Raise your level of abstraction
Tom and I where thinking about this question and I'd like to show present your answers. OK the article is written in German... Will there be a cg2015? If yes I'll submit this article and will provide an English version of it:
What do do you think?
thank you for asking for feedback and raising that question here. I was already about writing something about the OS issue, but did not do so due to german/english.
I read your article about Meta-Architecture and learned a valuable classification of MDD-setups, corresponding well to my own experiences in the field. I think it provides very good guidance for respective projects.
Knowing the history and present of MDx well, I still was rather puzzled by the given premise about it. Specifically, I stumbled over the statement, that only domain specific abstractions are "true" abstractions over program code. Surely past UML usage in probably next to all cases did not meet this goal of being substantially more abstract than code, but abandoning the whole approach in favor of DSLs seems like throwing out the baby with the bath water to me.
More than that, my experience shows that there are levels of abstraction like this:
Technology here is used a bit simplified, it can further be subdevided in things like architetural aspects down to programming language specific things and worse.
There is a telling slip of the tongue in the editorial of the OS issue; translated coarsly "...have doubts that MDx still has future... ...until they get rid of heavy-weightedness...". The "...stilll...until..." does not sound smooth to me - yet the sentence might reflect the current state of MDx very well: the first part (no future) reflects the present deep depression of MDx approaches, while the second might point to what I also sense, which is that under the burned earth of past approaches already fresh green MDx stuff springs again, since - to my opinion - the need for (improved) MDD approaches is as pressing as it never has been before.