Hi Alan,
we retrieved the IDL from the OMG web site
http://www.omg.org/spec/CORBAe/20080201/CORBA_TypeCode.idl
to regenerate the Helper classes ... this didn't compile properly and
had to be manipulated to get the relevant exceptions required
IvalidName is defined in
http://www.omg.org/spec/CORBAe/20080201/CORBA_ORB.idl
again we extracted the necessary exception and place it in a CORBA
module to regenerated the Helper classes
the definition for the exception Bounds at the top level CORBA module
couldn't be found
so it was hand crafted.
other IDL located at
http://www.omg.org/spec/CORBAe/20080201/
I've placed a reference, in the bug, to record the IDL locations.
The OMG also provides a jar, or zip, file containing these generated
classes, all of which will throw a not implemented exception
in the associated methods of the provided classes. But these form a
reference of what is required for a compliant implementation.
Furthermore, if we look at the IDL to Java language mapping
specifications, the following can be found, which alludes to the
inclusion of the referenced helper classes for a compliant implementation:
4.19.2 Certain Exceptions
The standard CORBA PIDL uses several exceptions, Bounds, BadKind, and
InvalidName.
No holder classes are defined for these exceptions, nor are they in the
interface repository. However, so that users can
treat them as “normal exceptions” for programming purposes, they are
otherwise mapped as normal user exceptions,
including the generation of helper classes.
They are defined within the scopes that they are used. A Bounds and
BadKind exception are defined in the
TypeCodePackage for use by TypeCode. A Bounds exception is defined in
the standard CORBA module for use by
NVList, ExceptionList, and ContextList. An InvalidName exception is
defined in the ORBPackage for use by
ORB.
// Java
package org.omg.CORBA;
final public class Bounds
extends org.omg.CORBA.UserException {
public Bounds() {...}
}
package org.omg.CORBA.TypeCodePackage;
final public class Bounds
extends org.omg.CORBA.UserException {
public Bounds() {...}
IDL to Java Language Mapping, v1.3 65
}
final public class BadKind
extends org.omg.CORBA.UserException {
public BadKind() {...}
}
package org.omg.CORBA.ORBPackage;
final public class InvalidName
extends org.omg.CORBA.UserException {
public InvalidName() {...}
}
regards
Mark
On 15/07/2014 20:34, Alan Bateman wrote:
On 15/07/2014 18:27, Rob McKenna wrote:
Hi folks,
Looking for a review for the following:
bug: https://bugs.openjdk.java.net/browse/JDK-7130985
webrev: http://cr.openjdk.java.net/~robm/7130985/webrev.01/
As per the bug report: The IDLtoJava spec clearly specifies the need
of the helper class (not the Holder) for these files. But these are
absent in the JDK. So, if we try to send any of these classes across
in a client-server rmi-iiop model, it will throw an error complaining
about the absence of the helper class.
There is clearly a lot of history and would be useful to get
confirmation that these are required for OMG complication.
As this is generated code then I'm curious where the original source
is. Just wondering if it would make sense to check-in the original
CORBA_TypeCode.idl.
Do we have any standalone tests that can be used to exercise these
classes?
-Alan
