Re: [boost] Current Guidance on Compiler Warnings?

Subject: Re: [boost] Current Guidance on Compiler Warnings?
From: Emil Dotchevski (emildotchevski_at_[hidden])
Date: 2018-11-27 03:31:11

On Mon, Nov 26, 2018 at 3:44 PM Gavin Lambert via Boost <
boost_at_[hidden]> wrote:
>
> On 27/11/2018 12:05, Peter Dimov wrote:
> > Gavin Lambert wrote:
> >> As an index for a vector or array which cannot have a valid index
> >> below zero, it's perfectly fine.
> >
> > It's not perfectly fine, because you can pass a negative index to it and
> > there's no way to check for that (from within the function). If you take
> > a signed type, you can assert.
>
> You cannot pass a negative index to it without a warning at the call
> site, so you have to fix it there anyway.

If you have:

void f( unsigned );

void g( int x )
{
  f(x);
}

I don't think you'll get a warning. But I might be wrong, so let's say you
do get a warning, so you do:

void g( int x )
{
  f( static_cast<unsigned>(x) );
}

How does this help in detecting the logic error of passing a negative x to
g? Can you quantify the exact mechanics by which this cast makes your code
safer? By what measure is the code less safe without the cast?