SysML 2 support

[Guillaume]

Hi,

I read an interesting overview of the changes in the upcoming SysML v2 (https://issuu.com/ppisyen/docs/ppi_syen_123 -> see from page 35 ).
I noticed that SysML v2 will be based on the Kernel Modeling Language (KerML), a shift from v1 which is based on UML.
The article mentions that SysML v2 examples were developped from PlantUML pilot project so it doesn't seem this change prevents UML tools from supporting SysML v2. Even though SysML v2 is not completed yet, am I right to expect that it will supported in EA once released (estimated ETA 2024) ?
Is there any information that Sparx can share at this stage?

I'm not sure as whether users will see the need to move to SysML 2, and how long SysML 1 will remain. Any thoughts ?

[Eve]

Yes, from memory they are also creating a UML profile that is required to include the full compatibility with the metamodel itself.

If you want to keep up-to-date with the latest, look at https://github.com/Systems-Modeling/SysML-v2-Release/tree/master/doc.

Read what you want into me having that on hand because I can't give anything official from Sparx Systems.

[Takeshi K]

Hi Guillaume,

Just FYI, Tom mentioned SysML v2 in the session of the Global Summit.


In my opinion, SysML v2 seems similar to UML v2. That means it is an ideal and perfect language for some people, but I doubt if everyone wants it and/or everyone can use it. I think it is complex and we need to define more elements to describe a system with v2, compared to v1.x.

[Guillaume]

Hi,

Thank you for your feedback. I attended the Global Summit but forgot that Tom mentioned it
I will look into the doc for more details. One of the main changes in SysML 2 is the concept of Block which is replaced with the Part Definition. I did find meaningful the use of Blocks in the vocabulary, so I wonder what justified this move.

[Eve]

Thank you for your feedback. I attended the Global Summit but forgot that Tom mentioned it
I will look into the doc for more details. One of the main changes in SysML 2 is the concept of Block which is replaced with the Part Definition. I did find meaningful the use of Blocks in the vocabulary, so I wonder what justified this move.

What they have done is make a consistent separation between Definition and Usage and a consistent notation for each of the specializations of those concepts. It should make it easier to learn the language. You see something that is labelled as a part on your diagram you automatically know that it is a PartUsage and it has the equivalent of a classifier relationship to a PartDefinition. The same with an action, you no longer need to know that it corresponds to an Activity, it follows the same pattern and there is an ActionDefinition somewhere. (I know you've addressed that confusion many times here)

[wikitect]

Eve - when SysML 2 gets implemented by EA will there be a script for migration (where the new standard and implementation of that makes this possible) and will there also be a definition of what it is impossile to migrate by script and which will therefore need fixing by hand?

The earlier the latter is available the better as it could be used to inform the extent of work needed and also possibly how to model something as if we know well in advance what is going to 'break' this can be marked for attention beforehand (even if it means just addinfg a 'caution' note or stereotype.

[Eve]

We do have a history of providing migration scripts between major versions of profiles that we added. If you look in the scripting window you can still see 'Migrate SysML 1.2 to SysML 1.3' and 'Migrate SysML 1.3 to SysML 1.5'. SysML 1.4 is skipped in that because we chose instead to upgrade the 1.4 profile to support 1.5.

Having said that, I can't give you an answer about what will be done for SysML 2.

[PDC]

Sorry to resurrect an old topic but I thought it more appropriate than raising a new thread on the same subject.

To be clear - does the introduction of SysML v2 make v1 obsolete? I can't imagine this is really the case, but I can't find any clear commentary on this anywhere online. Or maybe it depends on ones' definition of 'obsolete'...?

[Eve]

You will still be able to model SysML 1.x in Enterprise Architect.

My knowledge of SysML 2 says that there will be a significant retraining cost to get people ready to use SysML 2. As a result, there will be definitely still be places where SysML 1 will remain in use for some time to come.

[PDC]

You will still be able to model SysML 1.x in Enterprise Architect.

My knowledge of SysML 2 says that there will be a significant retraining cost to get people ready to use SysML 2. As a result, there will be definitely still be places where SysML 1 will remain in use for some time to come.

Thanks. That was my assumption really, in my head they are almost 2 different languages so it wouldn't make sense for OMG to completely bin off 1.x. It will be interesting to see if they continue to let 1.x evolve (e.g. will there ever be a 1.8 if deficiencies are confirmed in 1.7, or would OMG simply point people to using 2.x).

[jonesphedra]

Hi Guillaume,

Just FYI, Tom mentioned SysML v2 in the session of the Global Summit.


In my opinion, SysML v2 seems similar to UML v2. That means it is an ideal and perfect language for some people, but I doubt if everyone wants it and/or everyone can use it. I think it is complex and we need to define more elements to describe a system with v2, compared to v1.x.mapquest

It's possible that SysML 1 will remain in use for some time, especially if there are large existing repositories of SysML 1 models that organizations continue to maintain.

[mwijns]

Any news on SysML v2 w.r.t. EA 17? I didn't get a chance to try out the beta yet but the release notes don't seem to mention anything related.

[Eve]

This forum is not a place where Sparx Systems will give any kind of product plans.

SysML 2 support is not included in EA 17. The specification itself isn't finalized, which is at least part of the reason why v17 doesn't include it.

[mwijns]

Thank you for your answer, not what I was hoping for but very clear.