Shell Programming and Scripting

Hmm, then I'm still a little lost.

First script:

Code:

#!/bin/sh
while sleep 1; do
	xdotool search --name Chrome 2>/dev/null | while read id; do 
		xdotool set_window --name "Chrome" $id
	done
done

Second command:

Code:

google-chrome-stable

Third command:

Code:

killall chrome_retitled

Right now, I'm running:

Code:

#!/bin/sh
/home/jake/Scripts_Icons/chrome_retitled ##chrome_retitled = first script## &
google-chrome-stable
killall chrome_retitled

Could you show me how to put all three things into a single script?

Thanks once more.

Replace the script call with the actual while loops?

I posted here, thinking that would probably be the way. But my scripting skills are not up to inventing such variables.

But I'm relieved that there isn't an obvious way to simply string those three things together--I spent much time moving them around, but I always ended up having to call the one while script, running that in the background (as you suggested), then executing the two commands, one after the other, in a separate script.

What RudiC is trying to show you is the following: you have one script:

And in another script you call it and put it in the background:

In fact you need to take out your comments or at least put the "&" before the comment signs, otherwise it is just part of the comment, but that only as an aside. So, in fact, you have:

Code:

#!/bin/sh
/home/jake/Scripts_Icons/chrome_retitled  &
google-chrome-stable
killall chrome_retitled

The line i marked bold is where you call the first script, right?

Now, a script is basically a list of commands, bput together into a file. You can do that inside a shell too, using what is called "process substitution". You write ( .... ) and what is inside the braces is considered one command to the outside. You can redirect its input or output and everything else you do with a separate script - and that includes putting it in the background:

Code:

#!/bin/sh
( while sleep 1; do
	xdotool search --name Chrome 2>/dev/null | while read id; do 
		xdotool set_window --name "Chrome" $id
	done
done) &
google-chrome-stable
kill -15 %1

Notice that "%" is using the job control of the shell: if you have background jobs they are assigned consecutive small numbers which you can list with the jobs command. You can address the processes using these small numbers instead of their PIDs.

Furthermore notice that there is a difference between "process substitution" (( ... )) and "command substitution" ($( ... )) as they are easily confused. They (can) serve similar purposes but are different things.

You should NOT use killall anyway, because this sends signal 9 to the process(es). You should always try signal 15 first and only use signal 9 if that doesn't work.

I hope this helps.

bakunin

Code:

#!/bin/sh
( while sleep 1; do
	xdotool search --name Chrome 2>/dev/null | while read id; do 
		xdotool set_window --name "Chrome" $id
	done
done) &
google-chrome-stable
kill -15 %1

Your instructions are extraordinarily clear and helpful. First off, you combined things into a single script that almost terminates itself appropriately, which is close to what I was hoping for. Secondly, your explanations for why/how parentheses are used make sense and I see how I can apply the concept to other scripts I'm working on.

While I understand perfectly why users might choose

Code:

kill -15

rather than the more aggressive

Code:

killall

it turns out that

Code:

kill -15 %1

doesn't work in the script. This script, called "chrome," appears to terminate itself successfully:

Code:

#!/bin/sh
( while sleep 1; do
	xdotool search --name Chrome 2>/dev/null | while read id; do 
		xdotool set_window --name "Chrome" $id
	done
done) &
google-chrome-stable
killall chrome

Because of your clear explanation, I was able go into xubuntu's task manager and see that the script called "chrome" generates itself several times during actual browsing. Before exiting the chrome browser, you can watch "chrome" kill itself until there're only two "chrome"'s running, but it never gets down to one "chrome."

Is that why

Code:

killall

is necessary here? Because there are multiple "chrome"'s? I hope I'm right, because that would mean you've taught me something that I used to troubleshoot on my own.

Granted, the pedagogy here is small, but folks like me have to start somewhere.

Yes, exactly. The command kill sends a signal to a (exactly one) process, identified by the given process ID. Issue a

Code:

kill -l

to see a list of signals your system is able to send. killall, on the other hand, works on groups of processes: it will send the same signal (if i remember correctly signal 9) to all processes identified by a common command. You can try the difference with a little experiment: Issue the command

Code:

sleep 1000 &

three times. In the ps -er | grep sleep output you will see three times the same process "name" (in fact the command, as processes have no names, just IDs) but with three different PIDs, like this:

Code:

$ ps -fe | grep [s]leep
bakunin      5897  5893  0 21:12 pts/0    00:00:00 sleep 1000
bakunin      5898  5893  0 21:12 pts/0    00:00:00 sleep 1000
bakunin      5899  5893  0 21:12 pts/0    00:00:00 sleep 1000

You can terminate each one of them by kill -15 <PID>, but killall sleep will first search for all processes with a command "sleep", find these three and then send one after the other a signal.

Web browsers traditionally wreak havoc in the process model for reasons i have not fully figured out. I.e. it is not possible to start a Firefox (or Chrome, for that matter) two times, like any other program. And each window (i hate tabs, so my browser instances are all windows) is not a separate process but - honestly, i have given up on trying to understand. I all programs would behave that way i'd already asked for a permanent place in the asylum. Why a process started once has two processes instead of one i don't understand. If you are interested - and i hope you are, because this is what makes you learn more and more - i suggest you investigate and tell us if you find out. I don't have this xdotool you are using.

I hope this helps.

bakunin

I guess that all the chromes you see "several times during actual browsing" are child processes created / spawned by the parent chrome. When killing that, all child processes will disappear as well.
Why use kill at all? Check the existence of the parent process in the while loop (pgrep available?), and quit it when that process exits. (Untested!) example:

Code:

( sleep 1
  while pgrep google-chrome-stable
    do  xdotool search --name Chrome 2>/dev/null | 
        while read id
           do xdotool set_window --name "Chrome" $id
           done
        sleep 1
    done
 ) &
google-chrome-stable

Adapt the first sleep's interval to allow for the chrome process creation.