After a rather thorough RTFM session, I tried:

 [1+0]~$ mkdir test-parallel
 [2+0]~$ cat > foobarbaz
foo
bar
baz
 [3+0]~$ cat foobarbaz | parallel --gnu --pipe dd of=test-parallel/out'{}' ::: 1 2 3
0+0 records in
0+0 records out
0 bytes (0 B) copied, 0.00162787 s, 0.0 kB/s
0+1 records in
0+1 records out
6 bytes (6 B) copied, 0.00293682 s, 2.0 kB/s
0+0 records in
0+0 records out
0 bytes (0 B) copied, 0.000407992 s, 0.0 kB/s
0+0 records in
0+0 records out
0 bytes (0 B) copied, 7.072e-06 s, 0.0 kB/s
 [4+0]~$ ls -l test-parallel/
total 0
-rw-r--r-- 1 itz itz 0 Sep  9 20:57 out
 [5+0]~$ cat test-parallel/out
 [6+0}~$

I expected rather something along the lines of:

 [4+0]~$ ls -l test-parallel/
total 0
-rw-r--r-- 1 itz itz 4 Sep  9 20:57 out1
-rw-r--r-- 1 itz itz 4 Sep  9 20:57 out2
-rw-r--r-- 1 itz itz 4 Sep  9 20:57 out3
 [5+0]~$ cat test-parallel/out1
foo
 [6+0]~$ cat test-parallel/out2
bar
 [7+0]~$ cat test-parallel/out3
baz

I am beginning to share Joey's feelings wrt this version of parallel ...

Hi Ole,

as you can see I'm walking through the remaining bugs in the Debian BTS.
I tried to reproduce the issue below and need to admit that I get a
slightly different result with the currently packaged version 20210822
(I just realised that there is even 20211122 which I will upload soon).
I get a file test-parallel/out1 which is a copy of foobarbaz.

I admit I'm not comfortable with parallel (just took over maintenance
since nobody else volunteered) and may be that's simply a
misunderstanding of the docs.  It would be great if you could clarify
this.

Kind regards

      Andreas.

Am Mon, Sep 09, 2013 at 09:04:21PM -0700 schrieb Ian Zimmerman: