On 2015-07-09 22:57:25 +0900, Michael Paquier wrote:
> On Wed, Jul 8, 2015 at 6:16 PM, Andres Freund wrote:
> >> In any case, we are going to need at least (void) in front of those calls.
> >
> > We're "needing" nothing of the sort.
>
> I don't really understand your reluctance here. As one example, see
> c831593 where similar fixes are done and even back-patched.
That doesn't make it a required thing. And the changes there we more
than just adding a (void).
To me this kind of changes are busywork. Analsys tools are there to make
our work easier, not to generate more. There's good reasons why, with
other tools, in the past we've rejected lots of bogus "issues", even if
we could have silenced them by changing the code.