CanonicalEditPolicy and persistedChildren vs. transientChildren [message #548640] |
Thu, 22 July 2010 10:21  |
Eclipse User |
|
|
|
Dear GMF folks,
we encountered a strange behavior of a GMF editor of ours which we think
is a GMF bug:
1. Create an empty diagram
2. Make changes to the semantic model
-> the canonical edit policy correctly adds views for new elements
3. save
4. re-open the diagram
Then we get a
"IRJS0007E Semantic refresh failed."
Looking at the serialized notation model, we find that some of the
Shapes that should be there haven't been serialized and hence some of
the connections do not find their source and target Shapes.
Debugging the issue reveals that the canonical edit policy seems to
create all necessary views, BUT it puts some of them in the containment
tree as "persistedChildren" and the missing ones as "transientChildren",
which do not get serialized when saving.
Are issues like this known? Why can it happen that some views are put
into the tree as transient??
We treat all semantic objects the same. We created them within one
single model-to-model transformation using Xtend.
Any hints?
Cheers,
Hauke
|
|
|
|
|
Powered by
FUDForum. Page generated in 0.04442 seconds