Skip to main content



      Home
Home » Archived » M2M (model-to-model transformation) » [ATL] model checking with ATL
[ATL] model checking with ATL [message #30528] Tue, 10 April 2007 06:21 Go to next message
Eclipse UserFriend
Hello,

I have a UML profile which is applied to a UML class diagram. This profile
has several invariant constraints which are applied to the stereotyped
elements. I would like to use ATL to verify the validity of the stereotyped
class diagram with respect to my profile.



I have seen in the paper "Using ATL for checking Models"
( http://www.sciences.univ-nantes.fr/lina/atl/www/papers/GraMo T05.pdf) that
the authors implement their strategy in an Eclipse plugin using EMF.

I am thinking in creating the ATL transformation which define the stereotype
constraints as is explained in the paper and applying that transformation to
my stereotyped class diagram. But I don't know if it is the right solution?

Thanks in advance
Bea
Re: [ATL] model checking with ATL [message #30740 is a reply to message #30528] Wed, 11 April 2007 14:21 Go to previous messageGo to next message
Eclipse UserFriend
Originally posted by: carrasco.ModelDrivenDevelopment.co.uk

Well, I
Re: [ATL] model checking with ATL [message #30774 is a reply to message #30740] Wed, 11 April 2007 14:31 Go to previous messageGo to next message
Eclipse UserFriend
Hello,

I would argue that representing a diagnostic (i.e., the result of
checking some constraints over a model) as a model is actually simpler
to integrate in a Model Engineering scenario (sequence of
transformations). This is especially the case with AMMA, which is guided
by the principle "everything is a model".


For instance, the TCS compiler (i.e., from TCS+KM3 to an ANTLR grammar)
uses this approach.
Before running the TCS2ANTLR.atl transformation, the TCS2Problem.atl
transformation is executed to check constraints (it is implemented
following the scheme presented in the "Using ATL for checking Models"
paper).
Because the source model (i.e., the .tcs file) uses TCS, the problems
can be reported within the editor using the am3.applyMarkers ant task.
See:
TCSSample.zip\TCSSample\TCSToolkit\TCS2JAR\build.xml
(zip available from http://wiki.eclipse.org/index.php/TCS).

The KM3 compiler (i.e., KM3 to Ecore or KM3 to MOF 1.4) also uses this
approach, but was implemented before the AM3 ant tasks had been created.
Therefore, it uses Java code to launch the KM32Problem.atl
transformation. However, this shows that you can also integrate this
scheme within Java code, should you prefer this over ant tasks for some
reason.


Best regards,

Frédéric Jouault


Antonio Carrasco wrote:
> Well, I´ll give this one a try, too:
>
> The approach from Jean Bezivin and Frederic Jouault is indeed sound,
> and specially interesting as the result is a verification model
> (an instance of a verification metamodel),
> that can be subject to storage, structured inspection,
> reased upon, and further processed.
> This is consistent with this Team´s work
> around "model buses", "modelling at large", and high complexity.
>
> A simpler approach, if you are only interested in knowing about the validity
> of the model,
> is to include the constraints as OCL, and request Eclipse to validate the
> model.
>
>
> Now, there´s an implementation of OCL that applies to model implementations
> you generate yourself with EMF, as described in
> http://www.eclipse.org/articles/article.php?file=Article-EMF -Codegen-with-OCL/index.html
>
> There is other validation framework, EMFT validation,
> I am less familiar with. May be some other in the newsgroup may be of help.
>
> Unfortunately, if your are working with a profile of UML,
> I do not know how to evaluate validation constraints
> defined for a profile of UML2 implementation in Eclipse.
> Again, somebody else may provide an answer.
> Cheers,
> ACV
>
>
>
>
>
>
>
>
> "Bea" <beaperez@unizar.es> wrote in message
> news:evfogo$bn2$1@build.eclipse.org...
>> Hello,
>>
>> I have a UML profile which is applied to a UML class diagram. This profile
>> has several invariant constraints which are applied to the stereotyped
>> elements. I would like to use ATL to verify the validity of the
>> stereotyped class diagram with respect to my profile.
>>
>>
>>
>> I have seen in the paper "Using ATL for checking Models"
>> ( http://www.sciences.univ-nantes.fr/lina/atl/www/papers/GraMo T05.pdf) that
>> the authors implement their strategy in an Eclipse plugin using EMF.
>>
>> I am thinking in creating the ATL transformation which define the
>> stereotype constraints as is explained in the paper and applying that
>> transformation to my stereotyped class diagram. But I don't know if it is
>> the right solution?
>>
>> Thanks in advance
>> Bea
>>
>
>
Re: [ATL] model checking with ATL [message #30879 is a reply to message #30774] Thu, 12 April 2007 06:19 Go to previous messageGo to next message
Eclipse UserFriend
Hello,

If it is possible I would like to use ATL for checking my stereotype
constraints in the class diagram. I think I am a bit lost, Could you correct
me if I am wrong?

I could create an ATL transformation for my constraints (as it is explained
in the paper) and use it as I use my other ATL transformation. The problem
of using this way is that I don't get the advantages expounded in the paper
(diagnostic as a model, the problems are not reported within the editor,
etc).



In order to use the approach of the paper (and get the diagnostic as a
model, etc) , what I would have to do? Have I to define a --2Problem.atl
transformation?

Could you help me?



