Skip to content

Instantly share code, notes, and snippets.

@jorgenpt
Created November 11, 2011 00:59
Show Gist options
  • Star 1 You must be signed in to star a gist
  • Fork 1 You must be signed in to fork a gist
  • Save jorgenpt/1356797 to your computer and use it in GitHub Desktop.
Save jorgenpt/1356797 to your computer and use it in GitHub Desktop.
Workaround for JRuby's timeout not working well with IO.popen
# Workaround for IO.popen.readlines being allowed to block indefinitely even if
# wrapped in a Timeout::timeout call.
#
# Test case:
# $ time jruby -rtimeout -e "timeout(1) { IO.popen('sleep 10').readlines }"
require 'jruby'
module Timeout
def self.timeout(sec, klass=nil)
return yield(sec) if sec == nil or sec.zero?
thread = Thread.new { yield(sec) }
if thread.join(seconds).nil?
java_thread = JRuby.reference(thread)
thread.kill
java_thread.native_thread.interrupt
thread.join(0.15)
raise (klass || Error), 'execution expired'
else
thread.value
end
end
end
include Timeout
@toddthomas
Copy link

Thank you so much for this! Fixes timeout for me in a case where it wasn't working. I had to change seconds on line 13 to sec.

@arielvalentin
Copy link

Did you file a bug with the JRuby team?

@headius
Copy link

headius commented Aug 18, 2014

The test case provided appears to work fine with JRuby master (9000) which has reimplemented process-launching. I'll see if there's anything that can be done with the interrupt logic on 1.7.x.

@headius
Copy link

headius commented Aug 18, 2014

Note: one problem with the above hack is that interrupting a blocking Java IO read (as is happening here) generally closes the underlying stream and it can no longer be used. That's why we've gone back and forth on whether to do a hard Thread interrupt or just block.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment