Re: draft-cmetz-v6ops-v4mapped-api-harmful-00.txt and draft-itojun-v6ops-v4mapped-harmful-01.txt

[Pekka Savola <pekkas@netcore.fi>]

On Fri, 12 Sep 2003, Francis Dupont wrote:
>  In your previous mail you wrote:
> 
>    Francis, could you clarify how your local credo translates to how you'd 
>    like to, for example, port applications to IPv6?
>    
> => I replace all AF_INET by AF_INET6, gethostbyname() by getaddrinfo(),
> gethostbyaddr() by getnameinfo(), etc. I use one socket at the
> server side because I consider there is only one IP, in the API it
> becomes IPv6 with IPv4 space injected (in the math meaning) into
> the IPv6 space as IPv4-mapped IPv6 addresses.

What about client apps -- do you set up AI_MAPPED flag when doing 
getaddrinfo?

When you log into a dual-stack system with SSH using IPv4, would you 
prefer the application would print that you connected from an IPv4 
address, instead of the mapped address?  How about programs which make 
lookups based on your address (e.g. sendmail doing RBL-like checks) -- 
should all the applications convert the mapped addresses to IPv4 
addresses?  Should there be a function for it?  Or should the kernel lie 
to the apps about it?

There are other very difficult problems with mapped addresses, I've seen..
 
>    We're in the process of updating 
>    draft-shin-v6ops-application-transition-02.txt, and it seems like there 
>    are about half a dozen (at least) ways to create IPv6 applications.
> 
> => there are two ways:
>  - consider that the two stacks are disjoined: the code is duplicated,
>    there is a version for IPv6 and a version for IPv4.
>  - consider that there is only one IP and use only the IPv6 representation
>    of any IP address.

I don't know whether you're talking about different ways to build
dual-stack implementation, or how to build dual-stack applications.  If
the latter, your two ways seem to be wrong.  There does not need to be any
code duplication at all, using getaddrinfo loops.
 
>    The 
>    different scenarios are quite complex.  Most of it seems to come from the 
>    different ways how the mapped addresses could be used.
>    
> => in fact the issue is more to use or not to use them.

That's correct yes, but the details of "to use them" are also a bit hairy.

>    Could you try to clarify?
>    
> => there is a zillion of messages about this in the archives...

.. with nothing properly written up, I guess.. :-(

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings