-
Key: MARTE-64
-
Legacy Issue Number: 11774
-
Status: closed
-
Source: Fundacion Tecnalia Research and Innovation ( Mr. Huascar Espinoza Ph.D.)
-
Summary:
[GQAM: Domain Model & Profile] The PrecedenceRelation concept (fig 15.3) could be used to specify delays due to e.g., queues. However, PrecedenceRelation only exists in the Domain Model. Its implementation is intended to be supported by existing UML concepts (e.g., forks, joints, etc.) I’d suggest to define the corresponding stereotype (<<PrecedenceRelation>>) in order to allow for modelling delays or other precedence patterns like event divisors: a number of event occurrences triggers an execution.
-
Reported: MARTE 1.0b1 — Fri, 7 Dec 2007 05:00 GMT
-
Disposition: Resolved — MARTE 1.0b2
-
Disposition Summary:
This enhancement is not necessary, since everything it would provide can already be done using scenario Steps, and done better. However the method for showing an event divisor using steps should be added to the text.
Since a precedence relation always leads to one or more steps, properties such as queueing etc can be attached to the step. The step is a better place for these properties because there may be multiple branches in a precedence relation (a branch or fork), each with its own properties. For example, a fork precedence would have a different waiting on each branch of the fork, which would be complex to describe or specify within the fork, but very easy on the following steps.
An event divisor or multiplier can be provided using the CommStep stereotype. We assume that the event is a call invoking a Step. The call can be stereotyped as a CommStep, which has a repetition attribute. If this is set to N, it means the call is repeated N times, giving event multiplication. If it is set to 1/N, it means that every Nth execution of the sender gives a call... an event divisor. If the repetition attribute is specified as a deterministic NFP, it should be defined to be interpreted as 1/N for an integer N, determined by rounding to an integer if necessary. A stochastic division is also possible. -
Updated: Fri, 6 Mar 2015 20:58 GMT