From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <002b01c47a47$0f5d5ca0$f5fb7d50@SOMA>
From: "boyd, rounin" <boyd@insultant.net>
To: "Fans of the OS Plan 9 from Bell Labs" <9fans@cse.psu.edu>
References: <7446ba65534f2ff0fd37373855b2b932@9netics.com>
Subject: Re: [9fans] rc mystery
Date: Wed,  4 Aug 2004 19:18:21 +0200
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Topicbox-Message-UUID: d1043506-eacd-11e9-9e20-41e7f4b1d025

order of execution.

> term% echo `{date -n && sleep 2} `{sleep 2 && date -n}
> 1091636548 1091636548

those two are run in parallel and their outputs are handed to echo
as arguments (fork being fast, as it was always meant to be :)

> term% cat <{date -n && sleep 2} <{sleep 2 && date -n}
> 1091637155
> 1091637158

no, that gives you:

    cat /fd/m /fd/n

and the cat can't read /fd/n until /fd/m is closed.  both fd's
are pipes to the subshells, so the executional is serialised.

eg:

    brahma% echo <{date -n && sleep 2} <{sleep 2 && date -n}
    /fd/6 /fd/5
    brahma% cat <{date -n && sleep 2} <{sleep 2 && date -n}
    1091639796
    1091639798