Defect reports?

[Martin Sebor]

Pétur Runólfsson wrote:

...
> * The resolution of DR 25 requires undefined behaviour.

What makes you think that? We implement the proposed resolution
(with the change that we store XXX_MAX or XXX_MIN on integer
overflow) but AFAIK we're not relying on undefined behavior
anywhere in our code.

> 
> * The output functions are inconsistent in how they handle output
> errors. The arithmetic inserters and unformatted output functions
> set badbit, operator<<(basic_ostream, basic_string) sets failbit,
> and the rest appear to do nothing.

This is not surprising. Issue 394 tries to deal with some of it,
but what really needs to happen is that somebody needs to sit down
and review all of iostreams for these inconsistencies and clean it
all up. Benjamin says he's working on something along these lines.

Martin