V15.2 - Is it possible to create a link to a text object WITHOUT automation?

[Paolo F Cantoni]

As you know, EA (erroneously in our view^[1]) treats Text (and other kinds of items) items as "owned by the diagram".  This means that when you copy a diagram using a shallow copy that says it creates links to the items in the browser, it will create copies, not links, to the "owned objects".  In the past, we used to be able to get around this limitation by copying and pasting from one diagram to another (since, although the items are in the browser, they aren't visible so drag and drop is unusable).

However, we've just noticed that (at least with v15.2) this is no longer possible.  It seems to always produce a CORRUPTED copy of the item (if you're going to produce a spurious copy of the item, at least produce a true copy!  - to be the subject of a defect report).

So, as the title asks, is there any way to create a link to a pre-existing text element without having to resort to the "black arts"?

TIA,
Paolo

BTW: The Use Case is that we need to put "notices" on diagrams that have had one or more relationships "suppressed" to so indicate.  Otherwise, viewers might get the possibly erroneous view that they are seeing the "full story".  We don't need 100,000 identical text objects added to the repository... 

^[1] I couldn't find any reference to this concept ("Diagram owned items") in a quick scan of UML 2.5 - but stand to be corrected.

[Eve]

In the past, we used to be able to get around this limitation

...

However, we've just noticed that (at least with v15.2) this is no longer possible.

...

So, as the title asks, is there any way to create a link to a pre-existing text element without having to resort to the "black arts"?

This is going to be a little circular, I know that. The limitation you describe is because EA has always treated those elements as owned by a single diagram. They have been treated as special and as a result they need to be treated special. Over the years, there were a few places added that didn't enforce this behavior. They were bugs in the sense that it created data that didn't match the requirements of other places.

In summary, the workaround that you were using was a bug. It was fixed in 15.2. If there still a way to do this, it's a bug.

Early in the beta period for 15.2 I believe you commented on this change in behavior. In follow up to that I said that we were looking at allowing the re-use of these items. Ultimately there was too much "specialness" to change that during the 15.2 beta, I still expect that you'll see it eventually. In the mean time 100,000 identical text objects are better than having the single element be deleted by a seemingly unrelated action and need to find all the 100,000 places that you need to put it back. (Or just train your modelers that every diagram is a partial view of the model)

^[1] I couldn't find any reference to this concept ("Diagram owned items") in a quick scan of UML 2.5 - but stand to be corrected.

If you're going to use that argument then you should probably not use any Text elements at all. You won't find them in UML 2.5 either.

[qwerty]

UML 2.5 also uses the term Comment rather than Note. I think not allowing free text on diagrams is a flaw in the UML specs. But I know that sending bug reports to OMG is as fruitful as doing so with Sparx :-/ Bad world.

q.

P.S. I fostered the specs and stumbled over p. 732

B.7.22 UMLLabel [Class]
The most general class for UML shapes that are rendered only as text.

So that might be an EA text element?

[Eve]

B.7.22 UMLLabel [Class]
The most general class for UML shapes that are rendered only as text.

So that might be an EA text element?

That's part of the Diagram Interchange Annex, that exists for the sole purpose of defining appearance and positioning of elements on a diagram. If that is a match for a Text element then it supports the way Enterprise Architect has handled them for years.

[qwerty]

That would be a happy ending then :-)

q.