RE: [openorb-devel] Windows compile...

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 454-5900

My current evalution result (not 100% sure if I'm right):

The problem seems to be that the parsing tree for org.omg.CORBA.Any,
created when parsing org.omg.CORBA.Any[], is registered as type
org.omg.boxedIDL.CORBA.Any and the parsing tree for org.omg.CORBA.Any,
created when just parsing org.omg.CORBA.Any is registered as
org.omg.CORBA.Any. In the second case the lookup (with
return_existing_object(org.omg.CORBA.Any)) succeeds whereas it fails in
the first case...

The question is now why is the parsing tree registered under different
names and what makes it fail under Windows in any case?

Michael

> -----Original Message-----
> From: ope...@li...=20
> [mailto:ope...@li...] On Behalf=20
> Of Rumpf Michael
> Sent: Wednesday, July 28, 2004 3:55 PM
> To: rg...@ze...
> Cc: openorb-devel
> Subject: RE: [openorb-devel] Windows compile...
>=20
>=20
> > It's its likely that its an issue with how the complation=20
> > tree is built.=20
> > I suspect it normally works under IBM JDK because the class files=20
> > contain the members a a particular order, hence why it breaks=20
> > when using=20
> > the Sun JDK. ..so I was on a wild goose chase worring about /=20
> > vs \ problems.
>=20
> No, I just checked your thesis but IBM and Sun VMs behave the=20
> same under
> Linux. The problem seems to be related to differences in the=20
> Windows and
> UNIX VM/compiler/JIT implementations...
>=20
> Michael
> =20
> > Richard.
> >=20
> > Rumpf Michael wrote:
> >=20
> > >Hi Richard,
> > >
> > >it is getting more mysterious:
> > >
> > >When I strip down the itf RemoteComplex to just the Any[]
> > >echo_any(Any[]) then the problem appears under Linux as well!
> > >
> > >But when I have both methods in the interface with the Any=20
> type being
> > >parsed first then it works even for the array type. I assume=20
> > that this
> > >is only the case when the simple Any type comes first!
> > >
> > >:(
> > >
> > >Michael
> > >
> > > =20
> > >
> > >>-----Original Message-----
> > >>From: Richard G Clark [mailto:rg...@ze...]=20
> > >>Sent: Wednesday, July 28, 2004 3:05 PM
> > >>To: Rumpf Michael
> > >>Subject: Re: [openorb-devel] Windows compile...
> > >>
> > >>
> > >>Micheal,
> > >>
> > >>Just a heads up: I'm going to commit some rather drastic=20
> > >>refactorings to=20
> > >>the java2idl compiler in an hour or so, to tidy things up a bit.
> > >>
> > >>Cheers,
> > >>Richard.
> > >>
> > >>Rumpf Michael wrote:
> > >>
> > >>   =20
> > >>
> > >>>Just forget the latest mail. I shouldn't be doing=20
> several things in
> > >>>parallel ;)
> > >>>
> > >>>I stripped down the RemoteComplex file and the method that is
> > >>>responsible for the problem is:
> > >>>
> > >>>   org.omg.CORBA.Any[] echo_any( org.omg.CORBA.Any[] any ) throws
> > >>>RemoteException;
> > >>>
> > >>>The previous one with org.CORBA.Any as type works, only the=20
> > >>>     =20
> > >>>
> > >>array type
> > >>   =20
> > >>
> > >>>is making problems here.
> > >>>
> > >>>I will investigate further.
> > >>>
> > >>>Michael
> > >>>
> > >>>=20
> > >>>
> > >>>     =20
> > >>>
> > >>>>-----Original Message-----
> > >>>>From: Richard G Clark [mailto:rg...@ze...]=20
> > >>>>Sent: Wednesday, July 28, 2004 12:39 PM
> > >>>>To: Rumpf Michael
> > >>>>Cc: openorb-devel
> > >>>>Subject: Re: [openorb-devel] Windows compile...
> > >>>>
> > >>>>
> > >>>>Micheal,
> > >>>>
> > >>>>Looks like another bug with general file handling.
> > >>>>
> > >>>>The problem I get is with parsing the classes with the=20
> > >>>>java2idl compiler:
> > >>>>
> > >>>>compile-test-idl:
> > >>>>[idl2java] IDL to Java Compiler, Copyright (c) 2002-2004 The=20
> > >>>>Community=20
> > >>>>OpenORB
> > >>>>[idl2java] compile : test.idl
> > >>>>  [delete] Deleting directory=20
> > >>>>E:\openorb-clean\OpenORB\build\gensrc\test\org\omg
> > >>>>   [javac] Compiling 199 source files to=20
> > >>>>E:\openorb-clean\OpenORB\build\test
> > >>>>   [javac] Note:=20
> > >>>>E:\openorb-clean\OpenORB\src\test\org\openorb\orb\test\rmi\str
> > >>>>eam\StreamTest.java=20
> > >>>>uses or overrides a deprecated API.
> > >>>>   [javac] Note: Recompile with -deprecation for details.
> > >>>>[java2idl] Java to IDL Compiler, Copyright (c) 2002-2004 The=20
