Subject: Re: [boost] [Review:Algorithms] Order of args to clamp
From: Andrew Sutton (asutton.list_at_[hidden])
Date: 2011-09-25 09:53:54
> Yes, it may be possible to convert to a common type - but that will involve
> extra work at runtime. If you need to prove the correctness of something
> and this makes that difficult for you, you're welcome to use a subset of the
Shouldn't the algorithm be provably correct on its own right? I hope
that we're not building libraries where our strongest guarantee of
correctness is, "It looks like it does the right thing".
> Consider the case where the types differ only in some attribute like their
> allocator that doesn't change the behaviour of comparisons.
There is a way to define the algorithm over different types and still
preserve the semantics of the ordering, and to do so without requiring
conversions. I suspect that the case where types differ by an
Allocator will satisfy those requirements.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk