localhost is not resolved with DNS. It is kind of an alias for an ip address. Fun fact: you can define your own names in /etc/hosts (c:\windows\system32\drivers\etc\hosts for windows)
@@youdontknowme5969 No but it does bring up the topic of a computer's local hostname resolver found in the locations mentioned by @Vladislav Kalmyikov, which would be another fun techquickie all in it's own. Setting up local hostnames to get to private in-network services from say a local server.
@@BulbaWarrior every domain is an alias for an ip yk. besides localhost is in fact a local domain and local domains are never resolved by a dns but instead specified on a local network level which what the video and this comment were referring to.
Also, dotless domains confuse your browser if you’re typing it into an omnibox (it doesn’t understand if it’s a search term or a web address, this is also why the video instructed you to put slashes at the end for the two known dotless domains btw)
@@toptierdiscordmod7806 part of the issue of using a domain-less hostname is for things like routers and embedded devices, so they may not support https (or as is often the case I see, ship with self-signed certs).
I've spent a lot of time in my career parsing email and reading the actual IETF RFCs for SMTP to write parsers. A "dotless" domain is totally allowed in the specification. Where it isn't allowed is when you start to add security features such as DMARC/SPF/DKIM headers, Message Digest, Signing, and TLS for SMTPS. RFC 5321, section 4.1.2 defines the syntax for email addresses. Another "fun" feature is in how the local-part of the email address is defined. There is a lot that is allowed via the spec that most web based email validity parsers mishandle. Sorry to be "that guy". As someone who has handled a lot of email, I wanted to let you know you are incorrect.
@@mareviq They're usually doing that intentionally, because they don't want you to abuse their system by signing up for a million accounts using trivial mailbox aliases.
Except in the DNS protocol internals there's always a dot at the end that you never see in other software. It's significant when you're writing zone files though.
The binary protocol does not have any dots tho'. If you go to the page 31 of the RFC 1035 you can see how it is actually structured. It looks like: ( sudomain_length subdomain_chars )+ \0
Anyone else old enough to remember AOL Keywords? AOL was it's own private system in addition to the internet. So while you could enter a domain, you could also enter a "keyword" like "travel" or "weather", and go to the AOL-only page that was there. The same was true with email. You could enter someone's "screen name" and only type what's left of the @
google still owns that. if i type into my url bar i get search. most people google google, and it works 99% of the time, i get exactly what i want without knowing the domain.
@@teebu nope that's not how this works. the "URL bar" of today's browsers is called the omnibar which can handle either URLs or if the input can't be resolved to an URL a preconfigured search engine will get the input instead and delivers a search result page.
I love Riley, he's great with voice acting. Ltt really starting to feel like a traditional linear tv channel clips come up on youtube so often each day. It's like you could loop through all videos in real time and not really notice there's repeats unless you try hard to look for them.
Actually, all domain names end with a dot, you just don't have to type it, but you can if you want. See RFC8499 § 2 "Names", subsection "Format of names", for the distinction between the "presentation format" and the "common display format" (the latter being what's discussed in this video).
All DNS-adresses end with a dot, but it can be implicit. So all of them have dots (actually I'm not sure about the top-level, it may be the empty string, but I think it is a dot).
Dns is a hierarchy, every dot represents the next leaf node, the root starts with the dot at the end which can be omitted. I suppose if you could link an A record to . Instead of the root servers resolving all tld's you could have a site just being "." Which will be the mother off all sites.
Without the trailing dot, intranets (local networks) could append the rest of your domain to whatever you typed. It's much like "relative paths" vs "absolute paths", for those who know what those are.
I remember when this was common, it was called "America Online" and I believe the single "domain names" were called just "keywords" I like my DNS hierarchy and see no reason to change it thank you! 😎👍
That was a browser-specific feature. As long as you didn't use AOL browser (e.g. you used IE or Netscape), you didn't have this feature. If your ISP wasn't AOL, you could still use AOL keywords by using the AOL browser or typing them into AOL's search engine.
Back in the olden days, before the DNS, chaos ruled supreme. Imagine having to download a text file listing ALL the internet sites that you might want to use and then keeping it up to date. At least now there's a clear decentralisation of naming and the TLDs are a major part of that. Of course, when there's a networking problem it is usually because of a DNS misconfiguration.
"Imagine having to download a text file listing ALL the internet sites" - welcome to FidoNet, where you'd regularly download an updated 'node list', basically a BBS phone book, containing not only name and phone number, but also their max speed, compression standards, and the like. But, yeah, prior to DNS, it's all been local and shared /etc/hosts files basically, helping you with not having to remember all those IP addresses. (DNS, basically, is but a networked database of a massive /etc/hosts )
The UK was on JANET. Someone managed to find a gateway to the internet (supposedly it was 'stolen' but I always suspected it was leaked for experimental purposes) but the gateway didn't have a hosts file so you needed to find a source for that to get anywhere. That's how I found Nyx.
Granted, I have not read every single RFC, but I don't know anything in the relevant SMTP, ESMTP, or DNS RFCs which would prevent a single component domain name from working for sending email. Write an MX record, just like you'd write an A, AAAA, SOA, NS, or indeed any other record type, and you're good to go.
Based on what I know, yeah, dotless domain names can be used for emails (an example I see quite often is example@localhost) but I think Gmail and probably other web-based email clients are the ones that don't work with dotless domain names.
