I'd like to use socat to redirect the STDIN/STDOUT of the process executing docker attach to a listening TCP socket on my Linux system. root@eve-ng:/opt/unetlab# tty /dev/pts/2 root@eve-ng:/opt/unetlab# root@eve-ng:/opt/unetlab# socat EXEC:'/usr/bin/docker attach afe29573-0030-4a69-8b3e-c04ededa8db7-0-4',pty TCP4-LISTEN:32000 & 1275378 root@eve-ng:/opt/unetlab# The options for socat are the following: - EXEC:,pty to say socat to start a process executing ` allocating a pseudo-terminal pty attached to its STDIN/STDOUT (socat actually opens the master side multiplexer /dev/ptmx and then the fd for the pseudo-terminal pair's master ptm`) - TCP4-LISTEN:32000 to say socat to open a tcp socket listening on port 32000 Indeed, in another terminal you can check: root@eve-ng:~# tty /dev/pts/5 root@eve-ng:~# root@eve-ng:~# ps -ef | grep attach root 1275378 1210230 0 08:43 pts/2 00:00:00 socat EXEC:/usr/bin/docker attach afe29573-0030-4a69-8b3e-c04ededa8db7-0-4,pty TCP4-LISTEN:32000 root 1275379 1275378 0 08:43 pts/2 00:00:00 /usr/bin/docker attach afe29573-0030-4a69-8b3e-c04ededa8db7-0-4 root@eve-ng:~# docker attach actually attaches to a process running inside the container executing the exec executable, by the way you can see pts/1 is allocated to it (the container indeed was created using docker create -it) root@eve-ng:~# ps -ef | grep exec root 1284441 731051 2 08:52 pts/1 00:00:00 exec root 1284583 1222614 0 08:53 pts/5 00:00:00 grep --color=auto exec root@eve-ng:~# The problem I'm facing to is that connecting to the tcp socket 127.0.0.1:32000 from a telnet client, the login process inside the container (exec) doesn't work as expected. It seems the username typed in (carlo) is repeated again not allowing to successfully login into the container. Username: carlo carlo Password: RP/0/RP0/CPU0:Jun 20 09:00:59.739 UTC: exec: %SECURITY-LOGIN-4-AUTHEN_FAILED : Failed authentication attempt by user '' from 'console' on 'con0_RP0_CPU0' User Access Verification Username: Can you help me in understanding why ? Thanks. P.s. digging into it using strace I noticed that socat actually reads from the relevant fd (of type tcp socket) twice instead of just once.

- Works: vsim, sh -c vsim - Doesn't work: echo "vsim" | sh, echo "vsim" | xargs -I {} sh -c "{}" I want to run ModelSim (vsim) with dmenu , which is triggered using xbindkeys. --- ### Details vsim is a executable for ModelSim, installed in /opt/altera/modelsim_ase/bin. When I run it directly, it runs. But when I run it with xargs (eg. from dmenu), it does not work at all - the script itself launches, but probably in the wrong directory or something, I'm really clueless what's wrong. My path (I added newlines for clarity): [ondra@x201 ~]$ echo $PATH /usr/local/sbin: /usr/local/bin: /usr/bin: /usr/lib/jvm/default/bin: /usr/bin/site_perl: /usr/bin/vendor_perl: /usr/bin/core_perl: /opt/altera/quartus/bin: /opt/altera/modelsim_ase/bin: /home/ondra/bin: /home/ondra/.gem/ruby/2.1.0/bin: /opt/altera/University_Program/Monitor_Program/bin/bin Where is vsim? [ondra@x201 ~]$ which vsim /opt/altera/modelsim_ase/bin/vsim Run it with xargs: [ondra@x201 ~]$ echo "vsim" | xargs -I {} sh -c '{} &' [ondra@x201 ~]$ Reading /opt/altera/modelsim_ase/tcl/vsim/pref.tcl # 10.1d # # ^C Run it directly: [ondra@x201 ~]$ vsim Reading /opt/altera/modelsim_ase/tcl/vsim/pref.tcl # --- and modelsim starts fine now --- Any ideas welcome.

We've seen a few posts here lately that use this:

var=$(

in an attempt to read the shell's standard input into a variable. That is however not the correct way to do it on Linux-based systems and Cygwin at least. Why? What are the correct ways?