# Start a program, get it's output, then kill it. while "multithreading" **URL:** \ **Category:** Ruby\ **Created:** [January 18, 2008, 11:31am UTC](https://www.ruby-forum.com/t/start-a-program-get-its-output-then-kill-it-while-multithreading/126456 "2008-01-18T11:31:20Z")\ **Posts on this page:** 7\ **Page:** 1
**Author:** ![Deadolus](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/8491ac/32.png) [@Deadolus](https://www.ruby-forum.com/u/Deadolus)\ **Post date:** [January 18, 2008, 11:31am UTC](https://www.ruby-forum.com/t/start-a-program-get-its-output-then-kill-it-while-multithreading/126456/1 "2008-01-18T11:31:20Z")
Hi everybody, i’d like to program a gui for iperf ([Best Open Source Mac Software 2023](http://sourceforge.net/projects/) iperf), it should run under windows And \*NIX and I’m using fox as my gui toolkit, I’ll use some extra calculations to get the parameters, that’s why I can’t use an already existing gui for iperf. I already designed the gui and I’m now moving to connect the different widgets to it’s functions. After you set all your parameters (size of packet, bandwith, time, …) I want to press a button and it should start iperf and parallely read the it’s output, so I can display some statistics (running time and, on server side, the packet loss, maybe more). So what should I program to do this? I read about multithreading, IO.popen would be perfect, but how can I kill iperf when I want to quit, before iperf ran out of time? I think threads could also work (especially because one can kill threads, if no longer used so the program would be terminated), but what do I do to start the iperf? I’m new to this subject and would appreciate any help on this! Thanks in advance, Deadolus ---
**Author:** ![Deadolus](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/8491ac/32.png) [@Deadolus](https://www.ruby-forum.com/u/Deadolus)\ **Post date:** [January 18, 2008, 11:57am UTC](https://www.ruby-forum.com/t/start-a-program-get-its-output-then-kill-it-while-multithreading/126456/2 "2008-01-18T11:57:09Z")
2008/1/18, Deadolus [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid): > So what should I program to do this? > I read about multithreading, IO.popen would be perfect, but how can I > kill iperf when I want to quit, before iperf ran out of time? > I think threads could also work (especially because one can kill > threads, if no longer used so the program would be terminated), but > what do I do to start the iperf? First of all killing a thread won’t help because the process is in no way attached to a Ruby thread (btw, this would also be true if Ruby was using native threads). You can however use a thread to read output from IO.popen. For example, you could stuff lines into a Queue which is read from your UI thread and displayed. Or whatever you want do do with it. About the killing I am not sure. With a quick check I did not find a proper way to get the PID of the child process with IO.popen. If there is no way you could use popen with “-” as command (i.e. start a child process), have that writ out the current PID (in $$) to stdout and exec the process you want to start. The parent would then have to read the child PID first and remember it somewhere. Kind regards robert ---
**Author:** ![Deadolus](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/8491ac/32.png) [@Deadolus](https://www.ruby-forum.com/u/Deadolus)\ **Post date:** [January 18, 2008, 9:18pm UTC](https://www.ruby-forum.com/t/start-a-program-get-its-output-then-kill-it-while-multithreading/126456/3 "2008-01-18T21:18:49Z")
On Fri, Jan 18, 2008 at 07:56:34PM +0900, Robert K. wrote: > 2008/1/18, Deadolus [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid): > > > … > > I read about multithreading, IO.popen would be perfect, but how can I > > kill iperf when I want to quit, before iperf ran out of time? > > … > > About the killing I am not sure. With a quick check I did not find > a proper way to get the PID of the child process with IO.popen. If > … It’s a little hard to find until you already know where it is, but IO#pid might come in handy here 🙂 ---
**Author:** ![Deadolus](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/8491ac/32.png) [@Deadolus](https://www.ruby-forum.com/u/Deadolus)\ **Post date:** [January 18, 2008, 7:57pm UTC](https://www.ruby-forum.com/t/start-a-program-get-its-output-then-kill-it-while-multithreading/126456/4 "2008-01-18T19:57:18Z")
On 18 Jan., 11:56, Robert K. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > > After you set all your parameters (size of packet, bandwith, > > First of all killing a thread won’t help because the process is in no > > child process), have that writ out the current PID (in $$) to stdout > > and exec the process you want to start. The parent would then have to > > read the child PID first and remember it somewhere. > > Kind regards > > robert > > – > use.inject do |as, often| as.you\_can - without end I think I found the solution, this is a example script: io = IO.popen(“ping google.ch”) #open a new io thread = Thread.new(io) {while !io.closed? do Thread.current[“line”] = io.readline end}#ALWAYS read current line of io i=0 #example code… while i\<10 do puts thread[“line”] puts thread.status #puts out “sleep”, so it’s running… i+=1 sleep(1) end io.close #close io thread.status #puts out “nil” so it’s terminated Seems like you easily can close a io just by “close”. At least it isn’t in my ps aux list any more… But I just tested under linux so far, so I can’t say if this also works under windows (I currently have some problems with my windows box). What do you think about this solution? If anyone has a better solution, I’m still interested… ---
**Author:** ![Deadolus](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/8491ac/32.png) [@Deadolus](https://www.ruby-forum.com/u/Deadolus)\ **Post date:** [January 19, 2008, 11:45am UTC](https://www.ruby-forum.com/t/start-a-program-get-its-output-then-kill-it-while-multithreading/126456/5 "2008-01-19T11:45:53Z")
On 18.01.2008 21:18, Stephen L. wrote: > It’s a little hard to find until you already know where it is, but > IO#pid might come in handy here 🙂 Thanks! I use #popen too infrequently and when I use it I do not normally fiddle with the process directly. Kind regards robert ---
**Author:** ![Deadolus](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/8491ac/32.png) [@Deadolus](https://www.ruby-forum.com/u/Deadolus)\ **Post date:** [January 21, 2008, 11:02pm UTC](https://www.ruby-forum.com/t/start-a-program-get-its-output-then-kill-it-while-multithreading/126456/6 "2008-01-21T23:02:15Z")
On 19 Jan., 11:42, Robert K. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > > > a proper way togetthe PID of the child process with IO.popen. If > > > robert > > > I wasn’t really happy about my solution, so I continued > > > experimenting. > > > I now have something, which I’m quite happy about, I’ll post it here, > > > maybe some people have the same problems as I had. > > > I expanded the normal IO class: #-------------------File OwnIO.rb-------------------------------------- module OwnIO def killed? #nil if not killed, pid of IO if killed #I think we do not need this, please correct me, if I am wrong: #if(self.class != IO) then raise “Wrong type of Input variable, is #{self.class} but should be IO!” end begin ``` Process.waitpid(self.pid,Process::WNOHANG) rescue Errno::ECHILD #rescue if process is already terminated self.pid end ``` end def kill #kill a IO (you still have to close it) Process.kill(“KILL”,self.pid) end end class IO #add my own IO module to the normal IO class include OwnIO end #---------EOF---------------------------------------------- And here is a short example script (how I’ll implement it in my GUI more or less): #-------------------test.rb--------------------------------- require “OwnIO.rb” STDOUT.sync = true old = nil #old output i = 0 io = IO.popen(“ping google.ch”) io.sync = true t = Thread.new() {while !io.killed? do Thread.current[“line”] = io.readline end}#first doesn’t output all lines (Thread creation too slow?) while !io.killed? do if t[“line”]!=old then#only react to changes of the line old = t[“line”] puts t[“line”]#puts the current line if i==20 then ##kill the io after 20 lines have been received io.kill end i+=1; end end puts “Killed” #---------------------------- EOF---------------------------------------- The two functions “killed?” and “kill” are self explanatory, and seem to work nice. But at least “killed?” could have some improvements… So if anybody has some improvements/suggestions to the code: go ahead! Regards, Deadolus ---
**Author:** ![Deadolus](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/8491ac/32.png) [@Deadolus](https://www.ruby-forum.com/u/Deadolus)\ **Post date:** [January 22, 2008, 4:36am UTC](https://www.ruby-forum.com/t/start-a-program-get-its-output-then-kill-it-while-multithreading/126456/7 "2008-01-22T04:36:53Z")
On Jan 21, 2008, at 2:59 PM, Deadolus wrote: > So if anybody has some improvements/suggestions to the code: go ahead! if you care about stderr this won’t work on windows or nix. threads, io, and guis will hang you on windows unless you really know what you are doing. for a portable way to start a process, capture it’s stdout/stderr, and have a handle on the pid see my systemu lib - it’s the only thing that’ll work afaikt (read archives) regards. a @ [http://drawohara.com/](http://drawohara.com/)