Information Technology Reference
In-Depth Information
transformation we created to generate oAW workflows from
decision models is available at [ARB 12].
1
2
3
4
5
6
7
8
9
10
11
<fineFeature toFeature=
"
Automatic
"
b oundMetaconcept=
"
Window
"
>
<transformationAspect
adviceTarget=
"
domainWindowsToFacilitiesWindows
"
>
<extensionAdvice
value=
"
particularWindowsToAutomaticWindows
"
/>
</transformationAspect>
</fineFeature>
<transform id=
"
domainWindowsToFacilitiesWindows
"
>
<invoke value=
"
domainWindowsToFacilitiesWindows
"
/>
</transform>
Listing 5.3.
Example of a generated oAW workflow
5.4. Identified limitations
In previous sections, we discussed how our decision models
are useful to capture the relationships between features and/or
bindings, and specific transformation rules, and how the
requiredexecutionschedulingoftransformationrulestocreate
products is based on feature configurations and/or binding
models. Our idea to obtain a final execution scheduling was to
construct a baseline scheduling,which is modified according to
execution conditions defined in terms of feature configurations
and/or binding models. We have identified at least three
limitations in our strategy of relating execution conditions
to specific transformation rules. Two of them occur when
conditions only take into account feature configurations (see
Table 4.1). The other one occurs when conditions take into
account not only feature configurations, but also binding
models (see Table 5.1).
5.4.1.
Features combinatorial
The first limitation of our approach is that the number of
valid feature configurations,which can be created based on one
feature model could be high. In our current approach,we do not
include mechanisms to guarantee that either for all possible
Search WWH ::
Custom Search