IPv6 strawman plan: Re: [OpenAFS-devel] Moving Forwards

[Simon Wilkinson]

On 10 Sep 2012, at 06:19, Jeffrey Altman wrote:

> While you may fork the code base, you may not fork the AFS3 wire
> protocols.

This is just not true. There are no internet standards police. In the =
same way as anyone can run a service that looks like HTTP, but isn't =
quite on port 80, there's absolutely nothing stopping Troy (or anyone =
else) running a mutant filserver on port 7000. If you believe that the =
AFS trademark remains enforcable, you might get into trouble if you =
called that service AFS - but that's not what Troy is proposing

> On 9/10/2012 12:33 AM, Troy Benjegerdes wrote:
>>=20
>> Is there anything else I'm missing (besides the flamewar that
>> will probably follow regarding the name change?)

The big things that you are missing are the ubik election protocol, and =
the vl database. The database uses fixed size records that can't be =
easily changed (they're tied to the ubik page size, amongst other =
things). You would need to figure out a way to store IPv6 addresses =
within this database.

However, I think all of this is somewhat premature. As I have said on =
many occasions before, there is low hanging fruit here. Making RX IPv6 =
capable is a relative straightforward operation. It requires no protocol =
changes, and can be done without affecting the existing API. If you, or =
anyone else, is serious about implementing IPv6 support, this would be a =
great place to start work, and to convince the community that you are =
prepared to back your words with action.

Cheers,

Simon.