wait [<max-time>]
Ordinarily Lens cannot be run in the background if it is idle. It may be run in the background if it is just going to process a script and then exit, which is the normal case when training on a single processor. However, because it attempts to read from standard input when it is idle, Lens will suspend itself if it is simply put in the background where it cannot access standard input.
There are cases, which primarily arise during parallel training, when a
Lens process must sit idle while it waits for network communications.
We would often like such processes to run in the background. To do
this, the wait command must be used. This causes Lens to
enter a loop in which it will process GUI and network events but will
not attempt to read from standard input. Typically, a parallel training
client would load a network, connect to the server, and then call
wait, never to return until the server tells it to exit.
The default behavior of the command is to wait indefinitely. Given a
max-time in seconds, wait will return at that point
and, if it was called in the shell, the shell will once more be active.
A wait can be terminated by an interrupt signal (Ctrl-C) or
by pressing the button at the bottom of the Main Window when it reads "Stop Waiting".
If "wait 0" is executed during a wait, the wait will end.
If several waits are occurring simultaneously, the top one on the stack
will be terminated. Telling the client to execute "wait 0"
is one way a parallel training server could tell a client to stop
waiting without actually killing the client.
To wait indefinitely:
lens> wait
To make a "perpetual client" that repeatedly tries to connect to a server
at a particular machine and port:
while 1 {catch {startClient spock.starfleet.edu 9999}; wait 60}
In this way, server processes could come and go and the client would
automatically connect into whatever server was listening at port 9999 at
the time. Once a connection to a server was established, it would not be
affected by attempts to reconnect.