> > >>>>Community=20
> > >>>>OpenORB
> > >>>>[java2idl] compile :=20
> > >>>>org\openorb\orb\test\rmi\complex\RemoteComplex.class
> > >>>>[java2idl] <!>mapping_api::map_array1-->Error1
> > >>>>[java2idl] An exception occured while parsing file:=20
> > >>>>'org.openorb.orb.test.rmi.complex.RemoteComplex'!=20
> > >>>>(java.lang.RuntimeException:=20
> <!>mapping_api::map_array1-->Error1)
> > >>>>
> > >>>>Looking at the verbose output it would seem that for some=20
> > >>>>       =20
> > >>>>
> > >>reason the=20
> > >>   =20
> > >>
> > >>>>compilation tree isn't being constructed correctly, however=20
> > >>>>there are no=20
> > >>>>errors related to opening files.
> > >>>>
> > >>>>The problem could be related to the fact that CORBA IDs=20
> > seem to be=20
> > >>>>created from the filename of the classes and filenames from=20
> > >>>>       =20
> > >>>>
> > >>IDs, the=20
> > >>   =20
> > >>
> > >>>>whole thing is a mess (as we all know) but no one=20
> > currently has the=20
> > >>>>resources to rewrite it. Compliant IDL compiling requires tight=20
> > >>>>integration between the preprossor and the compiler and=20
> > >>>>       =20
> > >>>>
> > >>combined with=20
> > >>   =20
> > >>
> > >>>>some novelty rules thrown into the spec makes this a non=20
> > >>>>       =20
> > >>>>
> > >>trivial task.
> > >>   =20
> > >>
> > >>>>Checking out the old separate RMIoverIIOP package reveals=20
> > >>>>that this bug=20
> > >>>>was present there also, so RMIoverIIOP was always stillborn=20
> > >>>>       =20
> > >>>>
> > >>on Win32.
> > >>   =20
> > >>
> > >>>>Regards,
> > >>>>Richard.
> > >>>>
> > >>>>Rumpf Michael wrote:
> > >>>>
> > >>>>  =20
> > >>>>
> > >>>>       =20
> > >>>>
> > >>>>>Hi Richard,
> > >>>>>
> > >>>>>is this the problem you have been facing or is that=20
> another one:
> > >>>>>
> > >>>>>compile-test-idl:
> > >>>>>[idl2java] IDL to Java Compiler, Copyright (c) 2002-2004=20
> > >>>>>    =20
> > >>>>>
> > >>>>>         =20
> > >>>>>
> > >>>>The Community
> > >>>>  =20
> > >>>>
> > >>>>       =20
> > >>>>
> > >>>>>OpenORB
> > >>>>>[idl2java] compile : test.idl
> > >>>>>[idl2java] file:/C:/Documents and
> > >>>>>Settings/mrumpf/openorb/OpenORB/src/idl/test/test.idl:13:=20
> > >>>>>         =20
> > >>>>>
> > >>Unexpected
> > >>   =20
> > >>
> > >>>>>error
> > >>>>>
> > >>>>>I guess it has something to do with the blanks in the path.=20
> > >>>>>    =20
> > >>>>>
> > >>>>>         =20
> > >>>>>
> > >>>>I will move
> > >>>>  =20
> > >>>>
> > >>>>       =20
> > >>>>
> > >>>>>my stuff to another folder and try again.
> > >>>>>
> > >>>>>Cheers,
> > >>>>>Michael
> > >>>>>
> > >>>>>
> > >>>>>-------------------------------------------------------
> > >>>>>This SF.Net email is sponsored by BEA Weblogic Workshop
> > >>>>>FREE Java Enterprise J2EE developer tools!
> > >>>>>Get your free copy of BEA WebLogic Workshop 8.1 today.
> > >>>>>http://ads.osdn.com/?ad_idG21&alloc_id=10040&op=3Dclick
> > >>>>>_______________________________________________
> > >>>>>openorb-devel mailing list
> > >>>>>ope...@li...
> > >>>>>https://lists.sourceforge.net/lists/listinfo/openorb-devel
> > >>>>>
> > >>>>>
> > >>>>>
> > >>>>>    =20
> > >>>>>
> > >>>>>         =20
> > >>>>>
> > >>>>  =20
> > >>>>
> > >>>>       =20
> > >>>>
> > >>>-------------------------------------------------------
> > >>>This SF.Net email is sponsored by BEA Weblogic Workshop
> > >>>FREE Java Enterprise J2EE developer tools!
> > >>>Get your free copy of BEA WebLogic Workshop 8.1 today.
> > >>>http://ads.osdn.com/?ad_idG21&alloc_id=10040&op=3Dclick
> > >>>_______________________________________________
> > >>>openorb-devel mailing list
> > >>>ope...@li...
> > >>>https://lists.sourceforge.net/lists/listinfo/openorb-devel
> > >>>
> > >>>=20
> > >>>
> > >>>     =20
> > >>>
> > >>   =20
> > >>
> > >
> > > =20
> > >
> >=20
> >=20
>=20
>=20
> -------------------------------------------------------
> This SF.Net email is sponsored by BEA Weblogic Workshop
> FREE Java Enterprise J2EE developer tools!
> Get your free copy of BEA WebLogic Workshop 8.1 today.
> http://ads.osdn.com/?ad_idG21&alloc_id=10040&op=3Dick
> _______________________________________________
> openorb-devel mailing list
> ope...@li...
> https://lists.sourceforge.net/lists/listinfo/openorb-devel
>=20