It is. But then, there's SO many "buggy" names out there
that others did start to accept them. Ever since MS Exchange
made it into being there's heaps of hosts with an _ in the name,
too, which is COMPLETELY illegal (not just by position).

However, many MTAs have started accepting mail from
those hosts because those servers often are not only illegally
named, but also just drop bounces, and their admins are
incapable of finding out why mails from them don't arrive
at some sites. The CEO of the site with the nasty name
rings the CEO of the site that followed RFC and, voila,
another idiot has won.


Cheers,
Tink

Why must MS constantly redefine the standards to suit themselves? I'm not trying to slam MS, but given their position as a market leader, they've always tried to create "extensions" to RFCs and IETF standards.

And remember ASCII... 1 more bit and (almost) all latin languages fitted inside. A teacher told me IBM was upset against MS about this heresy. And as Tinkster said, another idiot (a HUGE one) has won.

It's a "bug" with DeviantArt in the sense that they are doing what they shouldn't, if we believe in following RFCs. I think the different behaviours we are seeing on our systems result from differences in the local resolver library.

First, the problem surely can't be in the upstream DNS servers, because I can ping kat1e- from my Windows laptop here but it fails from the Debian system on the same LAN, and both machines are querying the very same DNS servers.

That's why I think the difference must lie in the local resolver routine. The systems which say they can't find the domain must be filtering out what they see as a badly formed domain, before even asking the dns servers about it. The (Windows) systems which find the domain ok obviously don't care if it's badly formed, and query for it anyway.

Also, by the way, I don't think the test with dig is definitive:

But dig can give an answer for a nonexistent subdomain. Trying dig anything-at-all.deviantart.com gives the same output as you got for kat1e-. So the computer might not be capable of getting the IP address in that case.

Most likely, deviantart has a record of the form:
However, I would agree that some resolvers probably pre-filter on bad domains to save traffic.

Hi,

I know I already posted here to say it succeeded for me with Firefox under XP, but, I gave it another try it, here at work and it didn't work.

All my web traffic goes through a Novell proxy (BorderManager or something like that ... )

At home, with my squid proxy, it didn't work, and without the proxy, it works well ...

This shouldn't work anywhere. People not following the RFCs leads to this pile of headaches.

This is definitely wierd. I can get to the Deviant Art site mentioned in the first link and the Wings On Flames link but none of the others.
This then prompted my memory about something I had read elsewhere and it had to do with redirection scripting in the web site coding and optimising the site for search engines.

The answer that the person gave was this:
I notice from your server headers that your site is hosted on Linux using Apache. This means you can easily drop in an .htaccess to '301' (permanently forward) your non www domain to your www subdomain. If you need the code to do this here it is:-

create a text file called .htaccess and within it create this:-

Code:
Options +FollowSymLinks
RewriteEngine on
RewriteCond %{HTTP_HOST} .
RewriteCond %{HTTP_HOST} !^www\.frenchsecrets\.co.uk [NC]
RewriteRule (.*) http://www.frenchsecrets.co.uk/$1 [R=301,L]

(I think that's correct!)

Basically it's saying, if the URL does not (!) begin (^) with www then make it so using a 301 (permanent) redirect.

Now could it be that the browsers we are using are trying to access sites that DO NOT have this coding, do not do an automatic redirect and therefore will not recognise the site, or am I talking out of my hat??

David