On Tue, 6 May 2008, YOSHIFUJI Hideaki / µÈÆ£±ÑÌÀ wrote:

Hi,

In article [ref] (at Tue, 6 May 2008 11:07:49 +0300 (EEST)), Miika Komu [off-list ref] says:

quoted

quoted

Miika, where are your other patches on top of this?
This kind of change might be okay, but it must be along with
its users, at least.

we are maintaining a separate kernel module (GPLv2) for the HIP socket
handler for experimentation purposes. If you want to have a look it, you
can e.g. download the nightly tarball and have a look at the hipsock
directory:

http://infrahip.hiit.fi/hipl/hipl.tar.gz

Even though, we basically do not accept patches without in-kernel users.

Anyway, I've quickly looked into the code, and IMHO
it is better to avoid using hip_select_socket_handler() and
to introduce new hip_ops (or something) in sock{}.

Ok.

And, are you working on IPv4 transport?

The IPv6 socket interface is more useful for HIP-aware applications. We 
are implementing LSI support for non-hip-aware applications. The other HIP 
implementations (BSD HIP and Boeing) support already LSIs for 
non-hip-aware applications.

-- 
Miika Komu                                       http://www.iki.fi/miika/