Vrftsjtryk

When I write the code the way below, I am able to run several commands after the else statement:

if [ "$?" -eq 0 ]

    then              

        echo "OK"

    else 

        echo "NOK"

        exit 1

fi

However, when I use another syntax I am unable to union 2 commands after the OR:

[ "$?" -eq 0 ] && echo "OK" || (echo "NOK" >&2 ; exit 1)

In my use-case I have a complex script based on "$?" == 0, so I'm looking for a way to abort (additionally to echoing the message) when it is not true.

The pair of ( ) spawns a subshell, defeating the goal of exiting the whole script with the exit command inside.

Just replace the ( ) with { } (and adjusted syntax because { } are not automatical delimiters but more treated like commands: a space after { and last command inside must end with some terminator:; fits): this will run the chain of commands inside in the same shell, thus exit will affect this shell.

[ "$?" -eq 0 ] && echo "OK" || { echo "NOK" >&2; exit 1;}

UPDATE: @D.BenKnoble commented that should echo fail, the behaviour won't be like the former if ...; then ... else ... fi construct. So the first echo's exit code has to be "escaped" with a noop : command (which being built-in can't fail).

[ "$?" -eq 0 ] && { echo "OK"; :;} || { echo "NOK" >&2; exit 1;}

references:

POSIX:

Grouping Commands

The format for grouping commands is as follows:

(compound-list)

     Execute compound-list in a subshell environment; see Shell

     Execution Environment. Variable assignments and built-in commands 

     that affect the environment shall not remain in effect

     after the list finishes.

[...]

{ compound-list;}

     Execute compound-list in the current process environment. The

     semicolon shown here is an example of a control operator delimiting

     the } reserved word. Other delimiters are possible, as shown

     in Shell Grammar; a <newline> is frequently used.

dash manpage,bash manpage,...

One way would be to use the following script line:

[ "$?" -eq 0 ] && echo "OK" || exit $(echo "NOK" >&2 ; echo 1)

That serves the purpose of reporting an error and returning an error code, whilst being silent if the "OK" path is chosen.

The problem is that you are executing the other branch in a subshell (that is what the parentheses mead - e.g. in bash(1) man page look for compound commands) - thus the exit quits the subshell and not the shell executing the script. To see it more graphically, put a long enough sleep in both versions of your code, and run ps faux (for coreutils version of ps) or something like pstree to see the process ancestry relationships.

Now, how to "fix" it:

1. there's nothing wrong about if ... then / else / fi. Especially for longer blocks it makes the code more readable.



2. 
   

   if you want a one liner, then you can do

   
   
   

   [ "$?" -eq 0 ] && echo "OK" || echo "NOK" >&2 && exit 1
   
   

   
   
   

   which relies on echo exiting gracefully, or use a group command

   
   
   

   [ "$?" -eq 0 ] && echo "OK" || { echo "NOK" >&2 ; exit 1; }
   
   

   
   
   

   Check for general availability in your shell and for any peculiarities though - e.g. it should be terminated by a newline or a semicolon.

   
   


3. 
   

   most comprehensive - use a signal/event handler (bash example, syntax for other shells might differ):

   
   
   

   #!/bin/bash
   
   
   
   exit_handler () {
   
       echo "error happened, exiting"
   
       ... fix things, that may be half baked ...
   
   }
   
   
   
   trap exit_handler EXIT
   
   
   
   ... do stuff ...
   
   

   
   
   

   This gives you the opportunity to:

   
   
   
   
   
   • use set -e (set -o errexit) which is often considered dangerous - the EXIT handler can do things like removing any temporary files you one doesn't want to leave behind
   
   
   • have the cleanup code in one place only
   
   
   • handle other signals (those sent by e.g. kill) as well