solution vs. problem space
types of variation in problem space:
offerted by feature models: optional mandatory alternatives set selection
other dimension features that go into code into process into
in general: everything that can be decided on
features lack semantics
types of variation in the solution space = Operations: substitutive variation (replace) additive variation remove
other dimension: functionality and non functional things
mechanisms for variation in solution space: ifdef aspects templates ...
We want to have: Patterns for doing variations
If you have this kind of variation with this kind of feature and are restricted to this set of mechanisms implement it like this.
exposing variation points
runtime variation on demand requirments changed during runtime (for high availability systems) change the code during runtime, not only configurate it
ways to do this: AO : offers runtime weaving/hot deployment flexible component systems: replace, add, remove components patching at runtim ...
what is missing?
same problem as evolution
mapping several feature models on each other? crosscutting feature over one feature model?
-- DaniloBeuche - 10 Nov 2004