The real fun might begin when this "vector" is not a real field either: it can be just ObjectStreamField declaration de/serialized manually into any other private representation. So don't expect this test to be implementation-independent.
BTW, what is the point to care about serialization test performance? 2006/11/27, spark shen <[EMAIL PROTECTED]>:
Nathan Beyer 写道: > On 11/26/06, Andrew Zhang <[EMAIL PROTECTED]> wrote: >> On 11/27/06, Spark Shen <[EMAIL PROTECTED]> wrote: >> > >> > Reflection may be feasible, but a downside hit on performance. >> > >> > I have a suggestion: writing a subclass, open that field using a >> getter, >> >> >> How can we get the field if it's private? > > Write an accessor that looks up the declared field, set it as > accessible and then read it. This requires that the security policy > allows it. Good idea. And IMHO, this may not over-perform reflection. :-) Best regards > > -Nathan > >> >> and >> > then customize serialization process for this subclass to deliberate >> > invoke >> > this getter. Is this possible? >> > >> > >> > 2006/11/26, Andrew Zhang <[EMAIL PROTECTED]>: >> > > >> > > Hi folks, >> > > >> > > I encountered a problem when writing serialization test for >> > > SerialJavaObject >> > > class. There're two fields are required by serialized form: >> Object obj, >> > > and >> > > Vector chain. But there's no public method is available to get chain >> > > variable. The question comes: >> > > How can I verify deserilized object, by reflection? How do we handle >> > such >> > > problem in other modules? Any suggestions? >> > > >> > > Thanks in adavance! >> > > >> > > -- >> > > Best regards, >> > > Andrew Zhang >> > > >> > > >> > >> > >> > -- >> > Spark Shen >> > China Software Development Lab, IBM >> > >> > >> >> >> -- >> Best regards, >> Andrew Zhang >> >> >
