From: Jeff Garland (jeff_at_[hidden])
Date: 2004-05-20 09:38:38
On Thu, 20 May 2004 07:14:33 -0700, Victor A. Wagner Jr. wrote
> >By default you will get the output operator of the underlying value type which
> >in my experience should normally suffice. Or is there some additional
> >functionality you had in mind?
> not for output.
> for input it should fail if the input would fail the value input
> would fail rangeAssert.
Ok, I see. Given the policy-based nature of the error handling this could be
a bit tricky. However, I think this is a use case that argues for a
consistent is_valid method on the constraint_policy. This way we can write
one operator>> that can stream into the underlying type, check the validaty
prior to attemting to set the value, and then set the ios flags on failure.
Another problem this use case brings up is how to construct these...
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk