Re: [Rfid] Re: XML vs. Text vs. Binary

On Fri, Jul 22, 2005 at 06:56:14AM -0400, Scott Barvick wrote:
> Stephane,
>  
> From the security perspective, the protocol processing is layered on 
> top of standard security mechanisms as discussed in Section 4.2 of 
> the draft.  Therefore, the implementation can safely access fields 
> in payload as efficiently as possible.

I will leave it to implementors to decide whether they want to be 
robust against broken implementations or not. However, I like to point 
out two things:

a) As long as security processing is done in software, this will weight
   more than any encoding/decoding required, whether that is a binary
   format or not does not matter.

b) Our SNMP experience is that encoding/decoding is not at all an
   issue when it comes to the amount of CPU cycles burned. Most of 
   the time is spend accessing the instrumentation and in the security 
   processing (if enabled). So if you want to optimize CPU usage, 
   make sure you look at the whole system and spend the efforts
   where you can maximize the benefit.

Human readability is a major enabling factor. In theory, all you need 
is a decent library to handle binary encodings well but in the real 
world it seems to help if people can do things without relying on 
such a library. Human readable encodings (and I include XML here) 
simply enable more people to get into the "game" and this is 
ultimately driving the success of a technology.

/js

-- 
Juergen Schoenwaelder		    International University Bremen
<http://www.eecs.iu-bremen.de/>	    P.O. Box 750 561, 28725 Bremen, Germany

_______________________________________________
Rfid mailing list
Rfid@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/rfid