serializing derived features : Quest for use-cases. [message #430366] |
Mon, 18 May 2009 14:33 |
Philipp Kutter Messages: 306 Registered: July 2009 |
Senior Member |
|
|
Hi.
I started a discussion with Ed, whether serializing derived features
with standard EMF by having a non-transient, volatile, derived feature
could make sense.
Here is the state:
>> Would you agree that derived, volatile, non-transient features are a
>> conceptually sound way to have derived features been serialized?
Ed:
It seems odd to have a derived thing be serialized.
>> The answer is important to me, as I want to continue these tests to more
>> advanced use cases. Agreement that this is a useful use case is thus
>> important to me ;-)
Ed:
It seems odd. Why serialize something that's derived. For example,
eAllSuperTypes is derived from eSuperTypes and it would be rather
useless (worse than useless, because it could lead to inconsistency) to
serialize it.
Thus Ed questions the usefulness of having serialized derived features
at all.
Is there anyone who has useful use-cases to serialize derived features?
Our use cases result from the following three scenarios:
- pre-existing XML formats
- XML that should be rendered (Through XSLT, Birt, e.t.c.) and where we
want to do the calculations inside the model.
- built in redundancy in the serialization form, to avoid non-valid
resources.
Regards,
Philipp
|
|
|
Powered by
FUDForum. Page generated in 0.02771 seconds