ipfs swarm connect with alternate address

[btc]

ipfs swarm connect works as intended the first time it is executed with valid args:

go-ipfs (master) λ. ipfs swarm connect /ip4/104.236.176.52/tcp/4001/QmaQujMiKngVVV6nRQsGRU1HVKriU7aGzUzL695zG6nqeW
connect QmaQujMiKngVVV6nRQsGRU1HVKriU7aGzUzL695zG6nqeW success

woohoo!

Now, let's connect via udp over a port that we're pretty sure isn't exposed.

go-ipfs (master) λ. ipfs swarm connect /ip4/104.236.176.52/udp/4312312/QmaQujMiKngVVV6nRQsGRU1HVKriU7aGzUzL695zG6nqeW
connect QmaQujMiKngVVV6nRQsGRU1HVKriU7aGzUzL695zG6nqeW success

What's the desired behavior in this case?

a) command merely ensures that a connection exists between local node and the peer

• b) command ensures that there's a connection over the specified protocol tcp/udp

[jbenet]

that's not even a valid multiaddr (udp port) . it doesnt error out on parsing?

b) command ensures that there's a connection over the specified protocol tcp/udp

correct behavior. maybe should change output to include the addresses.

[btc]

that's not even a valid multiaddr (udp port) . it doesnt error out on parsing?

• nope. however, this will panic and take down the daemon:

ipfs swarm connect /ip4/104.236.176.52

[whyrusleeping]

yeah, multiaddr likes to panic...

[jbenet]

:( i'll fix it.

[jbenet]

No longer panics (i fixed it), but the errors are unfriendly:

> ipfs swarm connect /ip4/1.2.3.4
Error: invalid peer address: protocol requires address, none given: ip4

> ipfs swarm connect /ip4/1.2.3.4/
Error: multihash too short. must be > 3 bytes

[jbenet]

Also to correct my statement above:

that's not even a valid multiaddr (udp port) . it doesnt error out on parsing?

Sorry, /udp/<port> is totally a valid multiaddr. i mistook udp for utp. /utp/<port> is wrong, it would be: /udp/<port>/utp