<p dir="ltr"><br>
Le 11 janv. 2013 11:54, "Ethan Jackson" <<a href="mailto:ethan@nicira.com">ethan@nicira.com</a>> a écrit :<br>
><br>
><br>
>> The one thing that seems just a little odd about this is that it<br>
>> embraces the new struct netdev_tunnel_config less than I would have<br>
>> guessed. parse_tunnel_config() is where I noticed it the most: I<br>
>> would have expected that it would transform the key-value pairs into a<br>
>> struct netdev_tunnel_config, then the struct netdev_tunnel_config into<br>
>> netlink attributes in a second step. However, the code seems OK to<br>
>> me, it just wasn't exactly what I was expecting.<br>
><br>
><br>
> I did it that way to make the diffs more "obviously correct" throughout the series. I didn't want to add a bunch of logic to convert netdev_tunnel_config into netlink, and then two patches later delete it when we get rid netlink entirely. I figure, this way is less disruptive and gives me more confidence that the final series is correct.<br>
><br>
> That said, it doesn't matter in particular, if my reasoning doesn't seem valid I'd be happy to change it. Especially since we may end up merging this patch some time before the final transition to flow based tunneling.</p>
<p dir="ltr">It's fine as is. No worries.</p>