The fallback.rs implementation generates incorrect down/up events on a loop

[cafedetal]

The fallback implementation is always reporting down/up events for all interfaces due to a bug in the code:

if-watch/src/fallback.rs

Lines 53 to 56 in 31c4780

	if addrs.iter().any(|addr| addr.ip() == old_addr.addr()) {
	self.addrs.remove(&old_addr);
	self.queue.push_back(IfEvent::Down(old_addr));
	}

Line 53 will return true if the interface is in both the stored old address list and the newly reported list, therefore generating an IfEvent::Down for all of the addresses in the old list. The addresses then get re-added with a IfEvent::Up in the next block.

A simple reversal of the logic fixes the bug

            if !addrs.iter().any(|addr| addr.ip() == old_addr.addr()) {

Tested on android - no more interface spam after fixing that logic.

[thomaseizinger]

Tested on android - no more interface spam after fixing that logic.

Sweet! Can you send a PR? :)

[cafedetal]

#31

[cafedetal]

@mxinden Since I see there was a new point release for libp2p, any chance this could be approved/added for the next one? I am using my own dependency patch but it's a cpu/battery concern for other Android users, especially depending on the processing they do of the spurious notifications.

[mxinden]

Just released your patch. Thanks @cafedetal. Again, impressive find.

Given that this is a patch release, downstream dependencies can update to it without intermediaries having to update. Does that make sense?

[stormshield-gt]

As #31 has been merged and a new version released, can we consider this issue closed?

[mxinden]

Thank you @stormshield-gt for the ping.