Re: read and env variables + POSIX => SEGFAULT

[Linda Walsh]

(Cc: Chet Ramey... forgot to send it to list...oop)


Chet Ramey wrote:
> On 10/10/15 11:01 PM, Linda Walsh wrote:
>
> > > a= read a <<< x;echo $?
> > 0
> > > declare -p a
> > declare -- a="x"
> > #  the manpage claims "one line is read from [the input], and the result
> > #  is split by words and assigns 1st word to 1st var and so forth, but
> > #  apparently the reading of 1 line is optional -- though this is 
> > consistent
> > #  with the fact that read can be told to read some number of characters
> > and #  return when the limit is reached.  So technically, read doesn't
> > "read one line",
> > #  but read whatever is on 'input' up to 1 line.  (DOC clarification?)
>
> This is terribly wrong.
>
> The command in question is `a= read a <<< x'.
>
> The here-string construct takes the following word and, like a here
> document, makes it the standard input to the command.  The standard
> input is then a file consisting of a single line: x\n.
>
> It's basically shorthand for
>
> read a <<EOF
> x
> EOF
>
> So, `read' reads the single line from its standard input and assigns it
> to the variable `a'.
----
I wasn't sure if it put the "\n" at the end in a 1-line example.

Does it also use a tmp file and use process-substitution, or is
that only when parens are present?  I.e.

read a < <( echo x)

I'm under the impression, uses a tmp file.

does the read a <<< x
also use a tmp file?

I.e. is
> readarray -t a < <( echo -e 'x\ny\n')
> declare -p a
declare -a a='([0]="x" [1]="y")'

implemented the same way as
> a=(x y)
> b=$(printf "%s\n" ${a[@]})
> readarray -t ar <<< "${b[@]}"
> declare -p a
declare -a a='([0]="x" [1]="y")'



>