Eclipse Community Forums: M2T (model-to-text transformation) » [Acceleo] In search of a clean solution for namespace/block generation

Home » Modeling » M2T (model-to-text transformation) » [Acceleo] In search of a clean solution for namespace/block generation

Show: Today's Messages :: Show Polls :: Message Navigator

[Acceleo] In search of a clean solution for namespace/block generation [message #632220]

Tue, 12 October 2010 07:34

Kirsten M. Z.

Messages: 132
Registered: July 2010

Senior Member

Hello,

I am searching for the best way for generating code like the following in Acceleo. Unfortunately, I haven't found a good solution/example:

namespace N1
{
        namespace N2
        {
                namespace N3
                {
                        ...specific-code...
                }
        }
}

Especially concerning reusability of the "namespace generation", a perfect indentation and clean code, I want to use a construct as follows. Unfortunately, something like this, does not exist in Acceleo (I think?)

[generateNamespaces(baseClass)]
        ...specific-code...
[generateNamespaces /]

Has anybody some advice? How to solve my use case in a good way?

Thanks!

Regards,
Kirsten

Report message to a moderator

Re: [Acceleo] In search of a clean solution for namespace/block generation [message #632229 is a reply to message #632220]

Tue, 12 October 2010 08:26

Laurent Goubet

Messages: 1876
Registered: July 2009

Senior Member

Hi Kirsten,

You won't be able to have a perfect indentation in both generated code and template, you'll have to make compromises Razz

.

The best way of handling this problem is if your namespaces and "baseclass" are of two different kind. For example if your namespaces are "EPackage"s and baseclass are "EClass"es, you could then use code like this :

[template public generate(element : EObject)]
  [comment this will not generate anything/]
[/template]

[template public generate(pack : EPackage)]
  namespace [name/]
  {
  [pack.eContents().generate()/]
  }
[/template]

[template public generate(class : EClass)]
  ...specific code...
[/template]

The three templates share the same name, so Acceleo will check the actual runtime type of the parameter in order to know which it should call. As you start your generation from the namespace (package "N1"), it will first generate the "namespace N1" part, then as it goes down the content encountering more packages or classes, generate the remainder. As soon as we find anything than a package or class, the placeholder "generate(EObject)" will be called and won't generate a thing.

There are more than a single way to do this, you could also iterate over the result of "baseClass.ancestors(EPackage)->reverse()" and generate all namespaces at once from this list ... You'll have to experiment to see which you prefer.

Regards,

Laurent Goubet
Obeo

Report message to a moderator

Re: [Acceleo] In search of a clean solution for namespace/block generation [message #632238 is a reply to message #632229]

Tue, 12 October 2010 08:54

Kirsten M. Z.

Messages: 132
Registered: July 2010

Senior Member

Hi Laurent,

thanks for your solution. It contains some interesting hints. However, the overall solution is not useful for me. The problem is the reusability of the "namespace generation" code. BTW, I need the searched construct not only for namespace generation, but it is a neat example everyone knows.

Imagine, you have to generate:

namespace N1
{
        namespace N2
        {
                namespace N3
                {
                        ... cheese ...
                }
        }
}
namespace N1
{
        namespace N2
        {
                namespace N3
                {
                        ... onions ...
                }
        }
}

In "my preferred solution", this would still look like:

[generateNamespaces(baseClass)]
        ... cheese ...
[generateNamespaces /]
[generateNamespaces(baseClass)]
        ... onions ...
[generateNamespaces /]

BTW: I know that indention in Acceleo code is... well... I gave up Smile

At least the resulting code must be indented perfectly, and this is, what I meant.

Report message to a moderator

Previous Topic:	[Xpand] EvaluationException, no main found in xpand template
Next Topic:	[XPAND] Workflow problem

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

]

Current Time: Tue Jan 21 11:00:09 GMT 2020

.:: Contact :: Home ::.

Breadcrumbs

Sign up to our Newsletter