From mboxrd@z Thu Jan 1 00:00:00 1970 From: Felix Kaiser Subject: Re: [PATCH] ip link/addr show: add empty line between interfaces Date: Mon, 07 Sep 2015 17:38:48 +0200 Message-ID: <1441639971-sup-8473@zero> References: <1441537269-13899-1-git-send-email-felix.kaiser@fxkr.net> <20150907144609.3ec9f4d0@griffin> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Cc: netdev@vger.kernel.org To: Jiri Benc Return-path: Received: from mail-wi0-f179.google.com ([209.85.212.179]:37126 "EHLO mail-wi0-f179.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753955AbbIGPiw (ORCPT ); Mon, 7 Sep 2015 11:38:52 -0400 Received: by wicfx3 with SMTP id fx3so87771217wic.0 for ; Mon, 07 Sep 2015 08:38:50 -0700 (PDT) In-reply-to: <20150907144609.3ec9f4d0@griffin> Sender: netdev-owner@vger.kernel.org List-ID: Excerpts from Jiri Benc's message of 2015-09-07 14:46:09 +0200: > Or makes the readability worse by requiring more scrolling, depending > on number of interfaces you have and your taste. Of course it's slightly subjective, but I think it improves readability quite a bit *especially* on servers with many interfaces. Without the additional lines, you can obviously fit more on one screen, but I don't think it's much, and I think that once you have a screenful of text, adding some whitespace makes it much, much easier to focus on the individual sections. > Many users prefer the current format. I've met users who prefer ifconfigs format (and have literally told me that they find ips hard to read), and I've met users who use a shell alias to hack newlines in (which is the motivation for this patch), and everyone that I've shown side-by-side comparisons has been supportive. > Also, as sad as it is, there are scripts out there that parse ip output > and I have no doubts this would break them. I'm aware. My opinion is that if someone parses that, they deserve the opportunity to fix their code. At the very least they could have used -oneline! The default output format is clearly intended for humans; it seems like no attention at all has been given to parseability. And that's as it should be - since, because it requires the fewest keystrokes, that's the format that nearly all interactive users are going to see, and the output should be optimized for them. If absolute compatibility even for badly written scripts that misuse the tool had to be maintained then that'd mean that no improvements could ever be made, and that'd be sad. Felix