Home » Modeling » Papyrus » Multi model or submodel approach ?
Multi model or submodel approach ? [message #1692567] |
Thu, 16 April 2015 10:22  |
Eclipse User |
|
|
|
Hello,
I want to design a system with papyrus and I want to implement the collaborative work with SVN. I see that in papyrus tutorials they explain how to create and share submodels, but what is the interest of this approach instead of create different models for every subsystem as every submodel create a new file in the view project explorer. Having every diagrams for the whole design in a same model is it interesting to make for example verifications or validations in te model, or is it just to keep the architecture of a system composed of subsystems ?
Thank you,
Lola
|
|
| |
Re: Multi model or submodel approach ? [message #1692905 is a reply to message #1692714] |
Mon, 20 April 2015 10:00   |
Eclipse User |
|
|
|
Hi Lola,
I'm not sure exactly what you are asking, but let me try an answer. I am also unsure whether this is directly related to Papyrus as it appears to be more of a system/SCM architecture question.
When working on large systems, it is often a good idea to separate the model into multiple files ("modules"). this typically helps minimize (but not necessarily eliminate, conflicts when contributions ("check ins", "pushes", etc.) are made by the developers.
The way the model is segmented will depend on many things: architecture of the complete system, hierarchy of the components/modules, development team organization, platforms targeted, etc. The goals are typically to minimize dependencies in general and eliminate cross-dependencies (e.g., module A depends on module B that depends on Module C that depends on Module A) and to minimize conflicts when developers "check in" their changes. The dependencies, would be represented both as UML dependencies and relationships, but also as model imports (so that the module's elements are visible in the importing model). Note that a module should never have dependencies (import) to a module that is higher in the hierarchy.
Typically, each module will have their own diagrams. Modules that "contain" other modules will also have diagrams that show how they use the "sub-modules".
There would also be a single, "system" model that would show the dependencies at the highest level of the architecture.
Does that help?
|
|
| | | |
Re: Multi model or submodel approach ? [message #1693300 is a reply to message #1692905] |
Thu, 23 April 2015 00:17   |
Eclipse User |
|
|
|
Charles Rivet wrote on Mon, 20 April 2015 10:00Hi Lola,
I'm not sure exactly what you are asking, but let me try an answer. I am also unsure whether this is directly related to Papyrus as it appears to be more of a system/SCM architecture question.
When working on large systems, it is often a good idea to separate the model into multiple files ("modules"). this typically helps minimize (but not necessarily eliminate, conflicts when contributions ("check ins", "pushes", etc.) are made by the developers.
The way the model is segmented will depend on many things: architecture of the complete system, hierarchy of the components/modules, development team organization, platforms targeted, etc. The goals are typically to minimize dependencies in general and eliminate cross-dependencies (e.g., module A depends on module B that depends on Module C that depends on Module A) and to minimize conflicts when developers "check in" their changes. The dependencies, would be represented both as UML dependencies and relationships, but also as model imports (so that the module's elements are visible in the importing model). Note that a module should never have dependencies (import) to a module that is higher in the hierarchy.
Typically, each module will have their own diagrams. Modules that "contain" other modules will also have diagrams that show how they use the "sub-modules".
There would also be a single, "system" model that would show the dependencies at the highest level of the architecture.
Does that help?
Hi Charles!
Interesting discussion, since I have, while trying to figure out what is wrong with my imports tried to visualize how my models needs to be interconnected. See the attached image. Actually this image is not exactly how things are at the moment, but rather a concept of a future structure, but still shows the principals of design.

BaseCS is a skeleton control system that my company adapts or refines for a specific customer by adding customer specific function or when necessary, modifying those already in the skeleton system.
Would you say the way it is modeled could be improved in any way?
Regards,
Tomas Sandkvist
[Updated on: Thu, 23 April 2015 00:21] by Moderator
|
|
| |
Goto Forum:
Current Time: Thu Jul 10 05:26:13 EDT 2025
Powered by FUDForum. Page generated in 0.80393 seconds
|