Re: [6lo] New I-D draft-delcarpio-6lo-wlanah-00.txt - HC

[Alexandru Petrescu <[e-mail]>]

Carsten,

Le 17/06/2015 20:01, Carsten Bormann a écrit :
> Alexandru Petrescu wrote:
>> The section 8 is about HC.
>>
>> I would like to ask whether HC means to eliminate the IPv6 Base
>> Header and if yes then the subsequent question is whether we can
>> still consider this to be "IPv6 over..." when there is no IPv6.
>>
>> Because RFC2460 says that the IPv6 Base Header is mandatory.
>
> Alex,
>
> you are trying to make this point again and again.

Ok, let me summarize  instead of turning in circles:

I would like to request a modification of that Internet Draft - please 
add a statement saying that the IPv6 Base Header is mandatory, in clear, 
in network byte order.

> Indeed, header compression is a concept that is usually not taught at
> universities.  My standard talk about header compression starts with
> pointing out that there isn't even a Wikipedia page about this
> subject!

I am not sure a wikipedia page on Header Compression is necessarily good 
for HC altogether.

If one wants to study Header Compression there is plethora of 
authoritative information in the Information Theory section of your 
University Library.

> However, as manifested in RFC 1144, there is a long tradition of
> using header compression in link layer mappings for the Internet.

I think RFC1144 is no longer in use.  It should be declared Historic.

ROHC (RObust Header Compression RFC6846) is not applied to IP layer, as 
far as I know, but to upper layers.  Despite being touted decades ago as 
absolutely necessary for WiFi and cellular, what we use today on these 
links is pure IP.

Am I missing something?

> The fact that you cannot make out the IPv6 header when looking at
> the bytes on the air in a 6LoWPAN doesn't mean it is not being sent.
> It is just encoded in an efficient way.

What is meant by 'encoded in an efficient way'?

If what is meant is substituting a shorter header for the IPv6 Base 
Header then that's obviously not IPv6.  Because the IPv6 Base Header 
must be transmitted in full in clear in network byte-order.  The only 
modifs the network can make to it is decrementing the Hop Limit.

To further understand what is meant by 'encoded in an efficient way' I 
looked at the wireshark dumps using 6lowpan headers on 15.4 at:
http://svn.tools.ietf.org/svn/wg/6lowpan/packets/
(e.g. dis_option7.pcap)
They exhibit that there is a 15.4 header, a 6lowpan header _and_ an IPv6 
Base Header.  Efficiency couldnt mean to insert an additional header:

Ethernet II header                    Ethernet II header
IEEE 802.15.4 Data header             IEEE 802.15.4 Data header
6LoWPAN header                vs.     IPv6 Base Header
IPv6 Base Header                      ICMPv6 header
ICMPv6 header

This is why I dont understand what is meant by 'encoded in an efficient 
way'.

> (This is different from the confused idea of "header stripping",
> where the IP header actually is not made available to the recipient
> in a way such that it can be decoded.)

Well, please clarify: is the IPv6 Base Header present or not?  If it's 
not then it's stripped, right?

Alex

>
> Grüße, Carsten
>
>