Re: CARD: Address length in L2-ID parameter

Hi Henrik,

Henrik Petander wrote:

>Hi Marco,
>
>On Fri, 10 Oct 2003, Marco Liebsch wrote:
>
>  
>
>>"...L2 id suboption should have address length field which MUST be
>>used at least with with L2 type = 0x00."
>>
>>Maybe I misunderstood the issue, but the actual address length can
>>be calculated from the L2-ID sub-option's Length indicator, which
>>is always present, isn't it?
>>    
>>
>
>Yes, but the sub-option is padded to be a multiple of 32 bits long, based
>on the way I understood 5.1.3. For example a 48-bit MAC address (and the
>L2 id subotion) would not end on a 32-bit boundary without 8 bits of
>padding.  And if you don't know the length of the address, you can't
>determine where the padding starts.
>
>To correct this there are IMO three alternatives: 1. the boundary
>restriction needs to be relaxed or 2. the length of the address needs to
>be specified in the packet format or 3. L2 type specification should be a
>MUST and type 0x00 should be removed.
>  
>
Thanks for the clarification. Of course, you are right.
Referring to the recent issue on the assignment of L2-type values, I 
must admit that
your option 1 or 2 are the candidates I would go for. Since we said that 
individual
sub-options should take care about 32-bit boundary alignment, relaxing 
this for the
L2-ID is not consistent. Hence, it won't be a problem to have another 
byte after the
L2-type field, which indicates the real length of the L2-ID field 
without padding.
However, having 2 length fields in one parameter doesn't look good, but 
might be the
best compromise.

What's the opinion of others? Any preferences here?

marco


>Henrik
>  
>