[plug] Wake-on-Lan

[James Devenish]

In message <20040909003629.GA24285 at mail.guild.uwa.edu.au>
on Thu, Sep 09, 2004 at 08:36:30AM +0800, James Devenish wrote:
> If we assume that the unicast, multicast and
> broadcast types are all referring to IPv4, then unicast should probably
> be very unreliable. On an Ethernet network, if the computer has been
> sleeping long enough, it will become unreachable except by Ethernet
> broadcast frames. Therefore, any attempts to do unicast waking will
> fail when the sender is unable to find the relevant neighbour by ARP.

By the way:

(a) This was just me guessing and speculating.
(b) I should also have said "in a typical environment", since
    some people might simply use static addressing tables that
    could make these things work.
(c) I have not seen any "Wake-on-LAN Specification", and have
    never used anything other than "MagicPacket(tm)".