Yeah, that bit was guff. In the wild west days of the early enterprise I worked with a number of internal single word domains, and we had internal email setup for, uh, internal things! Very uncommon now for a myriad of other reasons, but SMTP isn't one
Yes, if it doesn't work with a certain piece of software or service, it's most likely an issue with implying a certain domain format restriction which isn't in the RFC / spec. Just look up how people recommend RegEx expressions to verify email addresses. This may actually be the source of the issue that the software actively refuses the address for no other reason. Since I've written my own DNS server (in C# which is running on my raspberry pi with mono) just for fun, I actually was knee deep in the DNS RFCs and I almost went nuts^^. There are so many extensions nowadays that it's almost impossible to include them all. Though the funny thing is that DNS doesn't have any "dots" at all in the protocol itself :) The protocol transmits hierarchical labels. In the protocol labels have a byte prefix that contains the length of the label. There are no dots transmitted in a request or response. The label system is even smart enough to allow references to other labels in the same request / response to save some space. This is actually pretty neat. So if a response includes something like ns1.my.awesome.domain.com, ns2.my.awesome.domain.com, ns3.my.awesome.domain.com it may be encoded as [3]ns1[2]my[7]awesome[6]domain[3]com[0] The other domains would be encoded like this: [3]ns2[0xC0][offset to the "my" label] and [3]ns3[0xC0][offset to the "my" label]. Technically a single label can't be longer than 63 characters. Any "length" value greater than 63 indicates a reference to any other point in the whole message. Reading / resolving a name means you just follow the chain until you reach a 0 length label which indicates the end. Note that an offset contains the lower 8 bits in the second byte and an additional 6 bits in the first byte. So an offset can be 16k instead of only 255. Usually DNS is served over UDP so you're naturally limited to about 512 bytes per message (request / response). Though DNS also works over TCP where the "packets" can be as long as necessary. Most commonly used DNS config files that use the bind syntax actually include the "root dot" to the right of the TLD. Most browsers or software which do DNS lookups would not use / interpret that correctly, but in the config it's common to see that extra dot to distinguish a sub domain namespace from a top level domain. Those config files are also build hierarchical. So a zone file could just define records within the zone they belong to without the need to write out the full domain name.
You can add any custom domain on C:\Windows\System32\drivers\etc\hosts (Windows) or /etc/hosts (Linux) but it only works for you. Those pn/ ai/ didn't worked on my PC or Mobile. Maybe does not work on cloudflare DNS.
Google did break some people's local config. Some developers used to internally register .dev for internal development sites. Then google register .dev as a gTLD. That's fine as long as there are no collisions. Until google changed the chrome so that all .dev domains had to be https by putting them in the HSTS preload list.
This video shows there is always something new to learn, i had no idea that top level domains could be used independently of domains, but it does explain 'localhost' which in retrospect was a clue hiding in plain sight
@@scytob Par for the course with these, they always miss some of the finer details of the topic. Like for clarity they should surely have mentioned things like .local which AFAIK is most often used on a LAN rather than dotless domains.
@@scytob well tell me why my googling verified it is a top level domain, but one that can't appear in a public dns because it is reserved and has a few other oddities based around conventions formed around how it is treated There are 3 other reserved tlds they are 'test', 'example' and 'invalid' and this is all documented in rfc2606 I've given you enough to get you going now its on you because you spouted off without anything to support your argument
@@alexatkin i agree .local is interesting, it also isn't actually DNS, it is mDNS - which is an entirely different resolver system and technology with a confusingly similar name :-) - what people don't seem to realize is any hostname used in a URI is not implicitly a DNS name, it could be an mDNS, DNS, netbios, widows broadcast or one of another number of name resolution protocols !
Unfortunately even if they wanted to, it wouldn't work. Your operating system automatically resolves localhost to 127.0.0.1, which is the loopback address so the traffic never actually exits your computer's network card.
@@gsck5499 in linux you could redefine localhost and localhost could used for domains but it shouldn't and may break some network software. I don't know if localhost is still in the windows .hosts file but it may by as local .hosts ist still supported as name resolution. I'm happy that network 44 moved on from the local .hostfile in 2008? after they sold a /16 block to Amazon.
Odd that Google wanted to do that, with no trailing slash too, given that it would mess up search. Now that the browser address bar is now also the search bar, there's got to be a clear difference between a URL and search term. It's not uncommon to want to search for only one word.
Of course they wanted to register w/o the trailing slash. That slash is not part of the domain name. DNs even *must not* contain slashes. It's just one part of the URI. It's just that modern Browsers let you use incomplete URIs and try to figure out the rest for your covienence. Is it even a incomplete URI or a search term, is it http or https, on which port etc. But sometimes (especially on dotless domains) the Browser just guesses wrong. The trailing slash just gives the Browser a hint that you meant an incomplete URI instead of a search term and may or may not work on different browsers (and settings). The better and more correct part to give as a hint would be the protocol part i.e. or
@@Biergartenparadoxon For sure, you can even see this by adding the protocol, typing pn for instance, so that it doesn't get confused. Big problem though is that it does otherwise get confused and I really see no way to differentiate between a single word search term or a single word URL in the same field unless you include some special characters like that; so why on earth Google, who's known for search and for the Chrome browser, wanted to make this more confusing is beyond me
@@Jamiered18 Of course I agree that it was a dumb idea for those reasons in the first place. Just disagree with "Odd that Google wanted to do that, with no trailing slash" since the trailing slash is not part of what you can register but a detail about how to use it afterwards.
The company I work at uses quite a few "dotless domains" on our internal network to access various services. The biggest issue is that all the web browsers also use the address bar as a search bar now, and sometimes when they see a dotless domain, they see it as a search term instead of figuring out if it's actually a valid address.
Those are probably just hostnames, not dotless domains. The work computers are typically all part of the same corporate intranet so they behave the same as a local network.
Microsoft used to sell internet "keywords". You could buy a word and when someone typed it into Internet Explorer it redirected to your choice of web sites. They were sold like domain names.
@@harryjohnson7714 the top (root) level "." is implied if left off. So while a dns management tool or system may not display it, it is treated as present. I shall also point out that the root dot is in the spec. RFC 1034: Domain names - concepts and facilities section 3.1 Name space specifications and terminology "Since a complete domain name ends with the root label, this leads to a printed form which ends in a dot."
