Message for Bind-users

[Kevin Darcy]

Eric A. Hall wrote:

> > There is no "practical" reason other than "this is the standard we
> > agreed to way back when and we're afraid to change it because then
> > we might break some lazy programmers' code (possibly causing security
> > holes, cancer, famine, or maybe even global thermonuclear
> > devastation)".
>
> You are certainly free to use whatever characters you want to. Just
> don't post here complaining that your applications are no longer
> working. I have seen many brave fools try to use underscores, and then
> things like DHCP, SMTP and the other services that enforce the naming
> rules just stop working.

So, the rules exist to prevent people from causing incompatibilities. And
the incompatibilities happen because of blind enforcement of the rules.
Sounds circular to me.

> > As for the historical reason, I've heard second- or third-hand that
> > some old teletypes used underscores instead of backspaces, or
> > something primitive like that.
>
> When ARPAnet was first put together, there wasn't an ASCII standard.
> Every host used different character sets. Using an underscore in a name
> has always been a good way to keep people from connecting to you. It
> still is.
>
> > This restriction kinda reminds me of how the dimensions of some of
> > the space shuttle parts are supposedly influenced by the width of a
> > Roman warhorse's rear end (war horses -> roads -> wagons -> railroad
> > tracks, you get the idea). Everyone is always too cowardly to risk
> > breaking compatibility with the old junk, so we have to live with
> > these arbitrary restrictions forever...
>
> If you read the rest of that, you'd have seen that the rockets are that
> size because the tunnels are that size. So breaking tradition in that
> case would mean that the rockets wouldn't get to the pad. The analogy is
> pretty much the same here. You can certainly say and do whatever you
> want, but you will most likely regret having done it.

And that's where the analogy breaks down, as all analogies ultimately do.
It takes a lot of effort -- *physical* effort -- to widen a tunnel, but
just a few code tweaks to start accepting underscores. IMO, lifting the
underscore ban, which would increase the available namespace and offer more
naming flexibility, exceeds the cost/benefit threshold in the long term.
But, given the effort required to implement it, unless and until a proposal
to widen all of the tunnels can be found to produce some *overwhelming*
positive advantage, it does not.


- Kevin