|Ordering in Sequence Diagrams [message #1095105]
||Mon, 26 August 2013 12:29
| Belen Rolandi
Registered: August 2013
Hi, i know this is an uml question, but i posted it in the uml2 forum and i got no answer, so i thought that maybe someone here might help me.
I'm working with the UML metamodel for sequence diagrams and i have a question regarding the ordering of occurrence specifications. According to the UML spec, occurrence specifications are ordered along a lifeline, but it says nothing about how this ordering is actually determined.
Can it be drawn from the physical ordering (position) of elements in the coveredBy list of each lifeline? Or it needs to be specified using additional ordering constraints, i.e., by menans of the GeneralOrdering element?
Can anyone help me clarify this issue?
Thanks in advance!
|Re: Ordering in Sequence Diagrams [message #1095764 is a reply to message #1095105]
||Tue, 27 August 2013 09:45
I understand that OccurrenceSpecifications are ordered on a single
lifeline, but not across multiple lifelines (i.e., the ordering of
events on different timelines that are not connected via a message is
The GeneralOrdering is an (additional) constraint to specify the
ordering of two events that are (usually) not on the same lifeline.
Hence, you don't need this to find out the order of occurrences on a
Hence, I think you should use the coveredBy property for the ordering,
and GeneralOrdering does not need to be added unless necessary.
The UML 2.5 spec will clarify this further I think. From the beta spec
The set of valid traces is constrained by a partial ordering of the
event occurrences in the traces. Likewise, the set of invalid traces is
also constrained by a partial ordering of the event occurrences in the
In an interaction diagram each vertical line describes the time-line
for a process, where time increases down the page. The distance between
two events on a time-line does not represent any literal measurement of
time, only that non-zero time has passed.
The instances in an interaction in principle operate independently of
each other. No global notion of time is assumed. The only dependencies
between the timing of the instances come from the restriction that a
message must be sent before it is received.
Along each instance axis the time is running from top to bottom,
however, a proper time scale is not assumed. If no coregion or parallel
operator is introduced a total time ordering of events is assumed along
Events of different instances are ordered via messages, or via the
generalized ordering mechanism. See 17.4.3 (Message). A message must
first be sent before it is consumed. With the generalized ordering
mechanism "orderable events" on different instances (even in different
interactions) can be ordered explicitly. No other ordering is
prescribed. An interaction specification, therefore, imposes a partial
ordering on the set of events being contained. A binary relation which
is transitive, antisymmetric and irreflexive is called partial order.
On 26/08/13 18:29, Belen Rolandi wrote:
> Hi, i know this is an uml question, but i posted it in the uml2 forum
> and i got no answer, so i thought that maybe someone here might help me.
> I'm working with the UML metamodel for sequence diagrams and i have a
> question regarding the ordering of occurrence specifications. According
> to the UML spec, occurrence specifications are ordered along a lifeline,
> but it says nothing about how this ordering is actually determined.
> Can it be drawn from the physical ordering (position) of elements in the
> coveredBy list of each lifeline? Or it needs to be specified using
> additional ordering constraints, i.e., by menans of the GeneralOrdering
> Can anyone help me clarify this issue?
> Thanks in advance!
Powered by FUDForum
. Page generated in 0.11423 seconds