The execution time of one operation is unpredictable, and I want to
bound it so that if it’s not completed in n seconds it falls back to a
“default” result. This is what I got:
batsman@tux-chan:/tmp$ cat b.rb
require ‘thread’
def exec_bounded_time time
mutex = Mutex.new
finished = ConditionVariable.new
completed = “not complete”
th = Thread.new do
yield
# could be killed just here, but don’t care
mutex.synchronize do
completed = “completed”
finished.signal
end
end
Thread.new do
sleep time # up to time seconds
mutex.synchronize do
th.kill
$stderr.puts “Computation too slow, killing…” if defined? DEBUG
finished.signal
end
end
mutex.synchronize { finished.wait mutex } unless completed == "completed"
completed
end
DEBUG = true
puts "Status of computation: " + exec_bounded_time(5) { sleep 6 }
puts "Status of computation: " + exec_bounded_time(5) { sleep 1 }
batsman@tux-chan:/tmp$ ruby b.rb
Computation too slow, killing…
Status of computation: not complete
Status of computation: completed
I’m using 2 threads, 1 condition variable and 1 mutex, which seems
too much. I have the feeling that other solutions are possible, and I
kinda thought of using continuations, but I couldn’t get the semantics
completely right (I want everything to end when the time limit is
reached OR as soon as the computation is finished).
Improvements?
Where should exec_bounded_time be put in? Another module, top-level…?
Finally, any better name?
···
–
_ _
__ __ | | ___ _ __ ___ __ _ _ __
'_ \ /| __/ __| '__ \ / ` | ’ \
) | (| | |__ \ | | | | | (| | | | |
.__/ _,|_|/| || ||_,|| |_|
Running Debian GNU/Linux Sid (unstable)
batsman dot geo at yahoo dot com
Software is like sex; it’s better when it’s free.
– Linus Torvalds