Skip to main content

Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Modeling » OCL » choosing "Property::CompatibleDefaultExpression" for programmatically defined pivot types
choosing "Property::CompatibleDefaultExpression" for programmatically defined pivot types [message #1748493] Wed, 23 November 2016 15:27 Go to next message
Julien Deantoni is currently offline Julien DeantoniFriend
Messages: 45
Registered: July 2009

I'm extending the CompleteOCL grammar with some concepts amongst which a new "Event" type.
I implmented the visitors so that the parsing is almost ok. The problem I have comes from the Continuation of the post order of this Event Type. I have to install the pivot and I wanted to map it to a new OCLAny type (to be compliant with any type of body specification in a PropertyDefinition).
If I use the OCLAny type, it is OK:
AnyType anyType = context.getEnvironmentFactory().getStandardLibrary().getOclAnyType();

However, if I try to create a new OCLAny type if filas (even if I tried to copy the properties of the anyType):
AnyType anyType = context.getEnvironmentFactory().getStandardLibrary().getOclAnyType();
AnyType eventType = new PivotFactoryImpl().createAnyType();

I this last case I have a warning in the editor, for instance when trying to define a def of Event type:
The 'Property::CompatibleDefaultExpression' constraint is violated for xxxx 

any idea on how to define the default compatibility ?

Re: choosing "Property::CompatibleDefaultExpression" for programmatically defined pivot ty [message #1748502 is a reply to message #1748493] Wed, 23 November 2016 16:38 Go to previous message
Ed Willink is currently offline Ed WillinkFriend
Messages: 7663
Registered: July 2009
Senior Member
Hi Julien

Re-using one type e.g AnyType as a new semantic type seems to be very pragmatic; it may save you time today, but it will make life very difficult in the future. Even today you ask me why your code doesn't work. I cannot help without a fully executable repro.

The general Pivot OCL philosophy is that each semantically distinct type has a distinct metamodel class. (On GIT/master this has driven me to introducing derived LetVariable/IteratorVariable/ParameterVariable/ResultVariable to avoid the WFRs of Variable having to discover their context before using a case switch.)

I recommend that you emulate one of the QVTd extensions for which many lack of extensibility bugs have been fixed in OCL. QVTimperative is most independent.

[I find the Complete OCL syntax very unsatisfactory - insufficient keywords, no {}'s or ;'s to help error recovery..]


Ed Willink
Previous Topic:Performing queries in KDM with OCL
Next Topic:pre and post conditions in OCL metamodel
Goto Forum:

Current Time: Sun Jul 14 03:09:54 GMT 2024

Powered by FUDForum. Page generated in 0.03268 seconds
.:: Contact :: Home ::.

Powered by: FUDforum 3.0.2.
Copyright ©2001-2010 FUDforum Bulletin Board Software

Back to the top