Re: FW: [Rfid] SLRRP Vendor Extensions

At 10:28 AM -0400 7/22/05, Peter Spreadborough wrote:
>Setting the V bit will indicate the presence of the VendorID field in
>the message header. If the V bit is not set then the VendorID will not
>be present. On receipt of a message with the V bit set an implementation
>may decide to accept or reject that message.

BTW, once you start having binary fields that are or are not present 
based on the value of other bits, you lose one of the advantages that 
David Husak claimed for a binary format -- that you can just put the 
received buffer in memory and access fields directly without the need 
to parse anything

The "slap it in memory and just access it" plan also runs amok of 
other issues such as alignment constraints, data packing issues and 
endianness.  All of those can be worked around by proper coding, but 
it is an implementation consideration whether handling straight 
binary access will/won't be cheaper (in time, code size or runtime 
memory use) than parsing the binary data into internal structures 
upon receipt.  If you add variable-length or conditional fields, 
you'll quickly determine that parsing it once is much more efficient 
(and less prone to small incompatibility bugs all over your code).

Margaret

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