reconnect should fall back gracefully

[ConfusedVorlon]

I love how robust the reconnection is, however if connections fail repeatedly, then the default of attempting every 0.5 seconds is going to hammer a struggling server.

If the reason that the connection is failing is a server in trouble, then this potentially adds something close to a denial of service attack on top of whatever problem exists...

Standard practice would be an exponential fallback - so, waiting (twice?, 1.5x?) as long between each reconnection attempt.
Perhaps to a maximum of a 60 second delay between checks...

I'm happy to have a go if you'd accept a pull request...

[nerzh]

I'll be glad to accept your pull request, but please make the reconnect parameter data type enum instead of boolean. Let the enum contain three cases: none, standard, and your reconnect case

[ConfusedVorlon]

will do.

Before I get stuck in - could you explain a bit around the locking in PingWatcher ?

I understand the locking around finish - this is accessed from the public isFinish() function, and also from the watcher thread.

Other variables like lastTimePoint seem to be only accessed from the watcher thread (and are private).

Is this just a case of being overly cautious, or am I missing some subtlety?

[nerzh]

This library was made very quickly for my project, it still works there. I want to say that there was no goal to make a library. I realized that it might be useful to others, so I made it public. This means that I wrote the code a long time ago and there may be inaccuracies. Just make your code work, you can rewrite my watcher if needed