| just an idea [message #19607] |
Wed, 01 February 2006 06:19  |
Eclipse User |
|
|
|
Originally posted by: mgarcia912.yahoo.com
Why don't some of the EMFT OCL developers write a visitor for Octopus, so as
to translate its ASTs for the UML class model and for OCL expressions into
their input format? Or build in-memory instances of their metamodels and
then serialize. Then you will have more confidence when debugging OCL, for
you know it should work.
I mean, translating the subset of OctopusUML that EMF can handle. EMF cannot
do natively UML 1.4 association classes, for example. You translate then to
an additional class with two binary associations, and OCL the cardinality
constraints.
Just an idea, just an idea.
Miguel Garcia
|
|
|
| Re: just an idea [message #19697 is a reply to message #19607] |
Wed, 01 February 2006 12:07  |
| Eclipse User |
|
|
|
Originally posted by: info.dfm2html.com
>
> Why don't some of the EMFT OCL developers write a visitor for Octopus, so
> as to translate its ASTs for the UML class model and for OCL expressions
> into their input format? Or build in-memory instances of their metamodels
> and then serialize. Then you will have more confidence when debugging OCL,
> for you know it should work.
>
> I mean, translating the subset of OctopusUML that EMF can handle. EMF
> cannot do natively UML 1.4 association classes, for example. You translate
> then to an additional class with two binary associations, and OCL the
> cardinality constraints.
I thought that EMF is not meant to implement UML1.4, but the UML2 project
can do so.
By the way, does anyone know: EMFT OCL implements most of OCL2.0 spec.,
but why can this EMFT OCL be executed on a EMF model (for a model having
plain EMF as metamodel, and not UML2).
Is EMF formally considered as a subset of UML so that this is a formally
valid approach?
|
|
|
| Re: just an idea [message #565275 is a reply to message #19607] |
Wed, 01 February 2006 12:07 |
| Eclipse User |
|
|
|
Originally posted by: info.dfm2html.com
>
> Why don't some of the EMFT OCL developers write a visitor for Octopus, so
> as to translate its ASTs for the UML class model and for OCL expressions
> into their input format? Or build in-memory instances of their metamodels
> and then serialize. Then you will have more confidence when debugging OCL,
> for you know it should work.
>
> I mean, translating the subset of OctopusUML that EMF can handle. EMF
> cannot do natively UML 1.4 association classes, for example. You translate
> then to an additional class with two binary associations, and OCL the
> cardinality constraints.
I thought that EMF is not meant to implement UML1.4, but the UML2 project
can do so.
By the way, does anyone know: EMFT OCL implements most of OCL2.0 spec.,
but why can this EMFT OCL be executed on a EMF model (for a model having
plain EMF as metamodel, and not UML2).
Is EMF formally considered as a subset of UML so that this is a formally
valid approach?
|
|
|
Powered by
FUDForum. Page generated in 0.02638 seconds
Home