It seems there is a problem too with setting the predecessor. See issue 5697.
I really would appreciate some confirmation that my interpretations are correct for this and the previous issues I mentioned. I'm disappointed we found this so late. The new sequence diagram implementation was based very much on rewriting with all the restrictions on easily dragging messages removed and to generally improve the features for the user to interact more easily. That I think, although rather late, has been quite successful. Unfortunately though there was an assumption that the modelling done by the previous implementation was correct and that was not reworked in the new implementation. Aside from the fact that it simply gets this wrong, part of the problem seems to me is that model-mdr is trying to implement too much for itself. When it comes to sequence diagrams setting activators and predecessors the position of drawing the message becomes far more important and more data needs to be send to the model for it to do its work. At least releasing as we are now will not make the current situation. I would hope we can work on correcting this for the next release. Of course now the sequence2 is a module it could in fact me released on its own if we see fit. So please do test sequence diagrams as heavily as you can and create issues and comment on those I've recently created. I have some further questions that I'll start separate threads for. Bob. 2009/2/9 Bob Tarling <[email protected]>: > I've just created issue 5692, 5693 and 5694. > > Could those more familiar with sequence diagram usage confirm my > interpretation for those issues. > > One of my concerns is how to remedy this and manage the fact that our > previous sequence diagram implementation has been writing incorrectly. > > Any feeling as to priority for these issues? > > As this is behaviour of the existing implementation I partly lean to P3. > > But as we have a new implementation being released that I would hope > would be used far more it would be nice to get this acting correctly > from the start for the new implementation. > > Possibly we could try and fix 5692 but leave 5693 as is for now. > > Bob. > ------------------------------------------------------ http://argouml.tigris.org/ds/viewMessage.do?dsForumId=450&dsMessageId=1136322 To unsubscribe from this discussion, e-mail: [[email protected]].
