> sometimes<p>All fully-qualified domain names end with a dot. The dot is effectively the label for the DNS root. Most tools (e.g. browsers) automatically assume a trailing dot.<p>[Edit] Interestingly, if I add a dot to this URL, after the domain-name, it works, and I get essentially the same content; but the styling is quite a bit different. I imagine it must be to do with the way stylesheets are referenced from HTML.<p>[Edit-2] I deprecate the use of the term "DNS root" to refer to the domains .com, .net, .org etc. These are top-level domains, not "roots".<p>There is a single root, and it's name is dot.
A few years ago there was a big "internet outage" in my country. The people responsible for the national TLD had forgot to put the trailing dot in their zone file somewhere, meaning lookups of example.tld failed -- but due to how search domains were configured by their servers, example.tld.tld did resolve!<p>Sometimes me and my friends refer to it as "the .tld.tld incident".
One of my first 'learning a language' projects was to write a DNS query in python from scratch. Basically, writing and reading bytes directly. Read the RFCs, made sure the format was right, wrote a parser for the reply and...fail. I remember pouring over every detail for about an hour before realizing the qname was missing the trailing dot. So, apparently at least some number of resolvers really demand it.
Relative DNS can be super useful.<p>For example, you could have a project where, by convention, the app servers are just called as <a href="http://app/" rel="nofollow">http://app/</a> the db servers are called as db etc. Saves lots of configuration effort. All you have to do is have a local DNS zone that keeps the traffic within its current setup.<p>That way you don't even need separate app configuration for separate environments, and can transport an application and its configuration without changes into the next environment.<p>(Sole exception: you likely want to inject different cryptographic keys to deal with (mutual) authentication).
Thanks to Julie for this; but I'm still a bit confused.<p>Does anyone share a suspicion that these myriad of anachronisms and bits of weirdness creates unnecessary complexity that swim through the system like a goblin in the machine? And that, if with 20/20 hindsight we just simplified things a bit, that many of our systems would be much more resilient and easy to manage?<p>I feel that we hacked the internet together, and in the impending rush to build more and more on top of it, we just resigned ourselves to the mess.<p>None of this (well, most of it) should be inherently hard. If we 'did it with foreknowledge' I wonder if all of these IT level networking issues would be pedantic, i.e. something you read a few quick docs and away you go. Everyone an expert pretty quick.<p>Merely looking at the options for my DSL device, my god man, even after all these years in tech, I'm still a bit bewildered.<p>Even at the consumer level, if I go over to someone else's house, I can't 'change the channel'. No idea what I'm watching, or the abstractions behind how it got to the TV.
I have submitted this exact article three days ago: <a href="https://news.ycombinator.com/item?id=32821927" rel="nofollow">https://news.ycombinator.com/item?id=32821927</a><p>Can someone tell me how this apparent duplicate comes into being? I was under the impression that HN consolidates identical submissions.
Interestingly, on some websites with paywalls, placing a dot at the end of the domain name actually allows you to bypass the paywall.<p>It used to work on more sites, but it has since been fixed. I guess someone caught up to the little trick haha