Thanks






"Fr
Re: [ATL] model checking with ATL [message #31015 is a reply to message #30879] Thu, 12 April 2007 13:47 Go to previous messageGo to next message
Eclipse UserFriend
Hello,

If you follow the scheme presented in the paper (i.e., a --2Problem.atl
transformation), you will get the diagnostic as a model.

You may then do whatever you want with this model. For instance, you may
use TCS to get a textual representation of the diagnostic.


Regards,

Frédéric Jouault


Bea wrote:
> Hello,
>
> If it is possible I would like to use ATL for checking my stereotype
> constraints in the class diagram. I think I am a bit lost, Could you correct
> me if I am wrong?
>
> I could create an ATL transformation for my constraints (as it is explained
> in the paper) and use it as I use my other ATL transformation. The problem
> of using this way is that I don't get the advantages expounded in the paper
> (diagnostic as a model, the problems are not reported within the editor,
> etc).
>
>
>
> In order to use the approach of the paper (and get the diagnostic as a
> model, etc) , what I would have to do? Have I to define a --2Problem.atl
> transformation?
>
> Could you help me?
>
>
>
> Thanks
>
>
>
>
>
>
> "Frédéric Jouault" <frederic.jouault@univ-nantes.fr> escribió en el mensaje
> news:evj9ji$c87$1@build.eclipse.org...
>> Hello,
>>
>> I would argue that representing a diagnostic (i.e., the result of checking
>> some constraints over a model) as a model is actually simpler to integrate
>> in a Model Engineering scenario (sequence of transformations). This is
>> especially the case with AMMA, which is guided by the principle
>> "everything is a model".
>>
>>
>> For instance, the TCS compiler (i.e., from TCS+KM3 to an ANTLR grammar)
>> uses this approach.
>> Before running the TCS2ANTLR.atl transformation, the TCS2Problem.atl
>> transformation is executed to check constraints (it is implemented
>> following the scheme presented in the "Using ATL for checking Models"
>> paper).
>> Because the source model (i.e., the .tcs file) uses TCS, the problems can
>> be reported within the editor using the am3.applyMarkers ant task.
>> See:
>> TCSSample.zip\TCSSample\TCSToolkit\TCS2JAR\build.xml
>> (zip available from http://wiki.eclipse.org/index.php/TCS).
>>
>> The KM3 compiler (i.e., KM3 to Ecore or KM3 to MOF 1.4) also uses this
>> approach, but was implemented before the AM3 ant tasks had been created.
>> Therefore, it uses Java code to launch the KM32Problem.atl transformation.
>> However, this shows that you can also integrate this scheme within Java
>> code, should you prefer this over ant tasks for some reason.
>>
>>
>> Best regards,
>>
>> Frédéric Jouault
>>
>>
>> Antonio Carrasco wrote:
>>> Well, I´ll give this one a try, too:
>>>
>>> The approach from Jean Bezivin and Frederic Jouault is indeed sound,
>>> and specially interesting as the result is a verification model
>>> (an instance of a verification metamodel),
>>> that can be subject to storage, structured inspection,
>>> reased upon, and further processed.
>>> This is consistent with this Team´s work
>>> around "model buses", "modelling at large", and high complexity.
>>>
>>> A simpler approach, if you are only interested in knowing about the
>>> validity of the model,
>>> is to include the constraints as OCL, and request Eclipse to validate the
>>> model.
>>>
>>>
>>> Now, there´s an implementation of OCL that applies to model
>>> implementations
>>> you generate yourself with EMF, as described in
>>> http://www.eclipse.org/articles/article.php?file=Article-EMF -Codegen-with-OCL/index.html
>>>
>>> There is other validation framework, EMFT validation,
>>> I am less familiar with. May be some other in the newsgroup may be of
>>> help.
>>>
>>> Unfortunately, if your are working with a profile of UML,
>>> I do not know how to evaluate validation constraints
>>> defined for a profile of UML2 implementation in Eclipse.
>>> Again, somebody else may provide an answer.
>>> Cheers,
>>> ACV
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> "Bea" <beaperez@unizar.es> wrote in message
>>> news:evfogo$bn2$1@build.eclipse.org...
>>>> Hello,
>>>>
>>>> I have a UML profile which is applied to a UML class diagram. This
>>>> profile has several invariant constraints which are applied to the
>>>> stereotyped elements. I would like to use ATL to verify the validity of
>>>> the stereotyped class diagram with respect to my profile.
>>>>
>>>>
>>>>
>>>> I have seen in the paper "Using ATL for checking Models"
>>>> ( http://www.sciences.univ-nantes.fr/lina/atl/www/papers/GraMo T05.pdf)
>>>> that the authors implement their strategy in an Eclipse plugin using
>>>> EMF.
>>>>
>>>> I am thinking in creating the ATL transformation which define the
>>>> stereotype constraints as is explained in the paper and applying that
>>>> transformation to my stereotyped class diagram. But I don't know if it
>>>> is the right solution?
>>>>
>>>> Thanks in advance
>>>> Bea
>>>>
>
Re: [ATL] model checking with ATL [message #31050 is a reply to message #31015] Thu, 12 April 2007 14:06 Go to previous message
Eclipse UserFriend
That is true! I had my head in the clouds :)
Thanks


"Fr
Previous Topic:[ATL] 'null' values not explicitly set?
Next Topic:[ATL] sequencing references
Goto Forum:
  


Current Time: Thu May 01 13:57:29 EDT 2025

Powered by FUDForum. Page generated in 0.03958 seconds
.:: Contact :: Home ::.

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

Back to the top