From: Jim Meyering <jim@meyering.net>
Date: Tue, 05 Aug 2008 15:44:03 GMT
Subject: Re: [PATCH] git-cvsimport.perl: Print "UNKNOWN LINE..." on stderr, not stdout.
Message-ID: <87tzdzo4ak.fsf@rho.meyering.net>
In-Reply-To: <20080805152836.GB21901@sigill.intra.peff.net>

Jeff King <peff@peff.net> wrote:
> On Tue, Aug 05, 2008 at 04:54:42PM +0200, Jim Meyering wrote:
>
>> I'm used to filtering git-cvsimport's noisy stderr, but this
>> diagnostic appears on stdout.  Looks like an oversight.
>> Now that I'm using cvsps-2.2b1, I see tons of these.
>
> There are a ton of things that go to stdout:
>
>   $ perl -ne '/print (\S+)/ && print "$1\n" git-cvsimport.perl |
>     grep '^#' | wc -l
>   18
>
> though many are only activated via "-v". Maybe it is worth putting all
> of them to stderr? I really don't see why cvsimport should ever produce
> any output on stdout.

A quick scan (post-patch)

    $ grep 'print ' git-cvsimport.perl|grep -vE 'STDERR|opt_v'
    print $f "$_=$conv_author_name{$_} <$conv_author_email{$_}>\n";
    print $fh $line;
    print $fh $buf;
    print $cvspsfh $_;
    print $fh
    print "SKIPPING $fn v $rev\n";

suggests that if you don't count the "SKIPPING..." diagnostic you'd get
for each "-S REGEXP"-skipped name, the "UNKNOWN LINE" diagnostic is the
only one that goes to stdout but that is not protected by an $opt_v guard.