Extensibility

I am not precisely sure, how - in the TMRQL approach - functional extensions are handled. Semantically speaking, functions, but also predicates, more terminology, other rules, etc. are all part of an application domain. Ideally, one would use an ontology language for it, depending on the expressitivity needs.

Probably one would have to write more SQL functions and add other tables. Either using SQL stored procedures or - given the, uhm, elegance of stored procedures, these functions are then written as part of the application.

What I can see in TMRQL is that one table - tm_tc - is such an arbitrary extension. It holds the information which topics are linked by using associations (of particular type) and building the 'closure' (paths of any length). Yes, I agree, that such a thing is quite useful. But that is also very specific and seems to be already part of the infrastructure. And, can it be controlled which association types do have this 'transitivity' property? How many such special solutions would we have to add over time to such a language?