Re: any plans for command substitution that preserves trailing newlines?

[Greg Wooledge]

On Tue, Jun 01, 2021 at 11:46:38AM +0800, konsolebox wrote:
> On Mon, May 31, 2021 at 11:46 PM Christoph Anton Mitterer
> <calestyo@scientia.net> wrote:
> > I wondered whether there were any plans to implement proper support for
> > command substitutions that preserves trailing newlines?
> 
> As an alternative, you can use `IFS= read -rd '' var < <(cmd)`.

Not picking on this answer in particular, but... I'm skeptical of the
original question.  The OP is concerned about some theoretical failure
of the workaround for trailing newlines in command substitutions when
using a variable-width character encoding and receiving malformed data
as input.

In a freaking *shell script*.

Is the OP actually trying to store *binary* data in a shell variable?
This is doomed to fail because you can't handle NUL bytes.  I'd worry
about that long before I'd worry about malformed strings in Asian
character encodings interfering with one's attempts to read binary data
in a shell script.

The correct response to "I'm trying to read binary data" is "don't do it
in a shell script".

Is the OP actually concerned that the user may supply input that is a
malformed string in their own locale?  Just let it fail.  The user is
responsible for their own self-imposed problems in that case.