fix: wait to set socket to active after chain_sync transitions to the correct state

[daveminer]

Fixes the bug where a received message could be processed by the socket while in a state not meant for message processing.

[daveminer]

We might want to see if there is a yaci config that can help test this before we merge it.

[caike]

The reason for setting the socket to active here and then below is that we need it to be after invoking the read_remaining_payload function. Notice that line 277 manually calls client.recv(socket, recv_payload_length, @recv_timeout)

I'm running this branch and don't see new blocks displayed so I suspect the changes break the current functionality.

[daveminer]

Interesting, it was testing out for me. Am I missing something?

Even so, what you say makes sense and I realizing this probably isn't what this comment is talking about:
https://github.com/wowica/xander/pull/48/files#diff-61848f9c738b8acca1018367012bcf6f97ca7ac8de8e8a6efbcbf2df4765d5f5L397

[caike]

Fixes the bug where a received message could be processed by the socket while in a state not meant for message processing.

Not sure I understand the scenario. Left a comment on the code but am curious to hear more about the issue you had in mind.

[daveminer]

Fixes the bug where a received message could be processed by the socket while in a state not meant for message processing.

Not sure I understand the scenario. Left a comment on the code but am curious to hear more about the issue you had in mind.

I was looking to resolve the bug mentioned here:

https://github.com/wowica/xander/pull/48/files#diff-61848f9c738b8acca1018367012bcf6f97ca7ac8de8e8a6efbcbf2df4765d5f5L397

Regardless of the comment above, I think I misinterpreted what the TODO was saying. The issue isn't that there's a period of time in the new_blocks/3 where the socket is in the wrong state, it's that we could receive socket messages (:info) while we're in catching_up, and we don't have a method signature to handle it.

I've made a commit that should be closer to an actual solution for the problem for discussion purposes. The code is redundant and would need to be DRYed up, but right now I'm just trying to make sure it's addressing the actual issue.