| Large Model, minor changes [message #1007067] |
Mon, 04 February 2013 05:19  |
Eclipse User |
|
|
|
Hi,
I haven't done very much in M2M, but before I start, I want to have your opinion:
We have a very big Ecore meta-model (~500 classes), lets call it "A".
Now, we have another Ecore meta-mdoel ("B"), which is almost similiar to "A" but does have about 20 deviations.
What I now want to do is to transform "B" -> "A" and "A" -> "B".
Can you tell me what the best approach would be and how complex this would be?
Thank you very much
Matthias
|
|
|
| Re: Large Model, minor changes [message #1007075 is a reply to message #1007067] |
Mon, 04 February 2013 05:49  |
| Eclipse User |
|
|
|
Hi
500 classes is not 'very big'. The corresponding model with 500,000
objects might be.
Ideally you want a bidirectional transformation language such as QVTr,
however the main Eclipse implementations only support editing today. You
could try the Medini or ModelMorf implementations.
Alternatively you can implement two unidirectional transformations, in
which case your choices are at least
- QVTo (most standard)
- ATL (most mature)
- Xtend2 (most Java-like)
- Epsilon (most diverse)
- Viatra2 (most unknown to me)
With only 20 deviations, handcrafted Java and mapping tools may also be
options
Regards
Ed Willink
On 04/02/2013 10:19, Matthias N wrote:
> Hi,
> I haven't done very much in M2M, but before I start, I want to have
> your opinion:
>
>
> We have a very big Ecore meta-model (~500 classes), lets call it "A".
>
> Now, we have another Ecore meta-mdoel ("B"), which is almost similiar
> to "A" but does have about 20 deviations.
>
>
> What I now want to do is to transform "B" -> "A" and "A" -> "B".
> Can you tell me what the best approach would be and how complex this
> would be?
>
> Thank you very much
> Matthias
>
|
|
|
|
|
| Re: Large Model, minor changes [message #1007438 is a reply to message #1007095] |
Tue, 05 February 2013 15:56 |
| Eclipse User |
|
|
|
Op 04-02-13 13:42, Ed Willink schreef:
> Hi
>
> I don't really understand the question.
>
> Some transformations languages have a default copy behaviour so that you only
> need to worry about differences.
>
> Regards
>
> Ed Willink
>
>
> On 04/02/2013 11:16, Matthias N wrote:
>> Hi Ed,
>> thank you very much for your answer.
>>
>> Is it possible to only define the deviations in the M2M transformatino
>> language, or do I really have to specify every transformation rule of the
>> entire meta-model by hand?
>
For the ATL solution, see this thread:
http://www.eclipse.org/forums/index.php/m/535170/
Regards,
Dennis
|
|
|
Powered by
FUDForum. Page generated in 0.04302 seconds
Home