Hello, > That's a shame :(, but at least we have a sample to fix it. > Who is working on the XmlSerializer again?
Currently no person is working continuously with it. Whoever noticed anything is commiting. I'm not kinda serializer person, so sorry to say I may miss the point. > Well first of all, why should [XmlInclude(typeof(S))] not work, since it > works on an abstract class that cannot be constructed as well. I'm a > compiler constructor myself, and would argue that an abstract class with > just abstract methods/properties and an interface should be very similar if > not equal. > At runtime the class of the property is known and it can be serialized as > such. OK, now I know I should separate two problems. (1)If a target field/property is specified as an interface or an abstract class, then that serialization must be impossible. But (2)even if the instance-typed field is instantiated as a concrete class, and if the serialization attribute for it specifies a concrete class, then it may be (or should be) possible to serialize. You are talking about (2). Then it seems better that when it is going to serialize an interface field/property, XmlSerializer only rejects if the target is not accessible, right? > Thanks a lot for that link. I already saw that, and hoped that the mono > implementation would be better. Sorry to say, these areas hadn't proceed so fast as you expected ;-) Oops, I must write up plans for XML... Thanks, -- Atsushi Eno _______________________________________________ Mono-list maillist - [EMAIL PROTECTED] http://lists.ximian.com/mailman/listinfo/mono-list
