Summary

Overall, I think it is worth developing a dedicated language. TMQL, at its current complexity, does not overwhelm smaller development teams and first results show that it can be implemented more or less efficiently depending on your expectations. I also believe that TMQL is an industrial language, while it will hopefully not trigger regurgitation (yeah, I had to look that up) in purists.

The specification will be very declarative, leaving implementation strategies as open as possible. It will be perfectly ok, for example, to compile a TMQL query expression into a SQL select statement.

A valid concern is that developing a new language carries several risks.

• Getting it right is not easy. To make Java, C#, Perl, Python developers happy is virtually impossible, given their significantly different mindsets when it comes to programming languages. Still, TMQL offers different surface syntaxes which may strike different chords in different communities. And these syntaxes can be mixed, so that there is always a migration path.

• TMCL, the constraint language, is an imponderabilium. Ideally, both, TMQL and TMCL should fit together, but this is something which is not managed at this point.

• And, honestly, we do not really understand the range and scope of applications which will come in the next 5 years.