The following sleeps 10 seconds and starts a new process, continuing from that point in the child process.

{
    sleep 10;
    my $cpid = fork;
    exit 1 if not defined $cpid;
    exit 0 if $cpid;
}
[download]

Does anyone wonder why you would want to do this? It sounds like a commedian I once watched.

"I got home tonight and discovered I had been robbed.....
Cleaned out.....
It was really spooky.....
Everything in my house had been taken and replaced with an exact replica"

Is this really the question? Or is the question how do I terminate contact with a browser, or my script leaks like a seive but I think I can fix it is I keep restarting it?

cheers

tachyon

I remember it well :-))

Here, the furniture's been replicated, but the house has been moved to a new address. The process ID has been changed (making a browser connection think we're done) and all the kids are sent SIGHUP, which says "We've moved on, and we're not saying where!"

If we have some reinitialization to do, we can do it in an _init() function and set the HUP handler to call it, or else just call it directly,

{
    my $cpid = fork;
    exit 1 if not defined $cpid;
    exit 0 if $cpid;
    _init();
}
[download]

I would also like to know what OP needs to do.

After Compline,
Zaxo

Your question's far too vague to answer directly but:

• Load any existing state (see below)
• Use alarm to setup an alarm for 10 seconds, and put a suitable $SIG{ALARM} handler in place (see perldoc -f alarm)
• Wrap whatever is doing your work inside an eval {}
• Checkpoint whatever state you need to resume (using Storable, Data::Dumper, YAML, or maybe even DBI or Class::DBI)
• Use exec $0, @args to restart yourself.

Of course you might also want to look into using something like cron instead to start yourself periodically, although most implementations only have minute resolution (not second).