Issue 2709:  (sec-rev)
Source:  (, )
Nature: 
Severity: 
Summary: 
Resolution: 
Revised Text: 
Actions taken:
June 8, 1999: received issue
Discussion: 

End of Annotations:=====
Date: Fri, 17 Dec 1999 14:15:44 -0800 (PST)
From: Ken Cavanaugh <Ken.Cavanaugh@eng.sun.com>
Reply-To: Ken Cavanaugh <Ken.Cavanaugh@eng.sun.com>
Subject: Re: Announcement for interop 2000 vote 2
To: terutt@lucent.com
Cc: interop@omg.org, rip-dev@eng.sun.com
MIME-Version: 1.0
Content-MD5: hWKhG408M81L5DZRxPBM4Q==
X-Mailer: dtmail 1.2.1 CDE Version 1.2.1 SunOS 5.6 sun4u sparc 
Content-Type: TEXT/plain; charset=us-ascii
X-UIDL: WnY!!/O9e9AiN!!!95e9


> Jonathan Biggar wrote:
> 
> > Bob Kukura wrote:
> > >
> > > IONA votes as follows:
> > >
> > > 2709 - No - I think this may completely break messaging, which
> relies on
> > > being able to defragment a fragmented message without
> interpreting the
> > > data.  I can't see any value in switching byte orders between
> fragments,
> > > have never heard of an ORB doing it, and this we'd be much
> better off,
> > > regardless of messaging, not allowing the byte order to change
> for a
> > > message.  Similarly, we should require all fragments of a
> message to use
> > > the same GIOP version.  I apologies for not raising this sooner,
> and
> > > urge others to change their votes if they see fit.
> >
> > Bob, you are right.   I misread the resolution, and now that I
> have read
> > it correctly, I agree that we should explicitly disallow framents
> of
> > different byte-gender.  No sane sender would do it anyway, and it
> only
> > drastically complicates things for the receiver.
> >
> > Tom, please change Floorboard's vote on 2709 to no.
> 
> Tom,
> 
> Ditto for HP. Please change our vote on 2709 to NO.
> 
> Thanks,
> 
> Jishnu.
> 

After looking at this again, I agree.  While Sun's Java ORB could
easily
handle changing the byte order, there are some significant
optimizations
that we could make if we could assume that the byte order never
changes.  A quick look at the messaging spec also seems to indicate
that
we cannot change byte orders from one fragmnet to the next, as Bob
mentioned.

Please change Sun's vote on 2709 to NO.

Thanks,

Ken.