Subject: Re: [boost] [gsoc-2013] proposal for approximate string matching
From: Vadim Stadnik (vadimstdk_at_[hidden])
Date: 2013-04-27 17:40:02
On Sun, Apr 28, 2013 at 4:20 AM, Erik Erlandson <eerlands_at_[hidden]> wrote:
> > >
> > >
> > For those who use STL distance is signed integer. If you choose unsigned
> > integer, then I suggest reconsider function names to avoid potential
> > confusion for users. For example, is it "deviation" that you name
> > "distance"?
> I thought the signed result type was "difference", not "distance."
> Formally, any proper distance metric is always >= 0. String edit
> distance is such a metric, and I'd advocate both for continuing to call it
> edit_distance<>, and that it be an unsigned result.
I agree with both comments, STL distance does not represent a "metric".
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk