| Re: [higgins-dev] IdAS notes |
|
I'm now going to check in the changes for #8. I'm still not sure why this was critical for 0.8. The implementation for it is like this:
public ISingleValuedAttribute getSingleValuedAttribute(URI attrID) throws IdASException, NotSingleValuedAttributeException {
IAttribute attr = getAttribute(attrID); if (attr.isSingleValued()) { return (ISingleValuedAttribute)attr; } else { throw new NotSingleValuedAttributeException(attrID + "is not a single-valued attribute"); } } This is in fact what I did in most CPs (I haven't updated the Jena CP).
Jim
>>> "Jim Sermersheim" <jimse@xxxxxxxxxx> 6/14/07 2:29 PM >>> Sounds good. I plan to put these fixes in the HEAD as well as the M0-8 branch.
I'm not sure what to do about downstream dependency projects. Adding these will cause them some of them to break. I can visit each project (that I know about) and at least make fixes to them such that they will build. I guess if I do that, I can just update their HEAD (and M0-8 for any that have that branch), and let others deal with branching as they choose.
Jim
>>> "Sergey Lyakhov" <slyakhov@xxxxxxxxxxxxxx> 6/14/07 2:13 PM >>> Jim,
> I'm glad you pointed these out because I would have guessed they weren't absolutely necessary. I think if user gets some unexpected Exception he should have a possibility to clear changes he made after last commit and avoid saving of wrong changes.
> On #8: I assume one would expect some exception to be thrown when the attribute specified by attrID is in fact *not* modeled as a single-valued attribute?
Yes, it should throw IdASException if this attribute is not an ISingleValuedAttribute instance. I think using this is better then to have ClassCastException or adding some additional code to check class of object. Thanks,
Sergey Lyakhov
|