Re: IOR reference of ORbit2 not recognized by mico

[Mark McLoughlin <mark skynet ie>]

Hi Michael,

On 20 Mar 2002, Michael Meeks wrote:

> Hi Hing,
>
> On Wed, 2002-03-20 at 10:08, Hing-Wah Wan wrote:
> > IOP_TAG_MULTIPLE_COMPONENTS:
> >     IOP_TAG_COMPLETE_OBJECT_KEY: object_key (24)
> >     '000000003ba9eb0f4398b0c50100000047f8e840333a6774'
> >
> >         Unknown component 0x1
> >        ^^^^^^^^^^^^^^^^^^^^^
> > from the response in the mico mailing list,
> > it seem that mico fail to recognize the IOR because of this unknown
> > component ,bugs?
>
> 	Ok - please tell the Mico mailing list that this tag is the CODE_SET
> tag that tells the world that ORBit2 is a UTF-8 only ORB:
>
> /* we always marshal the same thing: see 13.7.2.4 */
> static void
> IOP_TAG_CODE_SETS_marshal(CORBA_Object obj, GIOPSendBuffer *buf,
> 			  IOP_Component_info *ci)
> {
> 	/* To get these magic numbers see the 'OSF Character
> 	   and Codeset Registry'; ftp.opengroup.org/pub/code_set_registry */
> 	CORBA_unsigned_long utf8_key  = 0x05010001;
> 	CORBA_unsigned_long utf16_key = 0x00010109;
>
> 	/* Marshal a CodeSetComponentInfo structure */
> 	CodeSetComponent_marshal (buf, utf8_key, NULL);
> 	CodeSetComponent_marshal (buf, utf16_key, NULL);
> }
>
> 	Mico should most likely just ignore it - it's only there for paranoid
> reasons, to satisfy a Sun chap that we're doing things right and that
> the JavaORB will know the strings are utf-8 and not something whacked
> out - which would be stupid anyway.

	The problem is we're encoding it wrong ... just testing the
fix right now ...

Good Luck,
Mark.