[lwip-devel] Re: LWIP and LWIPv6 projects

[Ivan Delamer]

Hi All,

Thank you for the vote of confidence on the work I'm doing on IPv6.

I do agree with your assessment of LWIPv6. It seems to be a native IPv6
version of LwIP, with IPv4-mapped addresses for "legacy" support of IPv4.
I think it is a clever approach for a dual v4/v6 stack. However, it looses
the possibility to have a v4-only implementation with smaller footprint.

So far I have been able to keep everything IPv6 related #if'd out, so the
footprint won't change for a v4-only setup. I've also managed to leave the
API practically untouched, which I think is desirable for everyone. The
implementation in general follows the same approach as for IPv4 parts
(e.g. Neighbor Discovery tables are similar to ARP, Multicast Listener
Discovery (MLD) is similar to IGMP, etc.)

I still need to test things a bit more, but I hope that I can report on
some v4 and v4/v6 footprint comparisons by the end of next week.

Cheers
Ivan


> Date: Fri, 19 Nov 2010 09:22:13 +0000
> From: Kieran Mansley <address@hidden>
> Subject: RE: [lwip-devel] LWIP and LWIPv6 projects
> To: lwip-devel <address@hidden>
> Message-ID: <address@hidden>
> Content-Type: text/plain
>
> On Thu, 2010-11-18 at 16:48 -0500, Bill Auerbach wrote:
>>
>> It may well be more effort to merge than to add IPv6 to the current
>> code
>> base.  (merge_time - IPv6_time_saved > current_release +
>> IPv6_time). :-)
>
> I agree, and I think the result of a merge is likely to be not as good
> (for lwIP) as implementing IPv6 for lwIP from scratch (as Ivan's doing).
>
> Kieran
>
>