Re: Add SHELL_EXIT_CODE to psql - Mailing list pgsql-hackers

From Justin Pryzby
Subject Re: Add SHELL_EXIT_CODE to psql
Date
Msg-id ZAkzhE9GpFzysdcJ@telsasoft.com
Whole thread Raw
In response to Re: Add SHELL_EXIT_CODE to psql  (Corey Huinker <corey.huinker@gmail.com>)
Responses Re: Add SHELL_EXIT_CODE to psql
List pgsql-hackers
On Wed, Feb 22, 2023 at 03:04:33PM -0500, Corey Huinker wrote:
> +
> +/*
> + * Return the POSIX exit code (0 to 255) that corresponds to the argument.
> + * The argument is a return code returned by wait(2) or waitpid(2), which
> + * also applies to pclose(3) and system(3).
> + */
> +int
> +wait_result_to_exit_code(int exit_status)
> +{
> +    if (WIFEXITED(exit_status))
> +        return WEXITSTATUS(exit_status);
> +    if (WIFSIGNALED(exit_status))
> +        return WTERMSIG(exit_status);
> +    return 0;
> +}

This fails to distinguish between exiting with (say) code 1 and being
killed by signal 1.

> -            if (ferror(fd))
> +            exit_code = ferror(fd);
> +            if (exit_code)

And this adds even more ambiguity with internal library/system calls, as
Tom said.

> +        if (close_exit_code && !exit_code)
> +        {
> +            error = true;
> +            exit_code = close_exit_code;
> +            if (close_exit_code == -1)
> +                pg_log_error("%s: %m", cmd);

I think if an error ocurrs in pclose(), then it should *always* be
reported.  Knowing that we somehow failed while running the command is
more important than knowing how the command ran when we had a failure
while running it.

Note that for some tools, a nonzero exit code can be normal.  Like diff
and grep.

The exit status is one byte.  I think you should define the status
variable along the lines of:

 - 0 if successful; or,
 - a positive number 1..255 indicating its exit status. or,
 - a negative number N indicating it was terminated by signal -N; or,
 - 256 if an internal error occurred (like pclose/ferror);

See bash(1).  This would be a good behavior to start with, since it
ought to be familiar to everyone, and if it's good enough to write/run
shell scripts in, then it's got to be good enough for psql to run a
single command in.  I'm not sure why the shell uses 126-127 specially,
though..

EXIT STATUS
       The exit status of an executed command is the value returned by the waitpid system call or  equivalent
function.  Exit  statuses
 
       fall  between  0  and  255,  though,  as  explained below, the shell may use values above 125 specially.  Exit
statusesfrom shell
 
       builtins and compound commands are also limited to this range.  Under certain circumstances, the shell will use
specialvalues  to
 
       indicate specific failure modes.

       For  the shell's purposes, a command which exits with a zero exit status has succeeded.  An exit status of zero
indicatessuccess.
 
       A non-zero exit status indicates failure.  When a command terminates on a fatal signal N, bash uses the value of
128+Nas the exit
 
       status.

       If a command is not found, the child process created to execute it returns a status of 127.  If a command is
foundbut is not exe‐
 
       cutable, the return status is 126.

       If a command fails because of an error during expansion or redirection, the exit status is greater than zero.

-- 
Justin



pgsql-hackers by date:

Previous
From: Michael Paquier
Date:
Subject: Re: Add pg_walinspect function with block info columns
Next
From: Michael Paquier
Date:
Subject: Re: [PATCH] Align GSS and TLS error handling in PQconnectPoll()