Computed properties - a question and a bug

[sargasso]

Ver 4.50.746 (beta)

Question:  Some short time ago there was a discussion of whether properties should be treated differently to methods in class diagrams.  I have tried various searches but cant find it.  Does anyone remember what the topic was?

Bug(s): Let P be a class property whose value is computed - i.e. there is no associated private attribute.  Reason dictates that it is a readonly property.  When I RevEng it from C# EA creates the correct representation:

<<property>> int P with tag readonly=true and no associated attribute.

Now here's the twist - we have now decided to expose a setter for the property. (Why is complex - let's just say we want to persist a value passed in)

I have tried both setting the readonly tag to false and deleting the tag.  On forward engineering - back into the existing source - the setter method is not added.

Can anyone see why this would not be a bug?


tia
bruce

[Eve]

Because the setter method is inside the body of the operation which is never synchronised.

[sargasso]

hmmmmm.