Re: s_lock.h default definitions are rather confused

Andres Freund <andres@anarazel.de> writes:
> On 2015-01-15 10:57:10 -0500, Tom Lane wrote:
>> While I'll not cry too hard when we decide to break C89 compatibility,
>> I don't want it to happen accidentally; so having a pretty old-school
>> compiler in the farm seems important to me.

> I'd worked on setting up a modern gcc (or was it clang?) with the
> appropriate flags to warn about !C89 stuff some time back, but failed
> because of configure bugs.

My recollection is that there isn't any reasonable way to get gcc to
warn about C89 violations as such.  -ansi -pedantic is not very fit
for the purpose.
        regards, tom lane