Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.unix.shell > #4452

Re: n>&m

Show all headers | View raw

>>>>> Andre Aitken <andre.aitken.a@gmail.com> writes:

 > In a shell

 > program n>&m

 > Does this mean, send the output that "program" is writing to
 > descriptor n to the *current* location of calling shell's descriptor
 > m?

	Not quite.  Consider, e. g.:

$ LC_ALL=C cmp -- /dev/null /dev/zero 5>&1 2>&5 1>/dev/null 
cmp: EOF on /dev/null
$ 

	There, 5>&1 redirects command's &5 to Shell's &1 (e. g., the
	terminal), then 1>&5 redirects command's &1 to the
	/just redirected/ &5, and thus to Shell's &1.

	After that, 1>/dev/null allows us to discard command's &1, so
	that we could see it clearly that the gratuitous error message
	we've asked for was indeed propagated to the Shell's &1.

 > This is what I gather from the way examples are ordered (that is,
 > they always put 2>&1 after > file , and some explicitly say never to
 > switch order).

	Note that although against the convention, 2>FILE >&2 does
	effectively the same.

-- 
FSF associate member #7257

Back to comp.unix.shell | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

n>&m Andre Aitken <andre.aitken.a@gmail.com> - 2012-03-13 22:16 -0700
  Re: n>&m Ivan Shmakov <oneingray@gmail.com> - 2012-03-14 13:14 +0700
    Re: n>&m Andre Aitken <andre.aitken.a@gmail.com> - 2012-03-22 19:43 -0700
      Re: n>&m Barry Margolin <barmar@alum.mit.edu> - 2012-03-23 02:21 -0500

csiph-web