Zsh Mailing List Archive
Messages sorted by:
Reverse Date,
Date,
Thread,
Author
Re: non-interactive set -m
- X-seq: zsh-workers 27144
- From: Eric Blake <ebb9@xxxxxxx>
- To: zsh-workers@xxxxxxxxxx
- Subject: Re: non-interactive set -m
- Date: Mon, 13 Jul 2009 01:48:15 +0000 (UTC)
- Mailing-list: contact zsh-workers-help@xxxxxxxxxx; run by ezmlm
- References: <200907121935.n6CJZ5K0018519@xxxxxxxxxxxxxxxxxxx> <090712141900.ZM14558@xxxxxxxxxxxxxxxxxxxxxx>
- Sender: news <news@xxxxxxxxxxxxx>
Bart Schaefer <schaefer <at> brasslantern.com> writes:
> } > Note that in pdksh the job in the subshell is added to the table of
> } > jobs inherited from the parent shell (testing with pdksh 5.2.14).
> }
> } Yuk. So you've got a real job you can manipulate, and one phantom job
> } from the parent shell you can't and which is presumably fixed like that
> } for eternity.
>
> No, actually, you have two jobs neither of which you can manipulate.
> PDKSH doesn't do job control in subshells. It turns off -m when the
> subshell starts, and silently ignores it if you manually turn it back
> on again.
My opinion is that pdksh is somewhat buggy when it comes to job handling, and
that you are better off trying to emulate David Korn's ksh93 than the buggy
pdksh; other better examples are dash and bash. I already quoted the parts of
POSIX that mention that subshells inherit the same options as the parent (so the
pdksh behavior of silently disabling -m in a subshell violates that rule).
--
Eric Blake
Messages sorted by:
Reverse Date,
Date,
Thread,
Author