> After some revision and a fair bit of research I have found the issue. I should add that this error can be observed (but, thanks to IDP-456, with diagnosable logging) simple by removing the friendly names from the attribute encoders in our sample configuration. R