|
|
Re: [ATL] Re: Refining Execution Mode [message #510956 is a reply to message #501193] |
Fri, 29 January 2010 10:04 |
Julio Messages: 5 Registered: November 2009 |
Junior Member |
|
|
This example is far too simple.
I tried "sorting/reordering" model elements (inside a "root" element) as a result of refining, and this transformation worked for ATL2*, whereas now (ATL3*), it does not change anything (no matter if I use Regular VM or EMF-specific). The result of the transformation is the same as the input.
Furthermore, the new ATL disallows the use of lazy rules in a refining transformation whereas the previous version was permissive.
Is there a source to know what is not considered by the new ATL, compared with the ATL previous versions, in the case of refining transformations?
J.
[Updated on: Fri, 29 January 2010 10:06] Report message to a moderator
|
|
|
|
|
Re: Refining Execution Mode [message #511650 is a reply to message #511629] |
Tue, 02 February 2010 12:05 |
Vincent MAHE Messages: 129 Registered: July 2009 |
Senior Member |
|
|
Julio a écrit :
> I do not understand the difference implied by just saying "in place"
> transformation, by contrast with a "copy". Is there some pointer
> manipulation assumed (in ATL VM)? That it is faster I cannot asses
> either, perhaps you know of a benchmark suite, or have a complexity
> analysis at hand that you would like to share with us ;)
> In any case, a copy is linear, sorting/rearranging-elements is another
> business!
Yes but as in EMF the order is the one the model have been created with,
you cannot expect a refining mode (which only change element to be
modified) to do your ordering work. You have taken benefit of a bad
respect of refining semantic in ATL2, when ATL3 is more effective on the
refining semantic.
>
> I do not understand either the remark about reproducing in ATL3 the ATL2
> refining operation. Do you mean to program the refinement of ATL2 as a
> standard transformation?
> My problem at hand is "simple" (I want to believe): Given a model sort
> the occurrence of its elements according to a given comparison relation.
> Clearly models are big trees, and I do mean to "reorder" arbitrary
> subtrees, not only those at the same level.
> I do not mean to add/change-values-of model elements, but their "order"
> of ocurrence.
>
> As far as I can see
> the question remains: What does refining mean in the ATL3 context?
> (possibly with respect to refining in ATL2)
>
> J.
--
Cordialement
Vincent MAHÉ
Ingénieur plate-forme - Cesar/Artemisia - Équipe Espresso
IRISA-INRIA, Campus de Beaulieu, 35042 Rennes cedex, France
Tél: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71
|
|
|
Powered by
FUDForum. Page generated in 0.03359 seconds