On a related note, IP addresses also don't need dots. The dotted decimal format is the most common way to represent an IP address, but there are many other valid formats. You used to be able to craft IP addresses with 0, 1, 2, or 3 dots, using whatever mix of octal, decimal, and hexadecimal you wanted. You could even pad some of the numbers with extra data that would be ignored. It made it really easy to obfuscate IP addresses. And all web browsers and command line tools would decode them the same way. But modern computers are a lot more picky. It's a lot harder to find support for the alternative formats these days.
If it doesn't work in your browser, try put a dot anyway. So "pn." or "ai." The local domain that Riley talks about is also known as a "search domain". Putting a dot behind the hostname: no search domain tried.
There is one dot that is omitted from every public domain you mentioned: the one at the very end. Technically, the root of all domain names is a single dot. The TLDs are below that and therefore to its left when written. (The domain name system is a tree structure with the root at the right and branching to the left.) Thus, "com" is actually "com." and so on. It may very well work when included, but it is never required.
Starbucks claimed the trademark “twenty” for their large 20 oz size coffee and successfully sued another chain who was using the same name. Of course it was actually called a “venti,” which is just twenty in Italian. I guess if it’s a foreign word we can trademark it.
I'm pretty sure the SMTP protocol itself supports dotless domain names. Not sure if there are any SMTP implementations which don't, but they technically should.
This is for sure my favorite Techquickie video in terms of editing and writing the pacing feels actually better than other videos and feels like it's better to follow up and also the writing in some places with some good editing made me laugh so much I needed to replay it again
In reality, every domain has a dot at the end representing the root, but it's always added on to DNS requests for you because it's the only dot in a domain that you can infer. Thought the video might mention this, but NOPE LOL DOTLESS DOMAIN BRUH!
Of course, every domain (even "dotless" ones) actually ends in an implied "." which is the actual toplevel domain. You just rarely see it outside of DNS configuration files because it's the case 100% of the time, and thus can be implied.
SMTP does not care, what domain (or not) you type in ther... your mail client and/or your free-mail service would mind... maybe, in the webui... but smtp works fine with anything - the RFC specifies only @
I will add the dotless domain is required to work - because the local delivery agent would then be expecting that to be a hostname of a system on the local network
don't forget that you can have ipv6 instead of a hostname even with partial numbers. or user:password in front of the @. also you can use the user:password for regular websites that have htaccess as well to authenticate yourself directly. URLs are fascinating.
Technically the root oft the DNS tree, which all top level domains are relative to, is marked by a dot as well. So if you wrote a hostname in a URL without omitting anything, it would end in a dot even if it was just a TLD. It's just that the last dot can be committed without introducing ambiguity, so Browsers have come to accept domains that ommit the dot denoting the root. But technically speaking TLDs have a trailing dot as well. (See RFC 2181; See also RFC 1123, RFC 952 and RFC 921;)
Huh. I thought that Google would've been against dotless addresses. Anytime someone types just 'amazon' on a browser that has Google as its default search engine, it leads them to a Google result page. But I guess people typing 'search' could be more profitable for them.
In the current world where everyone has a single address/search combination bar with their default browser set to Google anyway, I think it would make more sense for Google to be against it, for the same reasons. It's hard to imagine more people search for "search" than they do external websites without the tld.
That additional profitability of .search .app .docs .eat .meme (and others) is exactly what google wanted (they also, likely, wouldn't bother paying other browser makers to have google as the default search provider).
Another reason dotless domains are bad is because they confuse browsers. If you type ‘london’ and hit enter, how does it distinguish that you want the google results page, or the domain under that name (say, a government website).
I want a tech quicky on what is a website. Who owns them? What is GoDaddy? Where is the information stored? When I buy a website, who do I pay? Can I make a random address or domain? If a website is seized, is it gone forever?
GoDaddy is a company that is accredited by ICANN (2:23) to register addresses for people, and they host content on their special computers that serve the content to the viewer from the network.
The recent set of Techquickies has been really good, well produced and very informative. Love seeing Riley stretch his legs. He's a great presenter. Well done.
Fun fact: no domains are dotless. However, the dot at the end is almost always hidden from the end-user. You can force your browser's omnibox to treat your LAN DNS as an address by putting a dot at the end (e.g. printer.lan. instead of printer.lan)
Dotless domains would also be a problem for web browsers, as the address bar is almost always used as a search bar (Firefox can still do a dedicated search bar, but I doubt anyone uses it anymore).
In firefox, even if you have the dedicated search bar enabled, searching is still performed from address bar (there might be about:config items to turn address bar search off, but I've not investigated).
local dotless domains usually work very differently than tld based one on the web. They rely on a search domain to get a fully qualified domain name, so you're actually visiting a shorthand dotless version of a site with a fully qualified domain name
Technically wrong. All domains need at least one dot. Namely, the one at the end that get added automatically. Try it out for yourself. Take any domain, add a dot at the end, and it will work just fine. P.S. 1:27 proves it.
pn/ doesn't works? It's because your browser probably doesn't treat it as a URL so it just googles it. *Try typing: http:/pn/* Edit: same with the other site, *http:/ai/* works.
ai./ (with dot) works for me in Windows (Firefox requires the slash, Edge doesn’t). pn. (with dot) works on iOS (it redirects to government.pn) but not on Windows.
No. Your firewall decides if a host is local or not based on the address returned, not the DNS name. .local is the only exception, at it implies handing the resolving off to mDNS (aka Apple Bonjour aka Zeroconf) instead of your proper DNS.
