Zsh Mailing List Archive Messages sorted by: Reverse Date, Date, Thread, Author

Re: Bug: reading from tty inside process substitution

X-seq: zsh-workers 52297
From: Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>
To: zsh-workers@xxxxxxx
Subject: Re: Bug: reading from tty inside process substitution
Date: Sun, 12 Nov 2023 13:50:58 -0800
Archived-at: <https://zsh.org/workers/52297>
In-reply-to: <CAH+w=7ZztRXt7R9fDK=yb1=fcvjwjiUGo98j73BgBG7NzWE+iw@mail.gmail.com>
List-id: <zsh-workers.zsh.org>
References: <CAA=-s3xCf4wx826U8jEjO_j5-1kcTi8v2=VZPiabKMo-meKq0Q@mail.gmail.com> <CAH+w=7ZztRXt7R9fDK=yb1=fcvjwjiUGo98j73BgBG7NzWE+iw@mail.gmail.com>

On Sun, Nov 12, 2023 at 10:09 AM Bart Schaefer
<schaefer@xxxxxxxxxxxxxxxx> wrote:
>
> I doubt there's any easy way to change this without breaking something else.

This is even worse:

read foo < <(read bar; echo $bar) </dev/tty

With MULTIOS, that command becomes un-killable, because "read foo"
sees a pipeline consisting of a catenation of both the process
substitution and the tty, but the process substitution is perpetually
TTIN'd and the interactive shell is ignoring interrupts.  Without
MULTIOS, the command substitution simply becomes a stopped zombie and
"read foo" sees only /dev/tty.  Without the </dev/tty, it gets stuck
but interrupt is possible, leaving behind a zombie.

References:
- Bug: reading from tty inside process substitution
  - From: Mark J. Reed
- Re: Bug: reading from tty inside process substitution
  - From: Bart Schaefer

Messages sorted by: Reverse Date, Date, Thread, Author