|
Hi,
Is it possible to get the formats in which different tools use to emit XMI while exporting file ?
As of now, there is no standard XMI format that all tools follow isn't ? Why is it so ? Is there any possibility that this could become possible in the future ?
Thank you.
| | vijayaraghava Wednesday, April 12, 2006 6:17 PM | XMI is a serialization of a MOF-based metamodel. MIcrosoft has not universally embraced MOF or UML. Instead, Microsoft is developing a what it calls a Domain Specific Language Toolkit. See http://blogs.msdn.com/stevecook/and http://blogs.msdn.com/billgibson/. | | Chris Lovett Tuesday, April 25, 2006 4:45 PM | Yes, we do provide an add-in to Visio that lets you export parts of a UML model in XMI format. However, the answer to your question of why there is no common standard that all uml tools follow is quite a complex one.Hereis an excerpt fromSteve Cook's article on Domain Specific Modeling and Model Driven Architecture that talks about XMI:
First of all, the XML generated is dependent upon the language definition. This means, for example, that the XMI serialization of a model defined in UML1.4 cannot be understood by an implementation of UML2.0, even in those areas of the language that are effectively identical from the user’s perspective. Secondly, the definition of XMI itself is subject to change. This means that there is XMI1.0 serialization of a model in UML1.3, an XMI1.1 serialization of the same model, etc. And, thirdly, the definition of MOF is also subject to change, which adds a further term to the set of possible combinations; indeed, the latest proposed version of MOF has two flavors, which are not fully compatible with each other. Thus, although XMI claims to provide interoperability of modeling tools, in practice, unless every tool supports every possible combination of each of the MOF, XMI and UML standards, interchange is problematical. A further problem with XMI, especially with early versions, is that its documents are instances of machinegenerated schemas that tend to be verbose and difficult to read; this goes against the grain for most developers who will want to apply the many readily-available technologies for visualizing, transforming, and manipulating XML documents.
As Chris mentions you may want to read the blogs by Steve Cook and Bill Gibson.
Cheers,
Ramesh. | | Ramesh Rajagopal - MSFT Tuesday, April 25, 2006 10:22 PM | You should ask this question of sombody who produce XMI tools or specs. | | Sergey Dubinets - MSFT Friday, April 14, 2006 2:09 AM | Hi,
Could be please answer the second part to the question ?
"Why isn't there a common standard that all tools follow ? Will it become possible in the future ? "
I searched around but itseems the formats aren't available for public use.
Thank you.
| | vijayaraghava Friday, April 14, 2006 10:01 AM | Sorry. I can't answer this question. I don't know anything about XMI.
System.Xml doesn't use or support it.
Do you know any tools from Microsoft that support XMI format? | | Sergey Dubinets - MSFT Friday, April 14, 2006 4:01 PM | Hi,
I think in MS Visio there is support for XMI. Diagrams can be imported / exported to XMI.
| | vijayaraghava Friday, April 14, 2006 10:08 PM | XMI is a serialization of a MOF-based metamodel. MIcrosoft has not universally embraced MOF or UML. Instead, Microsoft is developing a what it calls a Domain Specific Language Toolkit. See http://blogs.msdn.com/stevecook/and http://blogs.msdn.com/billgibson/. | | Chris Lovett Tuesday, April 25, 2006 4:45 PM | Yes, we do provide an add-in to Visio that lets you export parts of a UML model in XMI format. However, the answer to your question of why there is no common standard that all uml tools follow is quite a complex one.Hereis an excerpt fromSteve Cook's article on Domain Specific Modeling and Model Driven Architecture that talks about XMI:
First of all, the XML generated is dependent upon the language definition. This means, for example, that the XMI serialization of a model defined in UML1.4 cannot be understood by an implementation of UML2.0, even in those areas of the language that are effectively identical from the user’s perspective. Secondly, the definition of XMI itself is subject to change. This means that there is XMI1.0 serialization of a model in UML1.3, an XMI1.1 serialization of the same model, etc. And, thirdly, the definition of MOF is also subject to change, which adds a further term to the set of possible combinations; indeed, the latest proposed version of MOF has two flavors, which are not fully compatible with each other. Thus, although XMI claims to provide interoperability of modeling tools, in practice, unless every tool supports every possible combination of each of the MOF, XMI and UML standards, interchange is problematical. A further problem with XMI, especially with early versions, is that its documents are instances of machinegenerated schemas that tend to be verbose and difficult to read; this goes against the grain for most developers who will want to apply the many readily-available technologies for visualizing, transforming, and manipulating XML documents.
As Chris mentions you may want to read the blogs by Steve Cook and Bill Gibson.
Cheers,
Ramesh. | | Ramesh Rajagopal - MSFT Tuesday, April 25, 2006 10:22 PM |
|