Nothing you wrote is correct outside of your limited environment and experience. And you should use .internal for your lan dns suffix. It’s what Google uses and is generally agreed that it will never be used as a public top level domain. Your firewall may run a dns server. Your operating system does not force you to use it. You may have configured your firewall to capture dns traffic and redirect them to the local dns server it runs, but I’d wager a guess that you have not done this, as it would be implemented in an enterprise network for security, etc. That said, I can’t stand this Linus guy and his channels. I thought I had blocked them all. I’ve become stupider by wasting my time here. It as if I have attempted to bail water out of a sinking boat with a tablespoon.
A lot of people mentioning that it does not work on mobile or desktop. Most reports from Windows or Android users. Works for me on iOS 10.3.3, Safari. Saw one guy saying it works with Chrome on Mac
Complete fakenews dotless Domains dont exist. Each Domain or FQDN is follower by a dot at the end. So the Domain is not 'ai/' it is 'ai./' the Browser is just hiding this dot.
that's why hosts.ini and arp tables exist. You don't lax security just because there is no dot, you do so because hosts sends you to a 10.0 or a 192.168. additionally, local namespaces CAN have dots.
As far as SMTP not being able to deal with dotless domains goes, I haven't checked lately, but I'd hope SMTP still supports bang paths, to be backwards compatible with UUCP - no dots there necessarily, but some "!"s
My company does this internally. I can type certain words with a slash following it and it takes me to an address within our internal server architecture. Super handy
Surprised you guys didn't mention localhost -- possibly the most popular dotless domain.
wdym? by popular do you mean used alot? but im the only one who uses it wtf
it's not a public DNS domain tho
localhost is not resolved with DNS. It is kind of an alias for an ip address. Fun fact: you can define your own names in /etc/hosts (c:\windows\system32\drivers\etc\hosts for windows)
@@youdontknowme5969 No but it does bring up the topic of a computer's local hostname resolver found in the locations mentioned by @Vladislav Kalmyikov, which would be another fun techquickie all in it's own. Setting up local hostnames to get to private in-network services from say a local server.
@@BulbaWarrior every domain is an alias for an ip yk. besides localhost is in fact a local domain and local domains are never resolved by a dns but instead specified on a local network level which what the video and this comment were referring to.
Makes me happy to know someone stopped Google from abusing their position.
Google always abuses their position.
@@nclsDesign doesn't literally every big company lol
@@IceBro That's why they are a big company
Just don't Google Google it will break the internet
from abusing*
Also, dotless domains confuse your browser if you’re typing it into an omnibox (it doesn’t understand if it’s a search term or a web address, this is also why the video instructed you to put slashes at the end for the two known dotless domains btw)
pro-tip: you can also put "" before it to tell the web browser that this is a url and not a search term
@@mysteryboyee yeah, I kind of realized that after writing this comment
@@mysteryboyee yay! save the dot, use a :
@@toptierdiscordmod7806 part of the issue of using a domain-less hostname is for things like routers and embedded devices, so they may not support https (or as is often the case I see, ship with self-signed certs).
Or use a separate search box like Firefox has (default disabled).
I've spent a lot of time in my career parsing email and reading the actual IETF RFCs for SMTP to write parsers. A "dotless" domain is totally allowed in the specification. Where it isn't allowed is when you start to add security features such as DMARC/SPF/DKIM headers, Message Digest, Signing, and TLS for SMTPS.
RFC 5321, section 4.1.2 defines the syntax for email addresses.
Another "fun" feature is in how the local-part of the email address is defined. There is a lot that is allowed via the spec that most web based email validity parsers mishandle.
Sorry to be "that guy". As someone who has handled a lot of email, I wanted to let you know you are incorrect.
thank you kindly :)
If I had a dollar for every time a web form that said my email address can't have a + in it...
@@mareviq They're usually doing that intentionally, because they don't want you to abuse their system by signing up for a million accounts using trivial mailbox aliases.
just remember: the easiest way to validate an email address is to hand it to an MTA and see if it bounces.
Thanks! Was about to ask already.
Everybody who wrote a email validation regex knows the pain.
Except in the DNS protocol internals there's always a dot at the end that you never see in other software. It's significant when you're writing zone files though.
Exactly, there's always a dot at the end!
The binary protocol does not have any dots tho'. If you go to the page 31 of the RFC 1035 you can see how it is actually structured. It looks like: ( sudomain_length subdomain_chars )+ \0
Anyone else old enough to remember AOL Keywords?
AOL was it's own private system in addition to the internet. So while you could enter a domain, you could also enter a "keyword" like "travel" or "weather", and go to the AOL-only page that was there.
The same was true with email. You could enter someone's "screen name" and only type what's left of the @
isn't that still the same stuff as in the video, with normal dotless domains? just provider-only.
google still owns that. if i type into my url bar i get search. most people google google, and it works 99% of the time, i get exactly what i want without knowing the domain.
I vagely remember my ISP doing something similar.
@@teebu nope that's not how this works. the "URL bar" of today's browsers is called the omnibar which can handle either URLs or if the input can't be resolved to an URL a preconfigured search engine will get the input instead and delivers a search result page.
@@CathrineMacNiel *omnibox.
Riley quickly became one of my favorite show hosts
Addresses without dots are pointless
Underrated comment
I love Riley, he's great with voice acting. Ltt really starting to feel like a traditional linear tv channel clips come up on youtube so often each day.
It's like you could loop through all videos in real time and not really notice there's repeats unless you try hard to look for them.
Yes! he's the best!!
Literally my favorite person at LMG. Seems like a fun guy tbh
.coffee is my favorite
So do I!! He’s simply the best next to Linus. Not throwing shade, but he’s even better than Andrew, who I honestly think is grossly overrated.
theres enough uploaded it run 24/7 and you would never see it all. Not to mention each new channel every day .
The dotless domain is really a bad idea. It had to be GOOGLE pushing for this. What a surprise 😲
Registering "search" for Google also seems like a very monopolistic thing to do.
