Re: IGMPv3 for really dumb hosts

> So, the gain would be mostly from the host side, on hosts that
> haven't already implemented the complete IGMPv3.  I do not have
> enough information to know if enough hosts might benefit.

What I want to say is we must keep the interoperability of these
protocols on either of host- or router-side. We must avoid the
modification of the full version's spec.
Keeping a good condition for the protocol implementation is also
important. If the lite version only eliminates functions or
definitions from the current implementation (i.e. full version), it'll
be easy to make the lite version from the current one. However, if the
lite version defines new protocol messages or behavior, the
implementation cost becomes higher.

Talking about the host-side implementation, I think there is the case
that one needs to do the code maintenance to support both versions
(full and lite) on a single kernel code, in order that a developer or
a user of the system easily switches the version by the kernel
configuration or some command (e.g. sysctl), based on the hardware
equipment, circumstances, or other conditions. For instance, some
hardware vendor or user may want to use KAME for a high spec server or
a tiny equipment. It would be nice if the same kernel code can be used
for both only by chaning the configuration.

In addition, there are other difficulties on the complete version's
implementations. For instance, the MSF Advanced APIs and the message
retransmission procedure for robustness variable times (because the
report message should be merged if some state change occurs before the
message transmission completes) are quite complex. If the lite version
is designed, we should consider what was difficult (or too much
enough) for the host implementation of the full version were and how
eliminate it effectively. Adding new messages or behaviors is opposite
from my sense at this moment.

I'm sorry if I'm talking about host-side implementation specific
matter too much here, or the discussion is different from your points,
but this is my intention.
--
Hitoshi Asaeda