Interesting ... I've done a little poking around. Being able to hook up arbitrary hardware to your website (or expose your CD ROM player to anyone on the net) is indeed probably something not on the horizon here!
I find the home page claim:
MMBase separates content from layout, in this way information can be reused easily.
interesting when they then define SCAN, a scripting language not terribly unlike the ATS language in look-and-feel, EXCEPT that SCAN includes tags to embed SQL snippets directly.
So a web designer using this apparently is supposed to understand a little SQL on the side. If they use SCAN - they also say that the community focuses on JSP. This means that their target web designer is really a JSP scripter, much like ours is a Tcl scripter.
Anyway, while all the content is in the DB (sounds familiar) any SCAN page that retrieves content has to know something about how things are stored in the DB. So "content is separate from layout" but the folks doing layout at the "web designer" level need to know how to tease content out of the RDBMS.
It would be interesting to see what their object datamodel looks like ...
Their editor section (docs for people called "editors") talks about a bunch of canned editors (canned doc types) you can use to add content of various types. This sounds conceptually about what ETP+Templates wants to be when it grows up and has a lot more templated options available for it "out of the box"? Or what CMS+Templates wanted to be except maybe MMBase is a lot more user-friendly in the UI department?
This is all from a 15 minute reading of docs. I get the impression that they actually divide the world in much the way the OpenACS 4 world is divided except in our world if you want to hack packages you need to learn Tcl (while in practice it sounds like you probably need to be able to hack JSP, not just SCAN, pages in theirs). In principle non-hackers might be able to customize ATS pages a bit more easily than SCAN pages (or at least with less risk of mucking them up) simply because SCAN pages do have those embedded SQL snippets in them.
My other impression is that on the pure content managment level they're considerably farther along than ETP or CMS. It might be worth investigating for ideas on UI etc.
And as Jun says, they do some things that we just don't and probaby won't do in regard to distributed stuff including distributed hardware stuff.
