Re: Bogus reports from coverage.postgresql.org

Tom Lane wrote:
> I happened to notice that if you look at
> https://coverage.postgresql.org/src/backend/optimizer/util/predtest.c.gcov.html
> it claims that the two occurrences of
> elog(ERROR, "predicate_classify returned a bogus value");
> at lines 492 and 803 are reached.
>
> This would be quite astonishing if true, but it isn't true, since
> the regression tests aren't reporting any such failure.
>
> Needless to say, this puts quite a damper on my faith in the coverage
> reports. Maybe that machine needs a software update?

Hmm, wow. The count for that line is the same as the count at function
entrance, so maybe it is being considered as the normal output path of
the function somehow. I suspect this is a bug in gcov.

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