@bob @moonman @takeshitakenji My general approach has been to hold changes in Nightly and let them be tested.  If it ends up they screw something up, we can always revert it if it's bad enough to not be easily fixed, but the damage you do to people's willingness to contribute can be severe, if you reject them outright.  The important thing here is to make it collaborative, rather than advesarial.  If you're working with the contributor to resolve whatever problems arise, they're much more respected, and in turn that tends to make them much more likely to stick around, I'd think, than if you're disrespectful of them.