@@Crux161 oh ok
@@Crux161 what
@@Crux161 a broken bot it seems.
What you don't know is that ICAN already gave them a TLD! They own ".google" and you too can own one for about 200K (+ ongoing yearly costs)
Actually, all domain names end with a dot, you just don't have to type it, but you can if you want. See RFC8499 § 2 "Names", subsection "Format of names", for the distinction between the "presentation format" and the "common display format" (the latter being what's discussed in this video).
That wasn't a terrible approximation of a British accent. Well done.
As a brit, it was terrible
@@giraffes19 you brits have so many weird and disgusting local dialects, that no matter how poor an accent is acted, it still fits
@@giraffes19 as a Brit, you are terrible
@@giraffes19 meh not worse than most tries on yt
bri'ish
yes i just did type that
All DNS-adresses end with a dot, but it can be implicit. So all of them have dots (actually I'm not sure about the top-level, it may be the empty string, but I think it is a dot).
Dns is a hierarchy, every dot represents the next leaf node, the root starts with the dot at the end which can be omitted. I suppose if you could link an A record to . Instead of the root servers resolving all tld's you could have a site just being "." Which will be the mother off all sites.
@@olafbuitelaar I think it would be more of an empty string "" then ".", since the last dot is still a separator like all other dots.
Without the trailing dot, intranets (local networks) could append the rest of your domain to whatever you typed. It's much like "relative paths" vs "absolute paths", for those who know what those are.
you recall correctly it is a dot (and this is just about the worst video i have ever seen on domain names)
This video is so bad and full of bad information. LTT do better!
I remember when this was common, it was called "America Online" and I believe the single "domain names" were called just "keywords" I like my DNS hierarchy and see no reason to change it thank you! 😎👍
🤘😎🤘
That was a browser-specific feature. As long as you didn't use AOL browser (e.g. you used IE or Netscape), you didn't have this feature. If your ISP wasn't AOL, you could still use AOL keywords by using the AOL browser or typing them into AOL's search engine.
Jesus i feel old...
Back in the olden days, before the DNS, chaos ruled supreme. Imagine having to download a text file listing ALL the internet sites that you might want to use and then keeping it up to date. At least now there's a clear decentralisation of naming and the TLDs are a major part of that. Of course, when there's a networking problem it is usually because of a DNS misconfiguration.
DNS still supports CHAOS. But virtually everything uses the IN class now.
"Imagine having to download a text file listing ALL the internet sites" - welcome to FidoNet, where you'd regularly download an updated 'node list', basically a BBS phone book, containing not only name and phone number, but also their max speed, compression standards, and the like.
But, yeah, prior to DNS, it's all been local and shared /etc/hosts files basically, helping you with not having to remember all those IP addresses. (DNS, basically, is but a networked database of a massive /etc/hosts )
The UK was on JANET. Someone managed to find a gateway to the internet (supposedly it was 'stolen' but I always suspected it was leaked for experimental purposes) but the gateway didn't have a hosts file so you needed to find a source for that to get anywhere. That's how I found Nyx.
Granted, I have not read every single RFC, but I don't know anything in the relevant SMTP, ESMTP, or DNS RFCs which would prevent a single component domain name from working for sending email. Write an MX record, just like you'd write an A, AAAA, SOA, NS, or indeed any other record type, and you're good to go.
Based on what I know, yeah, dotless domain names can be used for emails (an example I see quite often is example@localhost) but I think Gmail and probably other web-based email clients are the ones that don't work with dotless domain names.
Yeah, that bit was guff. In the wild west days of the early enterprise I worked with a number of internal single word domains, and we had internal email setup for, uh, internal things! Very uncommon now for a myriad of other reasons, but SMTP isn't one
@@regiondeltas was the internal internet for internal things about things happening internally and handled by intern?
Only one of the things which made my brain itch due to being technically incorrect in this videos 😅
Yes, if it doesn't work with a certain piece of software or service, it's most likely an issue with implying a certain domain format restriction which isn't in the RFC / spec. Just look up how people recommend RegEx expressions to verify email addresses. This may actually be the source of the issue that the software actively refuses the address for no other reason.
Since I've written my own DNS server (in C# which is running on my raspberry pi with mono) just for fun, I actually was knee deep in the DNS RFCs and I almost went nuts^^. There are so many extensions nowadays that it's almost impossible to include them all. Though the funny thing is that DNS doesn't have any "dots" at all in the protocol itself :) The protocol transmits hierarchical labels. In the protocol labels have a byte prefix that contains the length of the label. There are no dots transmitted in a request or response. The label system is even smart enough to allow references to other labels in the same request / response to save some space. This is actually pretty neat. So if a response includes something like ns1.my.awesome.domain.com, ns2.my.awesome.domain.com, ns3.my.awesome.domain.com it may be encoded as
[3]ns1[2]my[7]awesome[6]domain[3]com[0] The other domains would be encoded like this: [3]ns2[0xC0][offset to the "my" label] and [3]ns3[0xC0][offset to the "my" label]. Technically a single label can't be longer than 63 characters. Any "length" value greater than 63 indicates a reference to any other point in the whole message. Reading / resolving a name means you just follow the chain until you reach a 0 length label which indicates the end. Note that an offset contains the lower 8 bits in the second byte and an additional 6 bits in the first byte. So an offset can be 16k instead of only 255. Usually DNS is served over UDP so you're naturally limited to about 512 bytes per message (request / response). Though DNS also works over TCP where the "packets" can be as long as necessary.
Most commonly used DNS config files that use the bind syntax actually include the "root dot" to the right of the TLD. Most browsers or software which do DNS lookups would not use / interpret that correctly, but in the config it's common to see that extra dot to distinguish a sub domain namespace from a top level domain. Those config files are also build hierarchical. So a zone file could just define records within the zone they belong to without the need to write out the full domain name.
