Re: quiet inline configure check misses a step for clang

On 2014-04-03 09:43:20 -0400, Tom Lane wrote:
> Andres Freund <andres(at)2ndquadrant(dot)com> writes:
> > The current quiet inline test doesn't work for clang. As e.g. evidenced in
> > http://buildfarm.postgresql.org/cgi-bin/show_stage_log.pl?nm=gull&dt=2014-04-03%2007%3A49%3A26&stg=configure
> > configure thinks it's not quiet.
>
> > Which means that postgres compiled with a recent clang will be noticably
> > slower than it needs to be.
>
> > The reason for that is that clang is smart and warns about static inline
> > if they are declared locally in the .c file, but not if they are
> > declared in a #included file. That seems to be a reasonable
> > behaviour...
>
> > I think that needs to be fixed. We either can make the configure test
> > considerably more complex or simply drop the requirement for quiet
> > inline.
>
> I object to the latter;

Do you have an idea how to make the former work? The whole autoconf
getup doesn't really seem to support generating two files for a
test. I've looked a bit around and it seems like it'd be a fair amount
of effort to do it solely in autoconf.
The easiest seems to be to just have a header for the test in the
sourcetree, but that seems fairly ugly...

Greetings,

Andres Freund

--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services