Labels on data object hidden

[skiwi]

Prompt work from sparx support

Here's the problem: the shape script lookup for the main shape searches for stereotype without worrying what the extended metaclass is, but the lookup for the label shape searches for the stereotype and extension. It shouldn't make any difference, until somebody applies stereotypes to the wrong metaclass.

The fix is to make the main shape search stricter, which won't help you and will in fact remove the shape script from this one object - you will have to replace your stereotyped class with a stereotyped artifact. The fix is also likely to break a number of diagrams that are currently "wrong, but look how you want them to look".

[Paolo F Cantoni]

Did Sparx say they were going to make the fix?

There's a pile of bugs where the fix is obvious, but haven't been implemented for years...

Nevertheless, an attempt to make EA more consistent!

You can't  beat:
Consistency, Consistency, Consistency! ^TM

Paolo

[skiwi]

The fix is also likely to break a number of diagrams that are currently "wrong, but look how you want them to look".

The issue has been passed on for a decision on whether to fix or not. If fixed it will be applied to a future release of EA.

The community might like to discuss this.
How does it respond to a fix that breaks existing diagrams.

[Paolo F Cantoni]

The community might like to discuss this.
How does it respond to a fix that breaks existing diagrams.

Yes!

But ONLY if Sparx takes note of the community view, otherwise we're just...

In my view, it deserves it's own topic.

I'm having enough trouble trying to make models that are RIGHT look right - never mind getting those that are WRONG...  See: Consistent Diagram Item rendering proposal

The whole point of a modelling program IS to produce valid models.  Otherwise you have just a drawing program of which there are many, MUCH better...

If your diagrams can ONLY look right by having an incorrect model, you got bigger problems...

Besides, in systems, "Two wrongs NEVER make a right".

My, initial, AU$0.05
Paolo

[KP]

The community might like to discuss this.
How does it respond to a fix that breaks existing diagrams.

Actually the message you were supposed to get from support was that your model is wrong and to fix it you should change your stereotyped class to a stereotyped artifact. We won't be breaking anyone's diagrams.

[Paolo F Cantoni]

Actually the message you were supposed to get from support was that your model is wrong and to fix it you should change your stereotyped class to a stereotyped artifact. We won't be breaking anyone's diagrams.

Thanks for that Neil,

As I implied above, I hadn't actually seen a wrong model produce a good output...

So - can you confirm exactly what the fix will be?  As I understand the discussion so far, there are implications for those of us creating MDGTs.

Paolo

[skiwi]

Actually the message you were supposed to get from support was that your model is wrong and to fix it you should change your stereotyped class to a stereotyped artifact. We won't be breaking anyone's diagrams.

Here are the messages I got, verbatim, the highlighting is mine

I'm assuming that you're referring to the BPMN DataObject element. The BPMN profile extends the "Artifact" metaclass with the <DataObject> stereotype. It does not extend the "Class" metaclass. If you try to apply the stereotype to the wrong metaclass then the results will be undefined.

Anyway, having said all that, I have tried applying the shape script associated with the DataObject stereotype to a variety of different metaclasses (including Artifact and Class). They all show their name in a label which can be hidden, and they all have a "Show Labels" command which brings the hidden label back. So I don't think the fact that one is a class and one is an artifact is the cause of the problem.

Here's the problem: the shape script lookup for the main shape searches for stereotype without worrying what the extended metaclass is, but the lookup for the label shape searches for the stereotype and extension. It shouldn't make any difference, until somebody applies stereotypes to the wrong metaclass.

The fix is to make the main shape search stricter, which won't help you and will in fact remove the shape script from this one object - you will have to replace your stereotyped class with a stereotyped artifact. The fix is also likely to break a number of diagrams that are currently "wrong, but look how you want them to look".

I have passed the issue on for a decision on whether we fix or not. If we do it will be applied to a future release of EA.

In fact no suggestion to change my diagram was made by support,
not surprising since I had already diagnosed the issue.

The point of this discussion, and bug report was that, when I first raised this problem, a problem that had myself and my colleagues stumped, and was not obvious at all to others who helped in this forum,
was the anomalous and unexpected behaviour of an object in a diagram, with no indication from the tool as to what might be causing the problem, ie the name/label not to show.
Neither was there any indication in the Help documentation that there were cases where the "Show Label" context menu item might not be available.

Its clear to me that this is a bug, and it needs fixing.
thanks
Simon