From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <1c146fcd200ff99cceed6495391ae5be@quanstro.net>
From: erik quanstrom <quanstro@quanstro.net>
Date: Fri, 24 Mar 2006 08:39:48 -0600
To: 9fans@cse.psu.edu
Subject: Re: [9fans] rc question
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Topicbox-Message-UUID: 1f28cf5a-ead1-11e9-9d60-3106f5b1d025

i think you're making this too complicated. you wouldn't want
to use <> with a real file. a fileserver like ndb/dns often ignores seeks.
why use a non-default file descriptor?

why wouldn't this work?

	<> /net/dns { echo google.com ip >[1=0] ; cat}

if you have a fileserver that does not ignore seeks, then this would work

	<> fspath { echo cmd ; syscall seek 0 0 0; cat}

- erik

rog@vitanuova.com writes

|
| geoff:
| > 	<>[4]/net/dns { echo google.com ip  >[1=4]; cat <[0=4] }
|
| if you're expecting the cat to start at offset 0, you'd be mistaken
| (the dup only ups the refcount on the underlying Chan,
| so the seek offsets are shared).
|
| cpu% echo one two three four five six > x
| cpu%
| cpu% <>[4] x {echo xxx >[1=4]; cat <[0=4]}
|  two three four five six
| cpu% cat x
| xxx
|  two three four five six
| cpu%
|
| you can probably do it with dd. the "read" command in inferno
| is a more direct analog of the system call, and allows a "-o" flag
| to set the offset. i found it quite often useful to know that one
| is getting one and only one (p)read.