From: Greg Colvin (gcolvin_at_[hidden])
Date: 2002-02-06 15:48:38
From: "David Abrahams" <david.abrahams_at_[hidden]>
> From: "Rainer Deyke" <root_at_[hidden]>
> > That just leaves you with the weak exception guarantee,
>
> This is a personal beef of mine: I regret ever introducing the term "weak
> guarantee". I started using the term "basic" instead after a very short time
> because "weak" tends to play into the common misconception that operations
> providing the basic guarantee are somehow not exception-safe. Normal vector
> inserts give the basic guarantee for types with throwing copy and/or
> assignment for very important reasons of efficiency. This guarantee is
> perfectly good for many situations and can be strengthened by the user when
> neccessary.
>
> So, I can't force you to say "basic guarantee", but I can ask nicely. Pretty
> please?
>
> Regards,
> Dave
Or don't say anything? When we rewrote your proposal in standardese
we never mention the basic guarantee, just because it is so basic.
Unless a construct specifically requires its users not to throw there
is no excuse for undefined behavior when they do.