Path: csiph.com!goblin2!goblin1!goblin.stu.neva.ru!usenet.stanford.edu!not-for-mail
From: Stephane Chazelas <stephane.chazelas@gmail.com>
Newsgroups: gnu.bash.bug
Subject: Re: Design question(s), re: why use of tmp-files or named-pipes(/dev/fd/N) instead of plain pipes?
Date: Mon, 19 Oct 2015 11:04:58 +0100
Lines: 32
Approved: bug-bash@gnu.org
Message-ID: <mailman.624.1445249112.7904.bug-bash@gnu.org>
References: <56218DA5.8030501@tlinx.org> <5622CDC8.2030102@case.edu> <5622EB23.6020700@tlinx.org> <56242D11.3050106@case.edu>
NNTP-Posting-Host: lists.gnu.org
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Linda Walsh <bash@tlinx.org>, bug-bash <bug-bash@gnu.org>
To: Chet Ramey <chet.ramey@case.edu>
Envelope-to: bug-bash@gnu.org
Content-Disposition: inline
In-Reply-To: <56242D11.3050106@case.edu>
User-Agent: Mutt/1.5.21 (2010-09-15)
Precedence: list
Xref: csiph.com gnu.bash.bug:11701

2015-10-18 19:36:49 -0400, Chet Ramey:
> On 10/17/15 8:43 PM, Linda Walsh wrote:
[...]
> >     ??? I've never seen a usage where it expands to a filename and
> > is treated as such.
> 
> Every example of process substitution ever given expands to a filename,
> and the result is treated as a filename.
[...]

Possibly (though unlikely) the confusion comes from the yash
shell where <(...), >(...) are for process *redirection* instead
of process *substitution*.

In yash.

ls >(process-stdout-cmd) 2>(process-stderr-cmd)

Does run those process-stdout-cmd and process-stderr-cmd
commands in background and ls's stdout and stderr are pipes to
their input.

See

http://unix.stackexchange.com/a/208133
http://unix.stackexchange.com/a/86372

for some more details and example usage.

-- 
Stephane