Aggregation vs reference links

Claus Jørgensen wrote:

Im working on the same project as described in the post by Christian

Svarrer. We have been working on it in a couple of weeks and haven't figured

out what the difference between an aggregation and a refence link is yet. I

can see that the language acts differently according to the type of link.

What is the diffence and when should you use which link?

    We use trees to store MPS nodes. Every node can have two types of links

agregation and references. We use agregation when we want to make one

node to be child of another and references in another cases.

    For example we have expression a = b. The lefthand side and righthand

side nodes are children, i.e. aggregated by assignment node, and are

stored in the same model as its parent (assignment) and within their

parent. The lhs and rhs themselves refer to variable declarations via

reference link, and targets of this link can be stored in different models.

   (   StatementList   )

        --

     reference   |            |

  ----

    reference

Thank you for the reply. I have one more thing. When building a tree

structure using aggregation how do the cardinality work? If you have 1..n at

the source node, what would that look like?

"Konstantin Solomatov" <kostik@jetbrains.com> skrev i en meddelelse

news:dnjt9d$ghh$1@is.intellij.net...

Claus Jørgensen wrote:

>> Im working on the same project as described in the post by Christian

>> Svarrer. We have been working on it in a couple of weeks and haven't

>> figured out what the difference between an aggregation and a refence link

>> is yet. I can see that the language acts differently according to the

>> type of link. What is the diffence and when should you use which link?

We use trees to store MPS nodes. Every node can have two types of links

agregation and references. We use agregation when we want to make one node

to be child of another and references in another cases.

For example we have expression a = b. The lefthand side and righthand side

nodes are children, i.e. aggregated by assignment node, and are stored in

the same model as its parent (assignment) and within their parent. The lhs

and rhs themselves refer to variable declarations via reference link, and

targets of this link can be stored in different models.

>

>

  (   StatementList   )

   |         |      \agregations

  (decl)   (decl)   (  =  )

  /   \    /    \    |    \ agregations

int  a   int   b   (var) (var)

      /|\      /|\     |     |

       ------|----     |

    reference   |            |

------------

   reference

Claus Jørgensen wrote:

Thank you for the reply. I have one more thing. When building a tree

structure using aggregation how do the cardinality work? If you have 1..n at

the source node, what would that look like?

Cardinality means how many nodes you can have in a role. For example

cardinality in variable declaration like int a; type reference has

cardinality is 0..1 to 0..1 this means that there is only one or none in

role of type, i.e. either variable declaration has no type or hasn one

type.

For example in StatementList Cardinlity 0..n to 0..1 means that

StatementList can have many Statemenent for example variable declaration

and if statement.