Re: WG next steps

[Masataka Ohta <mohta@necom830.hpcl.titech.ac.jp>]

Tony Li;

> |   The interaction is more complex.
> |   
> |   For example, if the current path fails, it is detected by the
> |   transport/applicaiton layer. Thus, it is necessary for the
> |   transport/application layer tell the network layer use an 
> |   alternative
> |   locator.
> |   
> |   Later, as the original path may be resumed, it is also necessary
> |   for the transport/application layer tell the network layer try an
> |   old locator. Note that there may be several old locators.
> |   
> |   To do so, transport/application layer and network layer must have
> |   common identification mechanism of locators.
> |   
> |   It is much easier for the transport/applicaiton layer to handle
> |   locators directly. Then, all the information necessary for
> |   transport/application layer control is included in the IP header.
> 
> 
> All of this could be handled by a set of opaque handles to locators,

That's what I'm saying.

> without the direct interaction of the transport layer and the 
> locators.

But, it is much easier.

Have an engineering point of view.

							Masataka Ohta