On 2015-03-25 19:11:06 -0400, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > Add macros wrapping all usage of gcc's __attribute__.
>
> I noticed that this commit attached pg_attribute_noreturn not only
> to the extern declarations, but to some actual function definitions.
Unless either Oskari or I screwed up, it should just have been a 1:1
translation from previous __attribute__((noreturn)) to
pg_attribute_noreturn. I looked through the commit just now and didn't
see any new locations.
> I think this is a bad idea, because it's going to look like heck after
> pgindent gets through with it. Do we actually need decoration on the
> function definitions?
Hm, I guess it should not look any worse than before? None of the
locations look like they've been introduced after the last pgindent
run. I only see plpgsql_yyerror, yyerror. That said, I see little reason
to add the noreturn thingy to the definition and not the declaration for
those. It actually looks to me like there's a declaration for
replication_yyerror, but a plain yyerror is used instead in repl_scanner.l?
Greetings,
Andres Freund