Ordered edges by index

[eclipse-viatra-bot]

| --- | --- |
| Bugzilla Link | 398737 |
| Status | NEW |
| Importance | P3 normal |
| Reported | Jan 22, 2013 08:42 EDT |
| Modified | Apr 19, 2016 05:25 EDT |
| Version | oldinquery |
| Reporter | Istvan Rath |

Description

Cloned from: 99: Ordered edges by index
https://github.com/ujhelyiz/EMF-IncQuery/issues/issue/99

Support ordered edge index path expressions, such as:

pattern myPattern() 
{
  Type.refType[X](Src,Trg);
}

where X would be the index variable (that can be replaced by a constant if necessary.)

[eclipse-viatra-bot]

By Istvan Rath on May 20, 2013 04:35

I'm setting version to "unspecified" as we currently do not (firmly) know when these future ideas will be addressed.

[eclipse-viatra-bot]

By Zoltan Ujhelyi on May 31, 2013 04:19

Henshin has now something similar, see corresponding blog post: http://www.ckrause.org/2013/05/henshin-098-working-with-lists.html

[eclipse-viatra-bot]

By Abel Hegedus on Apr 19, 2016 05:25

Updating to correct milestone.

[bergmanngabor]

Probably the simplest solution would be to introduce a constraint in the language and a new inputkey for a binary relation (element, valueList) identified by a structural feature, where each element is associated with the list of values that eGet() would return. IT is easy for the EMFScope to maintain such relations (if indexed, which should happen on demand), and it is possible to process such lists further in VQL: eval() could call e.g. head, while eval unwind could call a helper method to transform the list to a set of index-value pairs (similar to IterableExtensions.forEach).

Path.steps[](path, stepList);
lastStep == eval(stepList.tail);

Open question: semantics of this relation for elements where the result list is empty.

[bergmanngabor]

The original syntax suggestion is probably better:

• more straightforward type inference in the language (no generics, no mixing between EObject-valued and scalar variables)
• fewer VQL lines / simpler syntax in most use cases, with the exception of tail
• there is no difficult semantic question of whether empty lists are to be matched