org.jboss.ejb.plugins.cmp.jdbc.JDBCUtil Simon Stewart wrote:
> Standard JDK classes: java.lang.String, java.net.URL and > java.util.Date. > > It seems as if the code is not going through the steps necessary to > unmarshal (?) the Object before returning it. From your description, > it sounds as if some "instanceof" testing is done before the class is > returned. I would hazard a guess that a test to see whether the result > is of type "byte[]" doesn't then go on to check whether this is a > MarshalledObject? > > Out of curiousity whereabouts is the code? I can download the latest > CVS code and have a look at it given enough time. > > Also, looking at the interface I gave below, there is a slight error: > I used Serializable rather than Object to ensure that data _could_ be > stored in the database in the first place. > > Thanks for looking into this. > > Regards, > > Simon > > On Thu, May 16, 2002 at 09:31:45AM -0500, Dan Christopherson wrote: > >>I do remember looking at this code in JAWS and saying to myself 'I don't >>see how that can work, but nobody's complaining, so I'll leave it be.' >>Simon, I take it you're getting this problem with non-jdk classes? There >>may be some crufty instanceof testing in there. >> >>-danch >> >>Simon Stewart wrote: >> >>>On Wed, May 15, 2002 at 10:48:48PM -0500, Dain Sundstrom wrote: >>> >>> >>>>>I'm starting to suspect that JBoss might be the culprit here. If I >>>>>manually extract an object from the byte array, it turns out to be a >>>>>java.rmi.MarshalledObject when the expected class was java.lang.String. >>>>> >>>>>Can anyone shed some light on this? I'm using CMP 2, if that helps. >>>>> >>>>> >>>>JBossCMP stores objects in serialized form using a MarshalledObject. I >>>>copied this code from JAWS, so I am not sure of the reasons. I think it >>>>has something to do with portability of stored objects, but the real >>>>reason is it works. >>>> >>>> >>>Agreed, storing Objects as MarshalledObjects is the Right Way to store >>>things into the database so that they can be recovered, but that >>>doesn't explain why when I store an Object via CMP, a byte array is >>>returned on subsequent get()s. I can convert the byte array into an >>>ObjectInputStream, pull out the MarshalledObject and then do a get() >>>on that to restore the object, but that shouldn't be necessary. >>> >>>For an interface like this: >>> >>>public void setObject( String key, Object value ); >>>public Object getObject( key ); >>> >>>and then attempt to use it: >>> >>>SomeEJB.setObject( "key", new URL( "http://www.foo.com"; ) ); >>>URL url = (URL) SomeEJB.getObject( "key" ); >>> >>>I should _not_ be getting a ClassCastException (from memory: this >>>happened last night, and I've kludged a work-around using Strings) >>>when I call get(), but that's exactly what happens! >>> >>>I can extract the necessary bits from my code and send it to you as a >>>simple test case if that would help.... >>> > _______________________________________________________________ Have big pipes? SourceForge.net is looking for download mirrors. We supply the hardware. You get the recognition. Email Us: [EMAIL PROTECTED] _______________________________________________ JBoss-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-user
