Re: coproc issue

[alex xmb ratchev]

On Sat, Jun 17, 2023, 16:42 Greg Wooledge <greg@wooledge.org> wrote:

> On Sat, Jun 17, 2023 at 03:15:48PM +0200, alex xmb ratchev wrote:
> > whats the problem here
> > some timing issues ?
> >
> > ~ $ coproc a { echo foo ; read foo ; echo $foo ; } ; cat <&$a ; echo bar
> > >&${a[1]}
> > [2] 13957
> > foo
> >
> > ^C
>
> It's not clear what you expected.  You've launched a coproc, which is
> a background process.  This coproc writes "foo" to stdout, and then
> performs a blocking read, which will wait for a line to be sent to its
> stdin.
>
> After launching the coproc, you run a "cat" process which reads from
> the coproc's output and relays it to the screen.  Thus, you see the
> "foo" which the coproc wrote, and is subsequently written by cat, to
> the terminal.
>
> However, cat doesn't terminate until an EOF occurs.  Your coproc is
> still "running" (waiting on a blocking read), and therefore it has
> not closed its output.  This means the pipe that cat is reading from
> is still there, still ready to deliver more data.  It has not been
> closed, and therefore there's no EOF, and therefore cat does not exit.
>

ah thank u missed cat still on .. obvious error there

So, any commands that occur *after* cat do not get executed.
>
> If you want to interact with a coproc, you need to use commands that
> don't wait for EOF.  Use read instead of cat, for example.
>

i had the impression echo somehow eof s
cant send eof earily ? ah , close stdout/err in coproc .. maybe ..

read is alternative
ye

thxx , greets

>