Archimate 3: Cannot Create an assignment between a device and a system software

[matthew.james]

I remember a report about that particular issue and adding it in to our metamodel now.

Simon - given that this (i.e. Assignment between Device and System Software) is working in 1420, but not working in 1427 ...
- Is the XML change you've provided a post 1427 change? (I assume yes)
- Was this a regression or bug introduced post 1420, or is it the result of a re-engineering of the Archimate relationships in Sparx? (I'm guessing the latter given that our Archimate3 technology file doesn't have the change listed but it still works)
- To your earlier point about the ambiguity of the spec, is there a way Sparx can publish a human readable set of valid Archimate relationships as implemented in the Sparx MDG (something similar to the aforementioned appendix in the Archimate spec)?

The challenge we are facing is what to do when Sparx disallows a relationship we are trying to use (with strict connector syntax on) ... is the relationship actually invalid or is it an error in the MDG implementation

[Eve]

The relationships in the ArchiMate3 specification where completely reworked. Instead of defining a quicklinker table with one row for every letter in the relationship tables (manually) , we have a metamodel that defines available relationships as defined by the metamodel figures in Chapters 3 through 13. (And the corresponding textual descriptions)

[matthew.james]

Ok - cool. And that change happened somewhere between 1420 and 1427 ? (I can't find it mentioned in the release notes, although there are a few mentions of validation changes which may reflect the outcome of that change)

Is that meta-model shareable ... ?
I'm thinking as published diagrams and / or Sparx project (?)

[Eve]

The ArchiMate technology was converted to use a metamodel during the 14.0 beta period. (1414 to be precise)

Are you sure you haven't done something like replace the MDG technology with an older version?

We don't generally publish the source of our MDG technologies, even the ones that are in plain text in the installer.

[matthew.james]

The ArchiMate technology was converted to use a metamodel during the 14.0 beta period. (1414 to be precise)
Are you sure you haven't done something like replace the MDG technology with an older version?

Nope - definitely running 1420 and definitely standard install (our machines are locked down so we can't make any changes, the Sparx installation package was run by our sysadmin). Strict Connector Syntax is on and I can definitely create Assignment relationships between Device and System Software.

So if that doesn't work in 1427 (which I can't verify) then something changed in between

[Eve]

Just tested 1420.

The only quicklinks that I get between Device and System Software are the base UML links. (1421 added the capability to suppress these)
Looks like it's not running any validation at all, which would be the big difference.

In my opinion, 1427 would be more useful for modeling ArchiMate. Even if you have to disable the strict connector syntax.

[Uli Schulze-Eyssing]

@ Geert: Many thanks! 

Where can I get a download link to 1420?
Many thanks,
Uli

[matthew.james]

Just tested 1420. The only quicklinks that I get between Device and System Software are the base UML links.

Ok - quicklinks is different.  In 1420 I also only get the UML ones there (Dependency, Trace, Information Flow).  I'm talking about adding relationships from the toolbox; I tend not to use quicklinks as (from experience) the right relationships just aren't there - probably a reason to upgrade to 1427 then(?)

Looks like it's not running any validation at all, which would be the big difference.
In my opinion, 1427 would be more useful for modeling ArchiMate. Even if you have to disable the strict connector syntax.

There is definitely some validation running as there are some relationships that get rejected when Strict Connector Syntax is on, but are allowed when it is off, eg Device > Service > System Software