Hi Stef -
Two notes on your message:
Our goal is to have a small image [...]
I will point out that since 3.6 (the first version where we had a distinction between basic and full and where an image got substantially smaller) the basic image has only grown; now to a size where it rivals the size of 3.6 full. Mind you, the *basic* 3.9 image is almost as large as the *full* 3.6. You can draw your own conclusions from that (easily verifiable) fact.
Now .cs are not the future. All the good project work with MC (croquet, sophie, seaside, tweak....
I'm sorry to disappoint you, but for our next project at VPRI we just pulled out of Monticello. In short, Monticello is a great way of doing things if you have an environment which is based on "builds" (e.g., you basically throw existing content away, build a new system and load that content back in).
For maintaining a live system Monticello is simply a nightmare. It's slow, it doesn't work in the right way for incremental migrations and you are spending 90% of your time to deal with situations that change sets solve in a nano-second. It's simply not worth the hazzle for maintaining a live system.
So I wouldn't declare change sets dead quite yet; neither would I claim they are "not the future" when it comes to maintaining a live system. In fact, I believe they are. For example, just compare how long it takes to update a 3.7 to 3.8 image vs. 3.8 to 3.9 - it literally takes *ages*, it requires "extra" change sets to do things that Monticello simply cannot do and by the end of the day updating a 3.8 to current 3.9 alpha doesn't even work. I cannot recall a single case of where this has ever happened with change sets.
I think a discussion about how Monticello fits a working style that can be used to maintain a live system is overdue by now. Having been there, having seen the immense pain Monticello inflicts on both sides of the maintenance chain (not only is it a pain for the person doing the maintenance, it is also a pain for the person on the receiving end of the maintenance) I think we can say with some certainty that Monticello fails in this regard and that another approach is needed.
Cheers, - Andreas