#ifndef __PING_H_
#define __PING_H_

// When the client expects to hear from the server, and hasn't heard anything
// in a while, it will send a "ping" command.  The primary purpose of this
// command is to test the network.  No response is expected.  This module 
// logs all such requests.
//
// The client will send a ping after about 5 seconds without getting a GetA
// response.  The thought is that the server might have broken the connection,
// but the client doesn't realize it.  By attempting to send a message, the
// underlying TCP/IP logic will either deliver the message or report an error.
// If there is a problem, we should know quickly.  On the other hand, if we
// don't send anything at the application level, the O/S might only send a ping
// about once every 24 hours!  (This is configurable, I don't know what we're
// really getting.)
//
// I don't know if this actually solves any problems, or not.  I know that this
// type of problem can exist.  But it usually happens on long time frames.  If
// I leave an SSH session open all night, and I don't tell it to send a
// keep-alive message, and I'm behind a standard home firewall appliance, the
// firewall will forget about the connection.  This is well documented.  There
// will be no error message until I try to type something.  I don't know if
// this happens on smaller timeframes.  The client already has a timeout and
// will disconnect after about a minute of waiting.  And the server will
// disconnect after about 5 minutes of idle time.
//
// This logging feature could also point out other problems.  If we see a lot
// of these it would mean that something was a lot slower than it should have
// been.  Of course, we should never see this.  We have other logic to try
// to send the GetA messages quickly, even if the database is slow and we
// can't provide real data.
//
// In summary, by logging the ping requests we are looking for unexpected,
// problems, which might otherwise be hard to debug.

void initPing();

#endif