Model-Driven and Software Product Line Engineering - Model-Driven and Software Product Line Engineering - page 110

generate products, there is a need to use a mechanism
that allows designers to select and execute transformation
rules automatically according to variants selected by product
designers through the configuration process. This mechanism
must also ensure the correct execution ordering, also called
execution scheduling, of the selected transformation rules. By
correct, we mean an execution ordering that allows derivation
of the required configured product.
There is, then, the need of introducing decision models
in the context of MD-SPL as a mechanism for composing
transformation rules based on product configurations. In
section 2.7.2, we presented a decision model as a model that
captures variability in a product line in terms of open decisions
and possible resolutions [BAC 00]. Each decision is expressed
in terms of a selected variation point and associated with a
set of possible resolutions, which in turn refer to variants of
selected variation points.A set of effects is associated with each
possible resolution. An effect indicates how a particular core
asset is re-used to create a product line member. Thus, in the
context of MD-SPL engineering, decision models must capture
(1) the relationships between variants and solution space core
assets, such as transformation rules or re-usable source code,
and (2) the required execution ordering of transformation rules
to weave and transform models and/or source code to create
products based on product configurations.
4.5. Variability expression and product configuration
The most common mechanisms used to capture variability
and configure products in current MD-SPL approaches are
metamodels and feature models. As part of our approach to
create MD-SPL, playing the role of product line architects, we
use metamodels and feature models as our base core assets.
Playing the role of product designers, we configure products by
creating (1) models that conform to metamodels and (2) feature
configurations.