Re: latest gcc vs lib/timespec.h:85

[Paul Eggert]

On 10/02/2017 06:24 PM, Jim Meyering wrote:
> Given all of the comments on that function, I'd be tempted to suppress
> this warning in that function.

That would work. Another possibility would be to include verify.h and 
add something like this to the start of timespec_cmp:

  assume (-1 <= a.tv_nsec && a.tv_nsec <= 2 * TIMESPEC_RESOLUTION);

  assume (-1 <= b.tv_nsec && b.tv_nsec <= 2 * TIMESPEC_RESOLUTION);

We might be able to make these 'assume' calls fancier, to exactly match 
the comments, but I'm not sure it's worth the bother.