I'm no expert, but my mindset is: If I'm going to be using Unix-style
commands, I'm going to think in a Unix style. And Unix doesn't have
anything corresponding to CCSID, so I'm not going to expect that from
a Unix command. CCSID is an IBM i concept, so I'm going to use an IBM
i command to set the CCSID as needed. For this purpose, I use CHGATR.
(And note that if I'm going whole-hog for the Unix style by logging in
via SSH, and working with the files using only Unix-style tools, it
really doesn't matter what the CCSID is anyway.)
The other approach I've seen is to always ensure that the target file
already exists and has the right CCSID beforehand. I think `touch` is
the QSH command of choice for that purpose.