[jboss-jira] [JBoss JIRA] Updated: (JBMESSAGING-361) WireFormatTest.testNullResponse(), testDeactivateResponse() fragments invalid

WireFormatTest.testNullResponse(), testDeactivateResponse() fragments invalid ----------------------------------------------------------------------------- Key: JBMESSAGING-361 URL: http://jira.jboss.com/jira/browse/JBMESSAGING-361 Project: JBoss Messaging Issue Type: Sub-task Components: JMS Facade Affects Versions: 1.0.0 Reporter: Ovidiu Feodorov Assigned To: Tim Fox Fix For: 1.2.0.Alpha1 Original Estimate: 4 hours Remaining Estimate: 4 hours Switching to jboss-serialization 1.0.0.CR7 uncovered what apparently seems to be an invalid test section in WireFormatTest public void testNullResponse() throws Exception { MessagingMarshallable mm = new MessagingMarshallable((byte)77, null); InvocationResponse resp = new InvocationResponse(null, mm, false, null); ByteArrayOutputStream bos = new ByteArrayOutputStream(); JBossObjectOutputStream oos = new JBossObjectOutputStream(bos); wf.write(resp, oos); oos.flush(); ByteArrayInputStream bis = new ByteArrayInputStream(bos.toByteArray()); DataInputStream dis = new DataInputStream(bis); // First byte should be version assertEquals(77, dis.readByte()); ........... } It doesn't make sense to me to write stuff to a JBossObjectOutputStream and then try to read the serialized content using a DataInputStream, and yet this is what this test section does. JBoss Serialization doesn't guarantee this in its public contract. Excerpt form a conversation with Clebert: flanker707: Does JBossSerialization guarantee that something written with on a JBossObjectOutuptStream can be read with DataInputStream? Clebert Suconic: no... I didn't mean to flanker707: is this in the public contract? Clebert Suconic: If you use the same thing with ObjectInputStream.. you would have the same problem testNullResponse(), testDeactivateResponse() sections commented out until clarified.