I was wondering if you show the servicelocator in your sequence diagram? My SQDs are pretty cluttered already and if I put a servicelocator in it, they will become even more nasty. I know that in Cades book a servicelocator is shown in the SQD. However, I was hoping it will be enough to decribe it in the accompanying text, that every EJB etc. must be looked up using a SL. I think the SQDs are on an architectual level and not on design or implementation level. Therefore I would hope that an SL is not really necessary in the SQD. Like I said, I mention it in text and in the component diagrams.
my SD's are also quite loaded, so I have decided not to insert Service Locator and so on. I will explain it in a document. It is very dependent on the detail level of your solution. You should show such components which plays the most import role in your design.
thanks for your quick replies. I am also wondering if I can name my SQD methods "conceptualy". Meaning, since we are doing the architeture of the system, it should be enough to give away conceptual names for the methods in each SQD. It is up to the designer to come up with the real method names. To take this to an extreme, I could see myself just writing "plain english text" on the method arrows instead of breaking my head on the final method names which probably will chance during the design phase anyways.