You can add any custom domain on C:\Windows\System32\drivers\etc\hosts (Windows) or /etc/hosts (Linux) but it only works for you. Those pn/ ai/ didn't worked on my PC or Mobile. Maybe does not work on cloudflare DNS.
RFC standard for email addresses even allows IP address instead of domains. Why wouldn't it work with a dotless donain?
You should do a techquickie on the many different pronunciations of Asus.
Haha. I second this!
As a website developer who thought I knew a lot about DNS and everything, I had no idea this was a thing!
If it helps, most system administrators don't even know everything about DNS. There's a lot.
@@robspiess tbf dotless domains is a pretty stupid idea
Google did break some people's local config. Some developers used to internally register .dev for internal development sites. Then google register .dev as a gTLD. That's fine as long as there are no collisions. Until google changed the chrome so that all .dev domains had to be https by putting them in the HSTS preload list.
lots of isps used to use internal domains like "mail" and "news" to make settings easy for new internet users.
This video shows there is always something new to learn, i had no idea that top level domains could be used independently of domains, but it does explain 'localhost' which in retrospect was a clue hiding in plain sight
no it doesn't because localhost is not a domain (dotless or otherwise) it is a hostname - this video is wrong and misleading in many respects
@@scytob Par for the course with these, they always miss some of the finer details of the topic. Like for clarity they should surely have mentioned things like .local which AFAIK is most often used on a LAN rather than dotless domains.
@@alexatkin unfortunately with this video they missed it completely, from the main topic down to the details
@@scytob well tell me why my googling verified it is a top level domain, but one that can't appear in a public dns because it is reserved and has a few other oddities based around conventions formed around how it is treated
There are 3 other reserved tlds they are 'test', 'example' and 'invalid' and this is all documented in rfc2606
I've given you enough to get you going now its on you because you spouted off without anything to support your argument
@@alexatkin i agree .local is interesting, it also isn't actually DNS, it is mDNS - which is an entirely different resolver system and technology with a confusingly similar name :-) - what people don't seem to realize is any hostname used in a URI is not implicitly a DNS name, it could be an mDNS, DNS, netbios, widows broadcast or one of another number of name resolution protocols !
Imagine the scammers paradise it would be if dotless domains were more common, literally any word could be used like "winner/"
How ironic.. you have a spam reply
haha the winner was the spambot's keyword
irs/
@@tomikun8057 HAhahahhaha
google wanted to register „localhost“ and getting traffic from all machines in the world 😂
Unfortunately even if they wanted to, it wouldn't work. Your operating system automatically resolves localhost to 127.0.0.1, which is the loopback address so the traffic never actually exits your computer's network card.
@@gsck5499 not fully automatically, it's defined by /etc/hosts file
@@gsck5499 in linux you could redefine localhost and localhost could used for domains but it shouldn't and may break some network software. I don't know if localhost is still in the windows .hosts file but it may by as local .hosts ist still supported as name resolution. I'm happy that network 44 moved on from the local .hostfile in 2008? after they sold a /16 block to Amazon.
@@gsck5499well done ya missed the joke
Odd that Google wanted to do that, with no trailing slash too, given that it would mess up search. Now that the browser address bar is now also the search bar, there's got to be a clear difference between a URL and search term. It's not uncommon to want to search for only one word.
Of course they wanted to register w/o the trailing slash. That slash is not part of the domain name. DNs even *must not* contain slashes. It's just one part of the URI. It's just that modern Browsers let you use incomplete URIs and try to figure out the rest for your covienence. Is it even a incomplete URI or a search term, is it http or https, on which port etc. But sometimes (especially on dotless domains) the Browser just guesses wrong. The trailing slash just gives the Browser a hint that you meant an incomplete URI instead of a search term and may or may not work on different browsers (and settings). The better and more correct part to give as a hint would be the protocol part i.e. or
@@Biergartenparadoxon For sure, you can even see this by adding the protocol, typing pn for instance, so that it doesn't get confused. Big problem though is that it does otherwise get confused and I really see no way to differentiate between a single word search term or a single word URL in the same field unless you include some special characters like that; so why on earth Google, who's known for search and for the Chrome browser, wanted to make this more confusing is beyond me
@@Jamiered18 Of course I agree that it was a dumb idea for those reasons in the first place. Just disagree with "Odd that Google wanted to do that, with no trailing slash" since the trailing slash is not part of what you can register but a detail about how to use it afterwards.
The company I work at uses quite a few "dotless domains" on our internal network to access various services. The biggest issue is that all the web browsers also use the address bar as a search bar now, and sometimes when they see a dotless domain, they see it as a search term instead of figuring out if it's actually a valid address.
Firefox has an option in the settings to split up search and address bar
Those are probably just hostnames, not dotless domains. The work computers are typically all part of the same corporate intranet so they behave the same as a local network.
Brit here - that phishandsiliconchips gag damn near killed me. Well played!
Microsoft used to sell internet "keywords". You could buy a word and when someone typed it into Internet Explorer it redirected to your choice of web sites. They were sold like domain names.
that's evil genius haha
4:05 We’ll be right back. I love this 😂
Except domain records are stored with tail dot. So there are always dot in domain name
Nope. Not all of them are stored with a tail dot
@@harryjohnson7714 the top (root) level "." is implied if left off. So while a dns management tool or system may not display it, it is treated as present.
I shall also point out that the root dot is in the spec.
RFC 1034: Domain names - concepts and facilities
section 3.1 Name space specifications and terminology
"Since a complete domain name ends with the root label, this leads to a printed form which ends in a dot."
