PATCH [3/n]: Add __snseconds_t and __SNSECONDS_T_TYPE

[Paul Eggert]

On 04/11/2012 09:47 AM, Roland McGrath wrote:
> But if everyone is agreed that departing from POSIX here, or having POSIX
> changed, is a somewhat likely outcome, then the change can go in.

I'm agreed.  I was persuaded by the argument that the change is not likely to
affect real code; I looked prety hard and found just one example of real code that
would be affected, in an obsolete version of a GNU utility that did something
like "&x.tm_nsec".

I think that if a push is made to change POSIX, the change would be accepted.