ifconfig output: ipv4 netmask format

[Garrett Cooper]

On Fri, Apr 8, 2011 at 9:54 PM, J. Hellenthal <jhell at dataix.net> wrote:
> On Fri, Apr 08, 2011 at 07:36:45PM +0400, Sergey Vinogradov wrote:
>>On 08.04.2011 19:23, Mike Oliver wrote:
>>>On Fri, Apr 8, 2011 at 08:08, Sergey Vinogradov<boogie at lazybytes.org>  wrote:
>>>>Hi, hackers.
>>>>I have a question: why ipv4 netmask is displayed by ifconfig in hex format?
>>>>Isn't dot-decimal notation more human-readable? Will the attached patch
>>>>break something in the very bad way?
>>>
>>>Who's using IPv4 anymore?  ;-)
>>Long live IPv4! :)
>>
>>>Seriously though, if you give a small amount of time to learning the
>>>hex ->  binary translations then you would see how convenient it is to
>>>use hex rather than decimal when representing what are ultimately
>>>binary numbers.
>>>
>>>See this blog entry by Jeff Doyle...
>>>
>>>http://www.networkworld.com/community/blog/how-are-your-hexadecimal-skills
>>The article is great, but dot-decimal notation is de-facto standard
>>for stand-alone network mask representation. Like CIDR is standard for
>>IP blocks represenation. That's the reason I've started this thread.
>>And despite the greatness of the article you've mentioned, I think
>>it's a bad itea to hardcode its URL into ifconfig's output. You know,
>>for every single user reading it, and choosing the "way of hex" ;)
>>
>
> This is the year 2011 right ? when are we going to support new users
> rather than supporting old outdated washed up "scripts" ?
>
> I for one am for this change, given that there are lots of users from
> the PC-BSD community that do not read hexadecimal, octal and other such
> forms like a programmer does.
>
> And just because the change can be made does not mean that a
> compatibility shim cannot be put into place that restores the old
> functionality.
>
>
> It is time to stop living in the past and start thinking about the
> future. These types of things are what causes forks of projects to
> happen ultimately yielding in less contributors and developers. I for
> one hate to see things like that happen.

I understand your pain and while I've been on your side of the fence
several times in the past dealing with different things of this sort,
having to maintain backwards compatibility is a painful reality of
past design choices or mistakes. pkg_install and sysinstall are one of
many examples, but there are of course other areas as well.

Although I see the value of your and Sergey's argument, the problem is
that it may cause unexpected breakage for other third parties that
depend on a particular behavior in FreeBSD as Bjoern and others have
suggested; I have a script at least that does properly parse out the
hex output in order to set IPs properly with ipmitool, and I would be
perturbed to have to hack around this further in my script. Mind you,
this script is in my workspace, but having to track minor output
changes like this across versions of FreeBSD, or interface changes
like the semi-gratuitous removal of /dev/<device>c in 8.x still had to
be worked out on many branches my group maintains at $WORK.

Getting back on track, I think that more 'user friendly' variants of
FreeBSD, e.g. PCBSD, etc, could and should diverge with this one line
patch as they're not mainline FreeBSD and the change is trivial for
less hex inclined users as you suggest.

Granted, this is just one of many bikeshed topics. Funny how many
lines of debate have been written and how many feathers have been
unnecessarily ruffled over the output of one printf.

Thanks,
-Garrett