It might help if you would let us know exactly what output you are seeing. Bev in TX

On Mon, Jun 11, 2018 at 08:40:59AM -0400, bruce wrote: Can you clarify what exactly you would like the output to be? Is there any reason to not just duplicate the line with "echo" in front of it? -- Matthew Miller <mattdm(a)fedoraproject.org&gt; Fedora Project Leader

On 06/11/2018 05:40 AM, bruce wrote: Bash doesn't have a mode in which it echos a command before expansion and substitution occur, as far as I know.  The command you're seeing, with the wildcard expanded, is the command that bash is actually executing.

On 06/11/2018 02:31 PM, Bev in TX wrote: Yeah, that also *kinda* works.  That'll print lines as they're read, not as they're executed, so you *do* get the lines before expansion, but you won't get quiet what is expected (I think) if those lines appear in a loop or a function.

On 11Jun2018 14:02, Gordon Messmer <gordon.messmer(a)gmail.com&gt; wrote: The shell has long had a -v option which echoes commands as they are read. Note that this is in the parse step, so: - you see the command before expansion - you see whole compond commands _once_ as they are encountered, not as they are executed Personally I like "set -x". I've even got a tiny line script called "set-x" which goes: #!/bin/sh set -x exec "$@" and many scripts which do variants on: trace= [ -t 2 ] && trace=set-x # at least during development ... $trace important command here ... ... For proper utility scripts I tend to add a -x option to explicitly turn it on, and -n to set it to "eecho", another tiny script which goes: #!/bin/sh echo "$*" >&2 which gets you do-nothing tracing mode, 90% of what a -n needs. Cheers, Cameron Simpson <cs(a)cskk.id.au&gt;

On 16Jun2018 10:24, Marek Howard <marekhwd(a)gmail.com&gt; wrote: All Bourne shells honour the -n "set" option. But this isn't a shell switch, it is a switch for scripts, putting the script into "dry run" (or "no action", far more memorable) mode. Same meaning for the switch though; this is very common for shell commands. So something like: doit=1 trace= ##[ -t 2 ] && trace=set-x # for debugging # parse options (incomplete boilerplate code) while [ $# -gt 0 ] do case $1 in ... other options ... -x) trace=set-x doit=1 # trace execution -n) trace=eecho doit= ;; # no action -q) trace= doit=1 ;; # quiet - silent execution ... esac shift done ... $trace some significant command here ... if [ $doit ] then ... some action ... fi So the script uses $trace or $doit depending what the use is. For a lot of things just setting trace to "set-x" (recites commands and runs them), "eecho" (recites commands to stderr and does not run them) or the empty string (runs commands without reciting them) covers things. But something you need to be implementing "dry run" logic yourself, so $trace isn't enough; you also want a flag you can test. And for that I use $doit, set to "1" to do stuff and the empty string to not do stuff. This makes for an easy to read idiom: if [ $doit ] then ... do stuff ... fi because the test "[ 1 ]" is true and "[ ]" is false. Cheers, Cameron Simpson <cs(a)cskk.id.au&gt;