We have a multi-user environment with different screen settings. It seems that every time Capella opens a diagram on a different screen resolution, it modifies the location of objects. Is there a way to prevent this?
This is extremely surprising. I use for example a dual screen and there is no difference when I open/modify a diagram on one screen or on the other. Could you elaborate on the problem? Maybe send screen captures?
I can’t send you screen captures, but in our environment it’s quite easy to reproduce. Whenever a session is started in Capella, opening a diagram automatically marks it as modified. When you save the model afterwards, all modifications get saved of course, causing a huge amount of changes which only contain relocations of objects.
Maybe there are some workbench preferences which we aren’t aware of?
Are all contributors using the same Operating System?
I know I have differences between Mac and Windows.
I understand your models are confidential. But if it is easily reproducible, maybe you can do it on a sample dummy model?
OS is almost the same for everyone (win7 and win10). We have a nightly job running which automatically exports diagrams, but this one does not modify the source files.
However, I think I’m getting to the problem, and it seems to be more related to my other question I posted. This morning I freshly pulled the source files, opened one of the bigger diagrams and directly saved it. Checking the differences shows that a lot of nodes and connectors have been added to the diagram, only by opening it in Capella. These connectors might have been added before within a different diagram, but not the big one.
So, would it be possible to inform the user about possible modifications in other diagrams whenever he does something? The impact analysis already gives hints about related elements, maybe this can be extended to affected diagrams?
Same answer as on the other post, implementing this feature is not planned. Because it is much more complicated that it seems (it would means having all diagrams loaded in RAM permanently, and interrogating all of them every time a change is performed on an element). I think the semantic browser is what gives you the best idea of the impacts: for one given element you see all the diagrams in which it appears and you can directly open them.
Would there be a possibility to use the query interpreter for this kind of actions?
I don’t understand what the query interpreter would give you that the semantic browser does not. Could you elaborate?
The query interpreter could be triggered independently of the GUI, e.g. as a Gerrit patchset trigger. That way we could block model changes with a poor quality.
What would be the query you would trigger (in spoken language)? What would you be looking for? All diagrams that display elements that have been modified? How would you get a list of the modified elements? And once you have a list of diagram, wouldn’t you need to open them all to check their layouts?