@@harryjohnson7714haha, you can’t go against RFC
On a related note, IP addresses also don't need dots. The dotted decimal format is the most common way to represent an IP address, but there are many other valid formats. You used to be able to craft IP addresses with 0, 1, 2, or 3 dots, using whatever mix of octal, decimal, and hexadecimal you wanted. You could even pad some of the numbers with extra data that would be ignored. It made it really easy to obfuscate IP addresses. And all web browsers and command line tools would decode them the same way. But modern computers are a lot more picky. It's a lot harder to find support for the alternative formats these days.
I see Riley. I laugh before clicking, then laugh again when he throws in jokes before 5 seconds. 🤣😂
I've been studying to get computer certifications, being able to understand these different services and records and ports makes me so happy.
If it doesn't work in your browser, try put a dot anyway. So "pn." or "ai."
The local domain that Riley talks about is also known as a "search domain". Putting a dot behind the hostname: no search domain tried.
Worked when I typed the full address: ai/
thanks this worked even when http formatting didnt
There is one dot that is omitted from every public domain you mentioned: the one at the very end. Technically, the root of all domain names is a single dot. The TLDs are below that and therefore to its left when written. (The domain name system is a tree structure with the root at the right and branching to the left.) Thus, "com" is actually "com." and so on. It may very well work when included, but it is never required.
Starbucks claimed the trademark “twenty” for their large 20 oz size coffee and successfully sued another chain who was using the same name. Of course it was actually called a “venti,” which is just twenty in Italian. I guess if it’s a foreign word we can trademark it.
you can trademark english words - Apple™️
Had no clue this was even a thing! Great video as always!
You forgot to mention that all domains have an implicit dot at the end, so technically there are no dotless domains.
I like how you've chosen to use the flags of Guernsey and the British Indian Ocean Territory at 1:05
I'm pretty sure the SMTP protocol itself supports dotless domain names. Not sure if there are any SMTP implementations which don't, but they technically should.
Most SMTP servers even support IP-addresses, etc.
This is for sure my favorite Techquickie video in terms of editing and writing
the pacing feels actually better than other videos and feels like it's better to follow up and also the writing in some places with some good editing made me laugh so much I needed to replay it again
In reality, every domain has a dot at the end representing the root, but it's always added on to DNS requests for you because it's the only dot in a domain that you can infer. Thought the video might mention this, but NOPE LOL DOTLESS DOMAIN BRUH!
Of course, every domain (even "dotless" ones) actually ends in an implied "." which is the actual toplevel domain. You just rarely see it outside of DNS configuration files because it's the case 100% of the time, and thus can be implied.
SMTP does not care, what domain (or not) you type in ther... your mail client and/or your free-mail service would mind... maybe, in the webui...
but smtp works fine with anything - the RFC specifies only @
I will add the dotless domain is required to work - because the local delivery agent would then be expecting that to be a hostname of a system on the local network
don't forget that you can have ipv6 instead of a hostname even with partial numbers. or user:password in front of the @. also you can use the user:password for regular websites that have htaccess as well to authenticate yourself directly.
URLs are fascinating.
Technically the root oft the DNS tree, which all top level domains are relative to, is marked by a dot as well. So if you wrote a hostname in a URL without omitting anything, it would end in a dot even if it was just a TLD. It's just that the last dot can be committed without introducing ambiguity, so Browsers have come to accept domains that ommit the dot denoting the root. But technically speaking TLDs have a trailing dot as well.
(See RFC 2181; See also RFC 1123, RFC 952 and RFC 921;)
Huh. I thought that Google would've been against dotless addresses. Anytime someone types just 'amazon' on a browser that has Google as its default search engine, it leads them to a Google result page. But I guess people typing 'search' could be more profitable for them.
In the current world where everyone has a single address/search combination bar with their default browser set to Google anyway, I think it would make more sense for Google to be against it, for the same reasons. It's hard to imagine more people search for "search" than they do external websites without the tld.
That additional profitability of .search .app .docs .eat .meme (and others) is exactly what google wanted (they also, likely, wouldn't bother paying other browser makers to have google as the default search provider).
Another reason dotless domains are bad is because they confuse browsers. If you type ‘london’ and hit enter, how does it distinguish that you want the google results page, or the domain under that name (say, a government website).
I remember long ago when Opera had /. shortcut for slashdot. But it doesn't work anymore.
I remember Opera from long ago..
Love the little Tim and Eric reference 4:06
This was really informative! I honestly had never given this much thought, but you can bet that I will be now!
the editing in this is a piece is memey art
I want a tech quicky on what is a website. Who owns them? What is GoDaddy? Where is the information stored? When I buy a website, who do I pay? Can I make a random address or domain? If a website is seized, is it gone forever?
This is a good idea. Someone should make this. I know the answers to this though.
GoDaddy is a company that is accredited by ICANN (2:23) to register addresses for people, and they host content on their special computers that serve the content to the viewer from the network.
@@davidparke8896 ummm does godaddy have ip ranges to allocate to customers? Did you mean to say register (domain) names?
@@dekeonus Yeah, they register domain names for people, yeah. I’ve heard of URLs being called “web addresses” for some reason.
You forgot the most important reason ICANN banned dotless domains. It is because THEY CANN.
The recent set of Techquickies has been really good, well produced and very informative.
Love seeing Riley stretch his legs. He's a great presenter.
Well done.
except this one was really bad. Just plain wrong from the top level down to the details
Is this video from the early 90s?
Fun fact: no domains are dotless. However, the dot at the end is almost always hidden from the end-user.
You can force your browser's omnibox to treat your LAN DNS as an address by putting a dot at the end (e.g. printer.lan. instead of printer.lan)
Dotless domains would also be a problem for web browsers, as the address bar is almost always used as a search bar (Firefox can still do a dedicated search bar, but I doubt anyone uses it anymore).
I do
@Octeract-[SG]:
I still use it :-)
In firefox, even if you have the dedicated search bar enabled, searching is still performed from address bar (there might be about:config items to turn address bar search off, but I've not investigated).
