Hi Sara, Your meta-model of chicken and eggs sounds really nice :). Did you also manage to have relationships between entities? Currently, we synchronize the meta-descriptions only manually. But, indeed it would be nice if we could recompute the meta-descriptions at compile time or unload time. This is part of the future work, and until then . I saw that you used System reinitializeDefaultSystem. However, at the moment in the case of Moose you need to call AbstractEntity initializeAllMofDescriptions, or just used from the VW main menu: Tools/Moose Utilities/Reinitialize meta descriptions. This deals with some specific meta-annotations related to Moose. Cheers, Doru On Jun 26, 2007, at 12:57 PM, Stéphane Ducasse wrote: -- www.iam.unibe.ch/~girba www.iam.unibe.ch/~girba/blog/ "There are no old things, there are only old ways of looking at them."

Hi Sara, Mondrian is a generic visualization framework and it is part of Moose Config. Mondrian has a graph model behind, it works at the level of Smalltalk objects and allows you to script your visualization. For example, if you have a FAMIX model in a variable called mooseModel and you want to show the hierarchy of classes you can do: view := Mondrian.ViewRenderer new. view nodes: mooseModel allClasses. view edges: mooseModel allInheritanceDefinitions from: #superclass to: #subclass. view treeLayout. view open. You can find more details here: http://smallwiki.unibe.ch/moose/tools/mondrian/ But, could you elaborate on what you have in mind when you say generic visualizations? Cheers, Doru On Jul 2, 2007, at 10:22 AM, sellossa(a)ensieta.fr wrote: -- www.iam.unibe.ch/~girba www.iam.unibe.ch/~girba/blog/ "One cannot do more than one can do."

Hi Sara, I understood that you would want a visualization that would work for any meta-model, but what I did not understand is: What would the goal of the visualization be? What you probably want is some visualization that would just interpret the EMOF description in some way. One generic visualization I can think of is to see what kind of instances are there loaded and how many of each are there. But, this would basically be not much more than the list we show in the browser. Another possibility would be to have all objects in the model, or at least a collection of them, and then show all the pointers between them. I quickly put together this visualization and I attached an example of the LAN model: each node is an entity colored according to its type, and each edge represents a pointer to another entity. The nodes are arranged according to a force based layout (which is very slow). As you can see, the problem here is that even with a small model, you can get complex pictures. The visualization is available in MooseModel>>viewAllEntititesOn: in the MondrianPaintings bundle. Many times the meta-model of the case study and and the meta-model of the visualization are not the same. For example, in a FAMIX meta- model, InheritanceDefinition is a first class, just like Class, but in a class hierarchy visualization, classes are represented as nodes and inheritances as edges. In this case, the naive generic visualization I described above would probably map inheritances on nodes as well. I mentioned Mondrian, because the idea behind it is to provide a kind of an infrastructure with which you can quickly script your visualization to show what you want from the model. With Mondrian you besically specify a meta-model transformation. Cheers, Doru On Jul 2, 2007, at 1:04 PM, sellossa(a)ensieta.fr wrote: -- www.iam.unibe.ch/~girba www.iam.unibe.ch/~girba/blog/ "In a world where everything is moving ever faster, one might have better chances to win by moving slower."