Re: Add "-Wimplicit-fallthrough" to default flags (was Re: pgsql: Support FETCH FIRST WITH TIES)

On 2020-May-11, Julien Rouhaud wrote:

> On Mon, May 11, 2020 at 4:40 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> >
> > Julien Rouhaud <rjuju123(at)gmail(dot)com> writes:
> > > On Mon, May 11, 2020 at 3:41 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > >> Why? It uses "fallthrough" which is a legal spelling per level 4.
> >
> > > GCC documentation mentions [ \t]*FALLTHR(OUGH|U)[ \t]* for level 4
> > > (out of the view other alternatives), which AFAICT is case sensitive
> > > (level 3 has fall(s | |-)?thr(ough|u)[ \t.!]*(-[^\n\r]*)?).
> >
> > Oh, I'd missed that that was case sensitive. Ugh --- that seems
> > unreasonable. Maybe we'd better settle for level 3 after all;
> > I don't think there's much room to doubt the intentions of a
> > comment spelled that way.
>
> Agreed.

Pushed, thanks.

I ended up using level 4 and dialling back to 3 for zic.c only
(different make trickery though). I also settled on FALLTHROUGH rather
than FALLTHRU because the latter seems ugly as a spelling to me. I'm
not a fan of the uppercase, but the alternative would be to add a - or
@s.

I get no warnings with this (gcc 8), but ccache seems to save warnings
in one run so that they can be thrown in a later one. I'm not sure what
to make of that, but ccache -d proved that beyond reasonable doubt and
ccache -clear got rid of the lot.

--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services