The most reasonable implementation would be to send a WOL packet just before the connection attempt, so that if the server is not running, it would restart its run, or send the WOL packet if the server is not immediately responding in a reasonable time, like a couple of seconds. The WOL packet, instead, is now sent after the connection timeout, if the remote host is not responding. The reason it was done like this was to overcome a few WOL problems and incompatibilities we encountered at the time of the first implementation. We actually retested all of this and it doesn’t seem that these problems and incompatibilities still hold. This means that a more reasonable implementation should come shortly, actually resolving the problem you detected. Thanks for reporting.