Testing Thali-iOS: Multiconnect resolves with port number 0.

[lesn1kk]

It causes tests to fail because of error that this port is not available. I tried to run Can shift data via parallel connections multiple times and at some point it failed.
Logs

[enricogior]

From the log, we can see that there are two different things causing this issue.

[ThaliCore] BrowserManager.connectToPeer(_:syncValue:completion:) peer:C64DDE67-6011-427C-BDC0-9E94B512EB7B
[ThaliCore] Browser.inviteToConnect(_:sessionConnected:sessionNotConnected:) Peer(uuid: "C64DDE67-6011-427C-BDC0-9E94B512EB7B", generation: 0)
[ThaliCore] Session.init(session:identifier:connected:notConnected:) peer:C64DDE67-6011-427C-BDC0-9E94B512EB7B:0
[ThaliCore] Session.session(_:peer:didChange:) peer:BD7F2AF8-F739-4D63-A46B-CB02E373FDE3:0
[ThaliCore] Session.sessionNotConnected failed to connect to peer:BD7F2AF8-F739-4D63-A46B-CB02E373FDE3
[ThaliCore] BrowserRelay.closeRelay() disconnecting:false
[ThaliCore] Session.disconnect() peer:BD7F2AF8-F739-4D63-A46B-CB02E373FDE3:0
2017-06-12 12:26:37 - DEBUG thaliMobileNativeWrapper: 'multiConnectResolved: {"syncValue":"CPHHLKlZWuMYW9Ciiz4peARRwH6qOjxl","error":"Connection could not be established","portNumber":null}'
2017-06-12 12:26:37 - DEBUG thaliMobileNativeTestUtils: 'Got multiConnectResolved -syncValue: 'CPHHLKlZWuMYW9Ciiz4peARRwH6qOjxl', error: 'Connection could not be established', listeningPort: '%s''
2017-06-12 12:26:37 - DEBUG thaliMobileNativeWrapper: 'Received peer availability changed event with {"peerIdentifier":"BD7F2AF8-F739-4D63-A46B-CB02E373FDE3","peerAvailable":false,"portNumber":null,"recreated":true}'
2017-06-12 12:26:37 - DEBUG thaliMobileNativeWrapper: 'Filtered out nonTCPPeerAvailabilityChangedEvent due to not being in started state'
2017-06-12 12:26:37 - DEBUG thaliMobileNativeWrapper: 'Received peer availability changed event with {"peerIdentifier":"BD7F2AF8-F739-4D63-A46B-CB02E373FDE3","peerAvailable":true,"portNumber":null,"recreated":true}'
2017-06-12 12:26:37 - DEBUG thaliMobileNativeWrapper: 'Filtered out nonTCPPeerAvailabilityChangedEvent due to not being in started state'
2017-06-12 12:26:37 - ERROR thaliMobileNativeTestUtils: 'Fatal connect error: 'Connection could not be established''
2017-06-12 12:26:37 - DEBUG thaliMobileNativeTestUtils: 'Issuing multiConnect for C64DDE67-6011-427C-BDC0-9E94B512EB7B (syncValue: zjAqWcagknXt2ETg9H6evvB4uUWDUymq)'
2017-06-12 12:26:37 - DEBUG thaliMobileNativeTestUtils: 'Got 'multiConnect' callback'
[ThaliCore] BrowserManager.connectToPeer(_:syncValue:completion:) peer:C64DDE67-6011-427C-BDC0-9E94B512EB7B
[ThaliCore] BrowserManager.connectToPeer(_:syncValue:completion:) found active relay
2017-06-12 12:26:37 - DEBUG thaliMobileNativeWrapper: 'multiConnectResolved: {"syncValue":"zjAqWcagknXt2ETg9H6evvB4uUWDUymq","error":null,"portNumber":0}'
2017-06-12 12:26:37 - DEBUG thaliMobileNativeTestUtils: 'Got multiConnectResolved -syncValue: 'zjAqWcagknXt2ETg9H6evvB4uUWDUymq', error: 'null', listeningPort: '0''

There is a first attempt to connect to peer C64DDE67-6011-427C-BDC0-9E94B512EB7B that fails because it's a zombie advertiser.
Then there is a second attempt to connect again to the same peer, this time instead of returning an error, ThaliCore returns a successful connection with a port number set to zero.

We need to change the test so it will not try to connect to a peer that has already returned an error.
We also need to fix ThaliCore so it will remove the BrowserRelay after a failed connection, so it will not find it in the list of active relays and return a success with a port equal to zero.
I can open a new issue to track the ThaliCore bug.

[lesn1kk]

In tests we can probably filter all connections with port 0, so it won't be a problem. However, the fix for ThaliCore itself is more significant, because it could have direct impact on Thali, so this is definitely something we need to resolve.

[lesn1kk]

Turned out it was an issue in tests code after latest changes and unavailable peers were not removed from array correctly.

[enricogior]

The ThaliCore bug is tracked here #1906
I'm working on it.