bin/99585: [PATCH] Add option to tee(1) to also copy to standard error

[Dan Ponte]

The following reply was made to PR bin/99585; it has been noted by GNATS.

From: Dan Ponte <dcp1990 at neptune.atopia.net>
To: Bruce Evans <bde at zeta.org.au>
Cc: FreeBSD-gnats-submit at freebsd.org, freebsd-bugs at freebsd.org
Subject: Re: bin/99585: [PATCH] Add option to tee(1) to also copy to standard error
Date: Thu, 29 Jun 2006 11:54:57 -0400

 On Thu, Jun 29, 2006 at 10:35:57PM +1000, Bruce Evans <bde at zeta.org.au> was witnessed plotting the following conspiracy:
 > On Wed, 28 Jun 2006, Dan Ponte wrote:
 > 
 > >>Description:
 > >	I recently found myself in a situation where having tee(1) copy to 
 > >	standard error and standard output at the same time would be useful, 
 > >	so I implemented it.
 > >>How-To-Repeat:
 > >	Find yourself in said situation.
 > >>Fix:
 > >	Diff below.
 > 
 > Why not use a standard utility (tee itself in this case)?
 > 
 >   genoutput | tee /dev/stderr | tee foo
 > 
 > copies the stdout of `genoutput' to stderr and to file foo.  This can be
 > generalized to any number of copies on any numbers of file descriptors
 > or files.
 > 
 > The main problem with the utility method is is that the plumbing is
 > very confusing even for only one extra pipeline stage.  /dev/stderr
 > in the above refers to the initial stderr so it isn't affected by
 > redirection unless the pipeline stages are run in subshells.  Running
 > some of the stages in subshells makes the plumbing even more confusing.
 > 
 > Bruce
 > 
 Hrm...this is a good idea. I never thought about that.
 -Dan
 -- 
 Dan Ponte
 http://www.theamigan.net/ -- ICQ: 175527699
 Jabber: dan at styx.theamigan.net -- AIM: amiganWS/amigan940
 To get something done, a committee should consist of no more than three
 persons, two of them absent.