>> Wow, really? Then why wouldn't RedHat or CentOS have a fixed updated
>> version in their repo? That seems egregious if what you say is indeed the >> case. > > RedHat (and CentOS, since their whole mission is to match RHEL > feature-for-feature and bug-for-bug) believes that their Enterprise Linux > customers value consistency over currency. They release updates to patch > security holes, but their general attitude is that if Red Hat 5.0 shipped > with foo_1.1.3 in 2007, then Red Hat 5.7 should also ship with foo_1.1.3 > because their customers may have whole workflows built around the way > foo_1.1.3 handles a specific command flag and foo_1.2.7 may have changed > that. If necessary, they'll backport security patches from later versions > of foo back to the current, leading to RPM names like foo_1.1.3-17.el5_7 > -- but they won't add feature changes unless absolutely unavoidable. Sure, but the point is that my spamassassin and per-Net-DNS (where the error is happening?) are up to date from the CentOS repo.... so shouldn't they work without an error when spamassassin restarts?