| Problems regenerating classDiagram.gmfgen... [message #470343] |
Mon, 07 May 2007 16:46  |
Eclipse User |
|
|
|
Originally posted by: gabmecu.fiv.upv.es
Hello. I'm trying to regenerate the classDiagram.gmfgen from
classDiagra.gmfmap (I wan't to make some changes in the looking of the
figures) but I'm having problems since the validation of the map gives
errors.
More specifically, the error appears in the Association Class because of
the Source/Target Features. The value of these two attributes is
EReference type which seems to not be correct for the Class Association.
How can I overcome this problem? Both ends of an Association are
represented with properties that must be set with the ends of the
association. How is this accomplished? (through the GMF templates?)
Thanks in advance. Any help would ver appreciated.
Gabriel.
|
|
|
| Re: Problems regenerating classDiagram.gmfgen... [message #470346 is a reply to message #470343] |
Wed, 09 May 2007 15:28  |
 Michael Golubev
Messages: 383
Registered: July 2009 |
Senior Member |
|
|
Hello Gabriel,
Yes, there are some validation problems for ClassD.
It is basically because features like association links / classes (and also
some of custom labels, etc) can not be expressed in pure GMF.
Say, association link that visually connects 2 classes A and B may be semantically
represented as association with 2 child properties, that are _typed_ as B
and A.
This case is not supported by pure GMF generator, and I doubt it will be
supported in near future (this question was discussed in GMF news group a
lot of times).
So, in order to make it work, we have added some fake link definitions into
the ClassD GMF map file, that allow us to generate at least some parts of
the code correctly.
All custom staff was then rewritten via @generated NOT code.
Unfortunately, there are no notion for "@validating NOT" in the GMF map model,
so GMF generator still tries to validate these fake mappings.
Thus, warning here most probably mean that "GMF can not generate code properly
for this case".
That is OK, because all of the code that is generated with errors is rewritten
as @generated NOT.
So, you can try to just ignore these validation errors.
It is certainly true in case if you have changed only gmfgraph resources,
without introducing changes to GMF map model.
Just in case, the same issue affects ProfileDef and Component diagrams.
If it does not help, please consider publishing of your changes here, I will
take a look and try to resolve the problem.
Regards,
Michael
> Hello. I'm trying to regenerate the classDiagram.gmfgen from
> classDiagra.gmfmap (I wan't to make some changes in the looking of the
> figures) but I'm having problems since the validation of the map gives
> errors.
>
> More specifically, the error appears in the Association Class because
> of the Source/Target Features. The value of these two attributes is
> EReference type which seems to not be correct for the Class
> Association.
>
> How can I overcome this problem? Both ends of an Association are
> represented with properties that must be set with the ends of the
> association. How is this accomplished? (through the GMF templates?)
>
> Thanks in advance. Any help would ver appreciated. Gabriel.
>
|
|
|
| Re: Problems regenerating classDiagram.gmfgen... [message #592856 is a reply to message #470343] |
Wed, 09 May 2007 15:28 |
Michael Golubev
Messages: 383
Registered: July 2009 |
Senior Member |
|
|
Hello Gabriel,
Yes, there are some validation problems for ClassD.
It is basically because features like association links / classes (and also
some of custom labels, etc) can not be expressed in pure GMF.
Say, association link that visually connects 2 classes A and B may be semantically
represented as association with 2 child properties, that are _typed_ as B
and A.
This case is not supported by pure GMF generator, and I doubt it will be
supported in near future (this question was discussed in GMF news group a
lot of times).
So, in order to make it work, we have added some fake link definitions into
the ClassD GMF map file, that allow us to generate at least some parts of
the code correctly.
All custom staff was then rewritten via @generated NOT code.
Unfortunately, there are no notion for "@validating NOT" in the GMF map model,
so GMF generator still tries to validate these fake mappings.
Thus, warning here most probably mean that "GMF can not generate code properly
for this case".
That is OK, because all of the code that is generated with errors is rewritten
as @generated NOT.
So, you can try to just ignore these validation errors.
It is certainly true in case if you have changed only gmfgraph resources,
without introducing changes to GMF map model.
Just in case, the same issue affects ProfileDef and Component diagrams.
If it does not help, please consider publishing of your changes here, I will
take a look and try to resolve the problem.
Regards,
Michael
> Hello. I'm trying to regenerate the classDiagram.gmfgen from
> classDiagra.gmfmap (I wan't to make some changes in the looking of the
> figures) but I'm having problems since the validation of the map gives
> errors.
>
> More specifically, the error appears in the Association Class because
> of the Source/Target Features. The value of these two attributes is
> EReference type which seems to not be correct for the Class
> Association.
>
> How can I overcome this problem? Both ends of an Association are
> represented with properties that must be set with the ends of the
> association. How is this accomplished? (through the GMF templates?)
>
> Thanks in advance. Any help would ver appreciated. Gabriel.
>
|
|
|
Powered by
FUDForum. Page generated in 0.02774 seconds
] 
Search
Help
Register
Login
Home
