Limitations of named row types in Universal

Named row types seem like a wonderful idea, but we've been disappointed in the poor integration of them in our IDS w/ UDO 9.14UC1 database. Limitations that I've encountered include (a) inability to index, sort on or group by contents of name row types; (b) various bugs in using NULLs and named row types (e.g., reproducibly crashing SQL Editor with certain combinations of NULLs in named row types, hanging dbaccess in other conditions, problems declaring named row attribute as NOT NULL in table definition, etc.); and (c) problems resolving procedures whose parameters are LIKE some table column that is a named row type (you have to explicitly say what the named row type is in the procedure declaration). Bottom line, I'm getting increasingly nervous about Informix's implementation of named row types. Is anyone using these features in a production application? Do later releases address any of these bugs? What are people concluding about named row types? Rob