glibc 2.22 -- Final testing for 32-bit x86 failing?

[Roland McGrath]

> Hmm, for some reason I can't reproduce this anymore, and unistd.h is
> already included indirectly.
> 
> I don't understand why that's happening. But it still seems like a
> good idea from header hygiene perspective.

It is definitely correct (and in the "obvious" category) to have each
source file explicitly #include the header that is the canonical declarator
of each symbol that source file uses directly.