DMARC-Filter: OpenDMARC Filter v1.4.2 delorie.com 56G1AROo2423668
DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 646B63858D32
MIME-Version: 1.0
Date: Tue, 15 Jul 2025 18:09:36 -0700
Message-ID: <CAAHpriMfK_qTRU=X7X+TAvGtaeL=vRcDXJ1gWhw=0YgKrgfSog@mail.gmail.com>
Subject: Re: Misleading (irritating) output from cp
To: The Cygwin Mailing List <cygwin AT cygwin DOT com>
Cc: Keith Thompson <Keith DOT S DOT Thompson AT gmail DOT com>
From: Keith Thompson via Cygwin <cygwin AT cygwin DOT com>
Reply-To: Keith Thompson <Keith DOT S DOT Thompson AT gmail DOT com>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Sender: "Cygwin" <cygwin-bounces~archive-cygwin=delorie DOT com AT cygwin DOT com>

Fergus Daly wrote:
> If you try
> $ cp -v existingfile existingdirectory
> you get feedback
> 'existingfile' -> 'existingdirectory/existingfile'
> which is just fine. (Omitting the switch -v silences the feedback: also fine.)
> Whereas: if you try
> $ cp -v existingfile nonexistingdirectory
> you get feedback (two lines)
> 'existingfile' -> 'nonexistingdirectory'
> /bin/cp: cannot create regular file: No such file or directory
> of which the 2nd line is useful advice but the 1st line is just a bit misleading / irritating / .. difficult to ignore .. whatever.
> (Try the same / similar thing with any other command e.g. rm e.g. md5sum and you just get the 2nd line advice.)
> Try
> $ cp -v existingfile nonexistingdirectory 2> /dev/null
> then you suppress line 2 as required; but you still get line 1 which now is _very_ misleading / _very_ irritating!
> This "feature" is matched in Linux.
> Maybe a small thing but not to me. (Where it occurs it rather screws up the smooth evolution of session logs.)
> Does anybody know of a way of suppressing the 1st line feedback?
> Thank you!

Cygwin's "cp" command is provided by the GNU Coreutils package,
currently version 9.0.

The two-line output you describe doesn't occur with the command
"cp -v existingfile nonexistingdirectory".

Since "nonexistingdirectory" doesn't exist, the cp command has no way
to know whether it's expected to be a directory or a file.  It's not
a nonexisting directory; it's just a nonexisting *thing*.  If the
current directory is writable, that command will copy "existingfile"
to a new file named "nonexistingdirectory".

But you would get that output with:
"cp -v existingfile nonexistingdirectory/"
where the trailing / requires the target to be a directory.

The `info cp` description of the -v / --verbose option is "Print the
name of each file before copying it" (which is a little misleading;
it actually prints both the source and the target).  The `man cp`
description is "explain what is being done", which is vague enough
to be accurate.

Experiment shows that it actually prints (to stdout) the source and
target of the copy that it's *about* to perform, and then attempts
to perform the copy.  If an error occurs, a message is then printed
to stderr.

Apparently the FreeBSD version of "cp" doesn't print the verbose
message if the copy operation fails.

Both behaviors strike me as reasonable.  I don't think there's any
way (other than rebuilding coreutils from source or ) to get "cp"
to behave the way you want.

-- 
Problem reports:      https://cygwin.com/problems.html
FAQ:                  https://cygwin.com/faq/
Documentation:        https://cygwin.com/docs.html
Unsubscribe info:     https://cygwin.com/ml/#unsubscribe-simple