[rbridge] Agenda items?

[Radia Perlman]

Michael Smith wrote:

>
>Someone please correct me if I'm mistaken.  My understanding of the proposal
>is that packets sent to an unknown L2 destination will be treated as
>broadcast and sent along the single spanning tree.  Once the destination is
>known, packets will follow the shortest path which may or may not be the
>same as the spanning tree path.  Packets in transit via the spanning tree
>may quite easily be passed by subsequent packets following the shortest
>path, hence the reordering.
>
>  
>
Ah yes, that might cause some reordering. I'm not sure how much we 
should/need to
stand on our head to avoid reordering, especially if it's OK to 
"sometimes" do it (after
a topology change), so it's clearly assumed that
reordinering sometimes is not fatal. But it's definitely an issue we 
should discuss.

It would be nice for applications to realize they are working on a 
network and be
written appropriately.

Radia