On Wed, Jul 01, 2015 at 10:03:44 -0400, John Ferlan wrote:
The following Coverity issues are flagged in Coverity 7.6.1 compared
to not being seen in 7.5.1 - this series just addresses those issues
before 7.6.1 is installed in our upstream Jenkins environment:
http://jenkins-virt.usersys.redhat.com/job/libvirt-coverity/
This is a Red Hat internal link where other users don't have access to
this definitely should not be published to a mailing list at least to
avoid frustration to non-redhatters.
It seems the bulk are essentially false positives, but we can do something
in order to work around them.
So why are we doing anything about that? Coverity is the software that
should be fixed here.
While coverity is a very useful tool in some cases where it reports
actual errors the noise it generates is sometimes quite unbearable and
as it looks the new version added just a few false positives, but no
actual fault.
Additionally if we continue to patch up the mistakes rather than
reporting it we might as well as end up by flagging something with the
sa_assert() macro that will change into an actual error in later
patches and then we won't be able to detect that.
As of such I think that libvirt should mostly fix just actual errors
found by coverity and people who run coverity on the libvirt code base
should rather report the errorrs to the coverity vendor to fix the false
positive notifications rather than working that around by silencing it.
</rant>
Peter