Home » Modeling » TMF (Xtext) » Scoping in different contexts depending on non-resolved parameters
| Scoping in different contexts depending on non-resolved parameters [message #815183] |
Wed, 07 March 2012 10:25  |
 Alex G
Messages: 96
Registered: January 2012 |
Member |
|
|
Hi @all!
I want to be able to specify a graph like:
Vertex v
Vertex u[1]
Edge u[1] -> v
for(n in {2,3,4}) {
Vertex u[n]
Edge u[n] -> v
Edge u[n] -> u[n-1]
}
Edge u[2] -> u[1]
The edges have to reference to states, s.t. a state can have a qualified name like "v" or "u[1]", which I can write in XText like
IndexedID: id=ID (index=Indexing)?;
Indexing: '[' index = ArithExpr ']';
ArithExpr = ...
(where ArithExpr can be a an arbitrary arithmetic expression consisting of numbers or parameters, like 4+m/(2-x))
My questions are:
(1) Is it somehow possible to create several qualified names for one object, here:
"Edge u[n]" shall create three qualified names "u[2], u[3], u[4]", which can be referenced then like in "Edge u[2] -> u[1]"?
(2) The references shall be context sensitive, for example:
in "Edge u[2] -> u[1]" shall reference to the edges with qualified names u[2] and u[1]
but in "Edge u[n] -> v" the item "u[n]" shall be splitted (value converter?) in "u" and "n", where "u" references to the "u"-ID in "Vertex u[n]" and "n" references to the index "n" in "for(n in {2,3,4})".
A nicer way to solve this problem would be to generate this model into itself, like
Vertex v
Vertex u[1]
Edge u[1] -> v
Vertex u[2]
Edge u[2] -> v
Edge u[2] -> u[1]
Vertex u[3]
Edge u[3] -> v
Edge u[3] -> u[2]
...
Edge u[2] -> u[1]
and do then cross-reference validation in the generated model. But I think, I would need for generation the correct references for the indices described above.
Does anyone has an idea, how I can solve this?
|
|
|
| Re: Scoping in different contexts depending on non-resolved parameters [message #816463 is a reply to message #815183] |
Thu, 08 March 2012 22:08  |
Jan Koehnlein
Messages: 760
Registered: July 2009
Location: Hamburg |
Senior Member |
|
|
In your example u and u[x] are not the same objects, otherwise it
wouldn't matter if you write
u -> v
or
u[1] -> v
so adding multiple names won't help.
Xtext is meant for statically typed languages, so you cannot just
parameterize a name. Instead, you have to move the instantiation and
parameterization to your execution environment (interpreter or code
generator). From the perspective of Xtext, u[2] cannot be an element in
your example, as its definition requires evaluating an expression (the
for loop).
So your grammar should look like this
VertexDeclaration:
'Vertex' name=ID ('[' index=ArithExpr ']')?;
Edge:
'Edge' source=VertexRef '->' target=VertexRef;
VertexRef:
name=ID ('[' index=ArithExpr ']')?;
and the creation of real Vertices should take place in your execution
environment, which should also be implemented to resolve a VertexRef to
a Vertex instance after evaluating the index expression.
Am 07.03.12 11:25, schrieb Alex G:
> Hi @all!
>
> I want to be able to specify a graph like:
>
>
> Vertex v
> Vertex u[1]
> Edge u[1] -> v
>
> for(n in {2,3,4}) {
> Vertex u[n]
> Edge u[n] -> v
> Edge u[n] -> u[n-1]
> }
>
> Edge u[2] -> u[1]
>
>
> The edges have to reference to states, s.t. a state can have a qualified
> name like "v" or "u[1]", which I can write in XText like
>
>
> IndexedID: id=ID (index=Indexing)?;
> Indexing: '[' index = ArithExpr ']';
> ArithExpr = ...
>
> (where ArithExpr can be a an arbitrary arithmetic expression consisting
> of numbers or parameters, like 4+m/(2-x))
> My questions are:
> (1) Is it somehow possible to create several qualified names for one
> object, here:
> "Edge u[n]" shall create three qualified names "u[2], u[3], u[4]", which
> can be referenced then like in "Edge u[2] -> u[1]"?
>
> (2) The references shall be context sensitive, for example:
> in "Edge u[2] -> u[1]" shall reference to the edges with qualified names
> u[2] and u[1]
> but in "Edge u[n] -> v" the item "u[n]" shall be splitted (value
> converter?) in "u" and "n", where "u" references to the "u"-ID in
> "Vertex u[n]" and "n" references to the index "n" in "for(n in {2,3,4})".
>
> A nicer way to solve this problem would be to generate this model into
> itself, like
>
> Vertex v
> Vertex u[1]
> Edge u[1] -> v
>
> Vertex u[2]
> Edge u[2] -> v
> Edge u[2] -> u[1]
> Vertex u[3]
> Edge u[3] -> v
> Edge u[3] -> u[2]
> ..
>
> Edge u[2] -> u[1]
>
> and do then cross-reference validation in the generated model. But I
> think, I would need for generation the correct references for the
> indices described above.
>
> Does anyone has an idea, how I can solve this?
--
Need professional support for Eclipse Modeling?
Go visit: http://xtext.itemis.com
---
Get professional support from the Xtext committers at www.typefox.io
|
|
|
Goto Forum:
Current Time: Sat Apr 27 02:46:33 GMT 2024
Powered by FUDForum. Page generated in 0.03401 seconds |
] 
Search
Help
Register
Login
Home
