Sequence diagram: Condition for all lifelines

[MrDX]

Hi,
i am searching for a way to define a condition for all lifelines at ones (it's a global state). The StateInvariant covers only one lifeline, so I have to resize it to cover more than one lifeline. The problem is, that the XMI-Exporter does not save all covered lifelines (UML says, that a StateInvariant only have one covered lifeline).

Is there an other way to add a condition, that covers all lifelines?

[«Midnight»]

I take it that you mean you want a quick way to establish a set of state invariants, one for each lifeline. This would not invalidate the diagram (according to UML) by creating a single state invariant that applied to more than one lifeline. Is this correct?

David

[MrDX]

I don't mean a set of state invariants.
Perhaps, a example helps to demonstrate my prob:
I have three objects for devices, every device having its own states (Ready, Failure, ...). And there is the complete system with all devices and a global state (Global_Ready, ...).

To add the StateInvariant over one lifeline can be missunderstood, a condition over all lifelines would be easier to understand.

Another example:
http://www.item.ntnu.no/fag/SIE5020/MSC_HTML-version/02-03_MSC-Basics/02-03-04-CoregionSubmessage/exercise1.gif
The instances X,Y,Z must be 'Idle', after sending some messages they must be 'Idle' too (for example for test specification).

[«Midnight»]

i am searching for a way to define a condition for all lifelines at ones (it's a global state). The StateInvariant covers only one lifeline...UML says, that a StateInvariant only have one covered lifeline

It sounds like you need to define something a bit different. You are definitely looking at a 'global' situation, rather than a 'common' one. Offhand, I'm not sure what exactly is needed.

David

PS: I cannot open the page you referenced.

[MrDX]

Oh, there is a '*' instead of an 'a' in the link. New correct one:
http://www.item.ntnu.no/fag/SIE5020/MSC_HTML-version/02-03_MSC-Basics/02-03-04-CoregionSubmessage/exercise1.gif

Perhaps, the picture helps you to understand my problem ;-)

PS: The forum seems to replace the first 'a' with an '*'.

[MrDX]

Anyone an idea?

[KP]

PS: The forum seems to replace the first 'a' with an '*'.

I've fixed that for you. It's the crafty mechanism built into the forums to prevent name calling

[MrDX]

I've fixed that for you. It's the crafty mechanism built into the forums to prevent name calling

Thank you

[KP]

Anyone an idea?

Is it a Continuation? They have the same notation as a State Invariant but may cover multiple lifelines.

[MrDX]

I think so too, but Enterprise architect can only cover one lifeline (snaps to one of them). The other problem:

The OMG UML specification (UML Superstructure Specification, v2.0, p. 414) states:
"A Continuation is a syntactic way to define continuations of different branches of an Alternative CombinedFragment. Continuations is intuitively similar to labels representing intermediate points in a flow of control."

The OMG UML specification (UML Superstructure Specification, v2.0, p. 415) also states:
"Continuations have semantics only in connection with Alternative CombinedFragments and (weak) sequencing. If an InteractionOperand of an Alternative CombinedFragment ends in a Continuation with name (say) X, only InteractionFragments starting with the Continuation X (or no continuation at all) can be appended."

=> The UML specification allows continuations only for CombinedFragments.