Civil Engineering Reference
In-Depth Information
Quite often weaknesses of the component specification are discovered, when
developing the related test scenarios. As such, even the process of TCs generation
per se makes sense, as for that purpose especially checks of the individual functions
of the implementation are done, which require a unique interpretation of the com-
ponent specification. When performing a conformance test only unique verdicts
are allowed, which discriminate clearly whether or not a test candidate performs as
required by the component specification. Thus precisely defined limits are given,
within which an implementation must perform, if it shall be compliant to the stan-
dard. The benefits of this kind of tests obviously help to discover and avoid interpre-
tation vagueness. This vagueness inevitably would lead to problems in a system, in
the event that components must be exchanged and/or implementations of different
manufacturers are used in the same networked system.
Well-known companies integrating those standardized components require con-
formance tests as the basic precondition, if those components of a specific supplier
are to be applied. This is not only the case in automotive industries, but also in most
technical areas. Especially, conformance tests apply when using such components
in communication-related areas such as telecommunication up to high-speed com-
puter interfaces. Again this results from the application of components in networked
systems, which are supplied by different manufacturers.
When applying tests in general and especially conformance tests, problems can
be detected in an early phase and not just in the late phase of integration into the fi-
nal end product. At that late phase, problems may be even difficult to discover; may-
be they even can only be observed under very specific conditions. When performing
conformance tests on a single component partially, very specific and detailed test
scenarios are applied, while systems tests typically focus on known application cas-
es and error cases. As such, a potentially non-standard-conform component may fail
when used by the end user. This can lead to a recall action and thus cause tremen-
dous cost for correcting that problem. Typically, before going into the integration
phase, the supplier of a component himself applies various tests. However, these
tests are based on the interpretation of that specific supplier. As such, these tests do
not guarantee this component to be compliant to the standard (see Fig.
6.3
).
For currently developed standards, the specification of related conformance tests
is done in parallel. Quite often, this is done by consortia, which are put together
especially for that purpose (in automotive industries, there are, e.g. FleyRay con-
sortium or AUTOSAR consortium). This procedure makes sure that from the begin-
ning, only components are applied which are conforming to the standard. In addi-
tion, this procedure avoids that standards must be altered in a later phase, to make
them compliant to the already-available implementations. Obviously this procedure
does not make sense.
• Some purposes of conformance tests
− Proof of functions and timely behaviour of different implementations as
required by the standard.
− When developing conformance tests, the related component specification
is analysed very deeply, discovering various ambiguities and problems as
