Skip to main content


Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Modeling » EMF » [Xcore] specifying a setter for a many-valued derived feature
[Xcore] specifying a setter for a many-valued derived feature [message #1751769] Fri, 13 January 2017 16:11 Go to next message
Stefan Winkler is currently offline Stefan WinklerFriend
Messages: 307
Registered: July 2009
Location: Germany
Senior Member
Hi Ed, hi all,

I have something like this in Xcore:

class Coordinate
{
  double x
  double y
}

class Shape
{
  contains Coordinate[] points
}


and I have the requirement to additionally have access to x and y coordinates as separate lists:

class Shape
{
  contains Coordinate[] points

  derived double[] xCoordinates
  get { 
    points.map[x]
  }

  // same for y ...
}


Until here, everything seems to be fine.
Now, what I need is a setter as well and this is where I struggle.

I've tried with something like this:
  derived double[] xCoordinates
  get { /* ... */ }
  set {
    for (i : (0..points.size)) {
	points.^get(i).x = (value as Iterable<Double>).^get(i)			
    }
  }


but this fails silently (no error marker; the setXCoordinates() method is just not generated in Java and the Xcore parser seems to work a lot because I get small UI freezes every few moments).

What would be the correct syntax? Or is it forbidden to have a setter to a many-valued derived feature?

Thanks in advance for any hints!

Best,
Stefan
Re: [Xcore] specifying a setter for a many-valued derived feature [message #1751791 is a reply to message #1751769] Sat, 14 January 2017 05:17 Go to previous message
Ed Merks is currently offline Ed MerksFriend
Messages: 33137
Registered: July 2009
Senior Member
Derived setters are really not well tested and given that a multi-valued features never have a generated setter, it doesn't seem so sensible. Also, even for the getter, there are parts of the framework that expect a multi-valued feature's return value to implement InternalEList, so even that part might case problems. You could define an operation that updates the points feature. Note that I think it's generally not a great idea to model a Point as a class. The overhead for an EObject seems excessive for two double values. And likely people programming against such an API will find it annoying to create a Coordinate set the two values, and they'll find it confusing that a Coordinate has a container and will be removed from one container when added to another...

Ed Merks
Professional Support: https://www.macromodeling.com/
Previous Topic:EMF Drag and Drop
Next Topic:Many-valued references in EMF
Goto Forum:
  


Current Time: Fri Apr 19 20:41:34 GMT 2024

Powered by FUDForum. Page generated in 0.02785 seconds
.:: Contact :: Home ::.

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

Back to the top