using xargs and read from standard input

[Billa-User]

hello,

is it possible to use command "xargs" , start a subprogramm and read in the sub programm from standard input ?
example below, it isn't possible to read from standard input in "sub programm". maybe any ideas to fix it ? command "xargs" exists in many shell scripts.

main programm:
#!/usr/bin/ksh
cat xargs.lis | xargs -l -x -t ./u_xargs-test.sh

sub programm:
#!/usr/bin/ksh
echo "this is script: `basename $0`"
echo "Number. Par: $#"
echo "Parameter: $*"

echo "Input: "
read input <= isn't possible to read from standard input

echo $input

regards,tom

[Billa-User]

sorry, i forget to describe the content of example file "xargs.lis" :

parameter1
parameter1 parameter2
parameter1 parameter2 parameter3

i called "main programm" : h_xargs-test.sh,
"sub programm" u_xargs-test.sh

regards,tom

regards,tom

[James R. Ferguson · Accepted solution]

Hi Tom:

One way is to change your subprogram to specifically associate STDIN with '/dev/tty' as:

#usr/bin/ksh
exec 0< /dev/tty # NOTE!
echo "this is script: `basename $0`"
echo "Number. Par: $#"
echo "Parameter: $*"
echo "Input: "
read input
echo $input

Regards!

...JRF...

[Peter Nikitka]

Hi,

besides JRF's solution, you can see it clearly, that such reading of stdin will not work plain, when you look on this equivalent form of your command:
xargs -l -x -t ./u_xargs-test.sh
Stdin is already redirected.

mfG Peter

[Billa-User]

thank you very,very much, great solution.

[Billa-User]

hello,

i have a new question to this old thread:

i detect following case:

i use this statement (u_xargs-test.sh is a sub-programm):
xargs -l -x -t ./u_xargs-test.sh
when a error occurs in "./u_xargs-test.sh" or i want to stop program with "exit 1" the "loop" of xargs doesn't stop. also "set -e" in "h_xargs-test.sh" or "u_xargs-test.sh" has no impact.

how can i stop sub program , when i use the statement "xargs" ?

regards

[James R. Ferguson]

Hi Tom:

> how can i stop sub program , when i use the statement "xargs" ?

I don't know of anyway to do this. The 'xargs' process invokes your script for bundles of '-l' size arguments --- in this case *one*. This means that a new process (your script) is instantiated for every line read from you input. Whether or not the script (process) succeeds or fails doesn't matter to 'xargs'; it merely reads another bundle of input.

Regards!

...JRF...

[James R. Ferguson]

Hi (again) Tom:

OK, duh! If you want your invoked process to terminate the 'xargs' statement, simply issue a 'kill ${PPID}' in your script.

Regards!

...JRF...

[Dennis Handly]

>JRF: If you want your invoked process to terminate the 'xargs' statement, simply issue a 'kill ${PPID}'

You may want to check to see that your parent actually is xargs(1).

[Billa-User]

hello,

> You may want to check to see that your parent actually is xargs(1).

xargs terminates if it receives a return code of -1 from command or if
it cannot execute command. When command is a shell program, it should
explicitly exit (see sh(1)) with an appropriate value to avoid
accidentally returning with -1.

the "loop" of xargs continues , when i try to stop the sub-shell-program like

if [ ! -d directory ]
then
echo "dir not found"
exit 1
fi

so i have to change command "xargs" to a statement like "while" , isn't it ?
it is no so easy to change those commands :

nl -ba -w1 -s" " xargs.lis | xargs -l -x -t ./u_xargs-test.sh par1

regards

[Dennis Handly]

>the "loop" of xargs continues, when I try to stop the sub-shell-program like
> exit 1

As documented, you need: exit -1

[Billa-User]

thx ! now the sub-shell-program stops with "exit -1"

i am sorry , i miss the exact meaning ...