@@dekeonus Yeah, I seem to recall the address bar working anyway.
Neither pn/ nor ai/ worked for me. Chrome on Win 11 using Cloudflare DNS.
ai./ (with dot) works for me in Windows. pn. works on iOS with Safari (it redirects to government.pn)
That's actually not the worst fake British accent I've ever heard
That push from Google in order to make "search" and "maps" redirecting to them is pretty smart
Scrolled a few comments and was surprised that no one has brought up AOL Keywords yet.
As a person who lives in Anguilla, can confirm .ai indeed works. Great vid!
Must be nice, but why is there a little dix next to cauls bottom?
@@jr2904 mann idk lool
Actually, the I is dotted and there's 2 dots on the colon ()
^
I thought it was about IPv6. What a surprise.
The Eric Andre bit at 4:03-4:10 has me in stitches 😆
My ISP seems to block those DNS lookups
This was an specifically good episode. Nice and informative and presented and edited well. Cheers.
Riley's voice acting is top notch
No it isn’t…but it is funny.
Tell me you're not a brit
@@Dudae_ I can’t tell you I am not a Brit.
@@RabbitStu-M7UTS [I meant the other guy]
I can reassure you, your wrongness is top notch.
I saw the thumbnail and got really excited for a 3rd game announcement
That uk accent got me bruh 😂.!
Idk why i was expecting a "boddle of waddar"
Harry Podcast?
local dotless domains usually work very differently than tld based one on the web. They rely on a search domain to get a fully qualified domain name, so you're actually visiting a shorthand dotless version of a site with a fully qualified domain name
Not gonna lie that email gag was unexpected and made me laugh a lot.
Technically wrong. All domains need at least one dot. Namely, the one at the end that get added automatically.
Try it out for yourself. Take any domain, add a dot at the end, and it will work just fine.
P.S. 1:27 proves it.
Why doesn't this work on desktop?
Your browser is probably trying to access an intranet site as suggested in the latter half of the video.
Browsers generally just assume you are trying to search for something
pn/ doesn't works? It's because your browser probably doesn't treat it as a URL so it just googles it.
*Try typing: http:/pn/*
Edit: same with the other site, *http:/ai/* works.
ai./ (with dot) works for me in Windows (Firefox requires the slash, Edge doesn’t). pn. (with dot) works on iOS (it redirects to government.pn) but not on Windows.
that email explosion was unexpected but hilarious
No. Your firewall decides if a host is local or not based on the address returned, not the DNS name. .local is the only exception, at it implies handing the resolving off to mDNS (aka Apple Bonjour aka Zeroconf) instead of your proper DNS.
No
Nothing you wrote is correct outside of your limited environment and experience. And you should use .internal for your lan dns suffix. It’s what Google uses and is generally agreed that it will never be used as a public top level domain.
Your firewall may run a dns server. Your operating system does not force you to use it. You may have configured your firewall to capture dns traffic and redirect them to the local dns server it runs, but I’d wager a guess that you have not done this, as it would be implemented in an enterprise network for security, etc.
That said, I can’t stand this Linus guy and his channels. I thought I had blocked them all. I’ve become stupider by wasting my time here. It as if I have attempted to bail water out of a sinking boat with a tablespoon.
Also makes sense because having a dot in every web address makes it easier to convey in text/speech
Ai/ didn’t work on mobile chrome
Did for me on android
It does, simply ignore the security warning and press, "continue to site"
Don't use wi-fi, use mobile data, i think he should've explained Properly
Title makes it sound as if no Internet address needs a dot
*Tries ai/ and pn/* ... does not work 😅
ai./ (with dot) works. pn. only on iOS (redirects to government.pn).
A lot of people mentioning that it does not work on mobile or desktop. Most reports from Windows or Android users. Works for me on iOS 10.3.3, Safari. Saw one guy saying it works with Chrome on Mac
Yea DNS_PROBE_FINISHED_NXDOMAIN with Chrome on Win10. Both work on Chrome for Android.
@1:05
reminds me of that joke, why is it prenounced bri - ish?
because they drank the T
lol thats soooo hilarious
That was actually super interesting.
Complete fakenews dotless Domains dont exist. Each Domain or FQDN is follower by a dot at the end. So the Domain is not 'ai/' it is 'ai./' the Browser is just hiding this dot.
Stellar editing job
Claim your “here within an hour” ticket right here❤️
ty
Within a minute buddy
Got it
shhhh. shut up, no one cares!
Ok
As someone that's recently moved to London I'm incredibly disappointed in the lack of "Riley's idea of a Londoner's accent" here
nice bi'ish accent 1:05
👆👆congratulations🎊you have been randomly selected among my shortlisted winners you just won a prize🎁🎁🎁...
Can't believe Techquicke dropped a hard R....
Using good old german Eltz Castle to explain networking stuff, Iike 👍🏻
Why am I not surprised that Google tried to do another antitrust thing? That's like AOL keywords. No one uses AOL anymore.
Acccttuaallllyyy... a fully qualified domain name ends in a dot. Try it, add a period to any domain name.
that's why hosts.ini and arp tables exist. You don't lax security just because there is no dot, you do so because hosts sends you to a 10.0 or a 192.168. additionally, local namespaces CAN have dots.
Thank you for educating me.
the best way to enable dotless domains on your PC, is to type it in the address bar, then press CTRL and F4
Thank you for the video!
Cool to know!
As far as SMTP not being able to deal with dotless domains goes, I haven't checked lately, but I'd hope SMTP still supports bang paths, to be backwards compatible with UUCP - no dots there necessarily, but some "!"s
My company does this internally. I can type certain words with a slash following it and it takes me to an address within our internal server architecture. Super handy
Those are likely just hostnames for the VMs those services live on
You don't need dots, you should be using colons like we did at Xerox! For instance OSBU:South :)