Databases Reference
In-Depth Information
and that provide limited information about how the OSS components are integrated.
Only a few studies have conducted large-scale field studies representing several in-
dustrial sectors [ 14 , 32 , 34 ]. However, all of these studies provide little concrete
advice related to OSS component selection and knowledge management activities,
which are two of the pressing problems that software integrators face every day.
The following subsections provide a brief background on OSS components se-
lection and knowledge management, and summarize the body of evidence that exist
in the area based on published surveys.
9.2.1 State-of-the-Art Component Selection and Knowledge
Management
Roughly speaking, component selection consists of three activities that are usually
staged [ 19 , 31 , 39 ]:
1. Identification of candidate components : It is aimed to locate one or more candi-
date components that may cover some of the system requirements (while avoid-
ing non-relevant components) and to acquire information that makes their evalu-
ation and comparison feasible.
2. Evaluating components with respect to the expected requirements : This activity
aim to assess to what extent the candidate component(s) covers/cover a major
part of the system requirements.
3. Choosing suitable component alternative(s) : This refers to the comparison of the
candidate components to choose the one(s) that “best” fits/fit the stated require-
ments.
However, in all the above selection activities, non-functional requirements (also
called quality attributes) - like reliability, security, usability or maintainability -
are hardly covered, as they express hard-to-capture and late-emerging system-level
properties. On the other hand, most OSS components seem to fare satisfactorily on
quality issues [ 34 ]. Evidence exists that the practitioners' perception of OSS in the
embedded systems area is also satisfactory [ 35 ].
9.2.1.1 Identification of OSS Components
Searching for reusable components was traditionally supported by centralized com-
ponent repository systems with specific classification and searching mechanisms
[ 21 ]. However, the free availability of OSS components has shifted this focus to a
global reuse approach [ 41 ]. Much effort to support component searching have been
devoted to classification structures and specialized search engines. Birkmeier and
Overhage provided a comprehensive overview of this in [ 8 ]. Several classification
Search WWH ::




Custom Search