Thanks for your suggestion, @atripp, that's definitely something we'll look into when we find time to do so. In the meantime we have implemented the workaround described above, so thanks again for your help.