In the past the question has been raised from time to time if umbrello should be refactored to use more of the Qt model/view approach. This bug is for tracking this ongoing work. Reproducible: Always
Emmanuel Lepage Vallée wrote some related stuff at https://bugs.kde.org/show_bug.cgi?id=87104#c9
Bug 345031 has been opened for the evaluation of a few model/view classes implemented by Emmanuel.
At https://bugs.kde.org/show_bug.cgi?id=87104#c5, Ralf Habacker wrote: > (In reply to Oliver Kellogg from comment #4) > > [...] > > How would this class hierarchy view be different from the current tree view? > The recent tree view shows a "package hierarchy" in the opposite to the > derived and base classes , which would form the hierarchy. Ah, I see. I suppose the hierarchy would hinge on the generalization/realization associations among classes. OTOH, generally I feel it unsatisfactory that Umbrello lacks a representation of the UMLAssociations in the list view. See for example the Logical View at http://www.uwgb.edu/breznayp/cs372/rational.htm : Rational Rose uses a simple line for (uni and bi directional) associations. Further, it uses a line with a solid arrow for generalization/realization, and a dashed line for dependency. The advantage is that this includes not only generalization/realization but all UMLAssoication types. The UMLAssociation can be manipulated in the tree view independent of diagrams, for example can be moved or deleted, or a double click could open a properties dialog for the association. Perhaps I should open a separate wish for this?
(In reply to Oliver Kellogg from comment #3) > At https://bugs.kde.org/show_bug.cgi?id=87104#c5, Ralf Habacker wrote: > > (In reply to Oliver Kellogg from comment #4) > > > [...] > > > How would this class hierarchy view be different from the current tree view? > > The recent tree view shows a "package hierarchy" in the opposite to the > > derived and base classes , which would form the hierarchy. > > Ah, I see. > I suppose the hierarchy would hinge on the generalization/realization > associations among classes. Looking into refactoringassistant.cpp shows, that the refactoring assistant is based on UMLAssociation instances, so there is no difference under the hood. > OTOH, generally I feel it unsatisfactory that Umbrello lacks a > representation of the UMLAssociations in the list view. > See for example the Logical View at > http://www.uwgb.edu/breznayp/cs372/rational.htm : > Rational Rose uses a simple line for (uni and bi directional) associations. > Further, it uses a line with a solid arrow for generalization/realization, > and a dashed line for dependency. > The advantage is that this includes not only generalization/realization but > all UMLAssoication types. looks good. > The UMLAssociation can be manipulated in the tree view independent of > diagrams, for example can be moved or deleted, or a double click could open > a properties dialog for the association. which would also fix bug 332371 :-) > Perhaps I should open a separate wish for this? yes, will be good to track related commits. There is still the question, if such an implementation would include refactoring the tree view to the model/view design pattern ?
With bug 73847 the recent tree view has been extended with a model/view based tree view showing stereotypes. The related git branch is here http://quickgit.kde.org/?p=umbrello.git&a=shortlog&h=43856d05914bd58c172702d84471408acafd6b24
With 2.17.2 (bug 345546) a model/view based dock window containing stereotypes has been added.
With 2.19.2 a model/view based dock window containing all diagrams has been added.