So is it an option to let IsSerializable extend the Serializable marker interface?
On 21 Nov., 16:16, John LaBanca <[EMAIL PROTECTED]> wrote: > That's a good question, and I has to ask around myself. Here is the best > explanation: > > "In early versions of GWT, IsSerializable was the *only* way you could mark > a class as serializable by the RPC subsystem. The theory was that Java's > Serializable interface implied semantics that GWT simply couldn't implement > (readObject(), writeObject(), etc), so it was better to be absolutely clear > that it wasn't precisely the same as Java Serializable. We were eventually > convinced otherwise by many people who had existing POJOs that implemented > Serializable and *didn't* require these specialized semantics, and really > wanted it to work out of the box. So we added support for both." > > So based on this, IsSerializable may be deprecated at some point in favor of > Serializable. In the meantime, we run into these unfortunate cases where > you want a class to extent either IsSerializable or Serializable, but there > is no way to specify that in Java. > > Thanks, > John LaBanca > [EMAIL PROTECTED] > > On Thu, Nov 20, 2008 at 6:34 AM, dflorey <[EMAIL PROTECTED]> wrote: > > > I don't know if it's the only place where this question comes up, but > > right now the SerializableResponse from the table model requires its > > wrapped row values to implement the IsSerializable interface. > > So every row value object that implement Serializable (and as such can > > be serialized) cannot be used in SerializableResponse as long as it > > will not extend the IsSerializable interface. > > When using Serializable instead the same problem will arise with > > classes implementing IsSerializable. > > Any ideas? --~--~---------~--~----~------------~-------~--~----~ http://groups.google.com/group/Google-Web-Toolkit-Contributors -~----------~----~----~----~------~----~------~--~---
