<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
<br>
Matthew D. Fuller wrote:
<blockquote
cite="mid:cbd4a981fd3ffff9493d0f3ebbe3955c,20090616133027.GH11838@over-yonder.net"
type="cite">
<pre wrap="">On Tue, Jun 16, 2009 at 01:31:20PM +0200 I heard the voice of
David Ingamells, and lo! it spake thus:
</pre>
<blockquote type="cite">
<pre wrap="">The fact that some other less well informed projects do it doesn't
make it an OK thing to do.
</pre>
</blockquote>
<pre wrap=""><!---->
You claim "less well informed". I say rather that it's a standard,
very common, and highly useful idiom. That doesn't mean it's the only
choice, or even necessarily the best in any case, but to claim it as
some sort of original sin, or unprecedented perversity of bzr, is
laying it out pretty thick. </pre>
</blockquote>
Where did I say it was some form of "original sin ..."? What I am
saying is that STDERR was though up by some very great men in the
history of Unix for an excellent - and still very valid - reason. By
such lack of respect for that reason in some command-line tools the
original purpose is effectively being lost and CLI users are the
victims of the developer's laziness and lack of respect. The fact that
one or 2 minor tools use it doesn't make it "very common". If svn or
CVS did it you might have some basis for your argument, but
monotone??? I have raised this here because I want bzr to be and
remain mainstream, and by such lack of respect for the correct use for
STDERR bzr risks losing the credibility it needs to remain mainstream.<br>
<blockquote
cite="mid:cbd4a981fd3ffff9493d0f3ebbe3955c,20090616133027.GH11838@over-yonder.net"
type="cite">
<pre wrap="">Any soi-disant "unix power-user" who
hasn't seen it before has been leading a very sheltered life.
</pre>
</blockquote>
Oh! I've seen - and suffered from - it before and every time it annoys
me. It also annoys me when people try to claim I said things that I
didn't, Matthew D. Fuller. Are you a soi-disant new-age thinker who
knows better than those men who invented STDERR ( along with UNIX, C,
...)?<br>
<blockquote
cite="mid:cbd4a981fd3ffff9493d0f3ebbe3955c,20090616133027.GH11838@over-yonder.net"
type="cite">
<pre wrap="">
</pre>
<blockquote type="cite">
<pre wrap="">Respect the meaning of the name: "Standard ERROR", not "any other
stuff you might want to output but can't think of the proper way to
do it".
</pre>
</blockquote>
<pre wrap=""><!---->
Respect the meaning of the name: TELETYPEWRITER, not "some random way
of entering input and getting output".
Things are rarely just the literal meaning of their name.
</pre>
</blockquote>
There you have shot yourself in the foot - a CLI program that uses the
correct UNIX paradigm for I/O will still work correctly with a TTY - as
far as the program is concerned it won't know the difference - and
won't need to care. That is one of the beauties of UNIX (and by
inheritance Linux). I don't think your diluted approach goes so far as
allowing output to STDIN or input from STDERR does it?<br>
</body>
</html>