyable() {}
> >
> > should be
> >
> > noncopyable() throw() {}
> > ~noncopyable() throw() {}
> >
> > Cheers,
> >
> > Miki.
>
> <steam rising...>
>
> We eschew exception-specifications at boost. For the reasons why, see...
>
> [Sheesh, Beman, I guess this comes up often enough that we ought to get the
> rationale on the site. Egroups is a lovely host but their search engine
> leaves something to be desired. I can't find the original posting.]
>
> In brief, they provide no compile-time safety and lead to worse generated
> code on many (if not most) compilers.
Pardon me for butting in...
I can see where this would be true in the general case of exception
specifications, but isn't a throw() specification on a trival function
an (ahem) exception to the rule?
Perhaps that rationale would be a good thing to see, particularly if it
covers such cases.
Don
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk