On Tue, Nov 12, 2013 at 9:38 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Kevin Grittner <kgrittn(at)ymail(dot)com> writes:
>> Does anything stand out as something that is particularly worth
>> looking into? Does anything here seem worth assuming is completely
>> bogus because of the Coverity and Valgrind passes?
>
> I thought most of it was obvious junk: if there were actually
> uninitialized-variable bugs in the bison grammar, for instance, not only
> we but half the programs on the planet would be coredumping all the time.
> Not to mention that valgrind testing would certainly have caught it.
>
> I'd suggest looking only at the reports that pertain to seldom-exercised
> code paths, as those would be the places where actual bugs might possibly
> have escaped notice.
Clang also has a page "FAQ and How to Deal with Common False
Positives," http://clang-analyzer.llvm.org/faq.html. It demonstrates
how to force analysis on a path.
Jeff