Scriptlet failures that happen on uninstall can be detected by doing a yum reinstall on the package. However, testers don't normally do this, so they usually don't get detected until the next update, and it can be difficult to find which of the many updates is generating the error message. Would it be possible to automate something like a yum reinstall on each new package and capture any resulting errors?
That would be possible to automate in AutoQA once we are in a position we can automate stuff easily (not yet) :)
But since you raised this up, I already wondered several times - when I see a scriptlet error during update, is there an easy way to tell which package generated it? Often it is not clear at all and I don't know how to find it out. Shouldn't be easy for yum to prepend the error message with a package NVR that caused it?