Can clients be reused after reactor shutdown?

[8573]

While looking into how I could implement reconnection after ping-timeout in irc-bot, I hit this question. Suppose that—

• several IrcClients are cloned;
• their clones are registered with an IrcReactor r with r.register_client_with_handler(...); and
• r.run() is called and, after running for a while, returns.

Can the same sequence of events safely be repeated, registering new clones of the same IrcClients with the same IrcReactor and running the reactor again?

I would guess that, if the reactor shut down because a connection failed, the IrcClient corresponding to that connection no longer would be usable, but what about the other clients running on the same reactor?

I would like to suggest that the answer be documented, covering whether an IrcClient c that was being run on an IrcReactor can be reused thus in the following cases:

• the run call returns Ok;
• the run call returns Err because of a problem with c; and
• the run call returns Err otherwise, such as because of a problem with a different IrcClient.

I note, largely for the sake of making GitHub create cross-references between these issue tickets, that this seems relevant to tickets #88 and #167.