[rhelv6-beta-list] No tui network setup on server install

[Nico Kadel-Garcia]

On Fri, Jul 2, 2010 at 1:50 AM, Bryan J. Smith <b.j.smith at ieee.org> wrote:
> There have been plenty of alleged "idiots" who have had to deal with
> countless combinations and support issues when it comes to /etc/hosts
> of countless combinations of programs and utilities in the "setup" and
> "anaconda" packages over the years.
>
> And most of the time these alleged "idiots" are just engineers and
> support individuals dealing with "incorrect," "non-standard" and sometimes
> just flat out "broken" assumptions and other coding non-sense in various
> applications and utilities.  It's been a continuing and plaguing issue with
> IPv6 as well, where the alleged "idiots" are trying to make the system IPv6
> ready, and too many other, upstream and ISV components just don't follow
> standards.

Hold: the problem isn't the addition of the hostname itself to the
loopback address space. It's the syntax.

It currently says:

127.0.0.1     localhost.localdomain localhost myhost.example.com myhost

This is awkward to parse and audit. It's useful to have the local host
on the loopback port, true. But that can be edited, and managed, by
this:

127.0.0.1    localhost.localdomain localhost
127.0.0.1    myhost.example.com myhost

This makes for considerably easier to manage syntax. in configuration
management tools. It's way too late to get it changed now,
unfortunately, it's embedded in far too many odd little tools.

> I.e., there are no "idiots" on this, just people trying to make most things
> work, and minimize the "broken."  A quick trip to Anaconda and the history
> on this shows a lot of the logic.  If systems were always configured correctly,
> if programs always followed standards, this would be a non-issue.

I'm occasionally a bit harsh. But it's the invention of undocumented
"standards" that can frequently cause problems.

> But many systems don't always have proper name resolution, name lookup and
> many other aspects, or programs don't always agree on everything the same.

Oh, I agree that the loopback listing for the system's default
hostnames is reasonable: it allows X to work when the network card is
dead, for example, and allows internal SSH or web services to work
properly. It's the syntax that's a longstanding problem.