Re: [Help-bash] Readig multiple null-separated values

[František Kučera]

Dne 29. 07. 19 v 10:01 David napsal(a):
> On Sat, 27 Jul 2019 at 23:56, František Kučera <address@hidden> wrote:
> > in my Bash pipelines I sometimes pass data separated by null byte \0 and
> > it is not only lists (like in find -print0) but
> > matrices/tables/relations with fixed number of columns/attributes. The
> > \0 is used to both separate values and records. The advantage is that
> > the only forbidden character in values is the \0 and there are no
> > collisions with text separators like ,;:\t\n etc. and no need for
> > escaping or quoting. So it is a safe and simple way to pass structured
> > data between processes.
> When you use the word "data" above, do you mean ASCII text,
> or something else?

It is usually UTF-8 text and it could come from various sources like 
databases, XML, filesystem (file names, extended attributes or file 
content).

> If ASCII text, are you aware of the ASCII delimiter characters?
> See https://en.wikipedia.org/wiki/Delimiter#ASCII_delimited_text

I know about them, but as already mentioned by others, such characters 
might occur inside the value. And the only character that could not 
occur in a filename or C string is \0. However it is a corner-case and 
ASCII delimiters would mostly work.

Franta