Hacker News Reader

Comment by eru | Hacker News Reader

A trailing dot is entirely within the spec, isn't it?

For some fun: http://gs.com./ gives an error. https://news.ycombinator.com./ works, but has me logged out (probably the cookie doesn't match?).

View on HN

https://news.ycombinator.com/item?id=23479435

In 2020, someone find if you add "." in host suffix of Youtube url: http://www.youtube.com./ it will works without AD (IIRC, because AD resource file is viewed as different domain and blocked)

Alas, now it just redirects to the dotless version.

Appending "." to domain names helps bypass naive DNS based censorship, too.

http://microsoft.com./ has an interesting 404 page

I used to work for a guy who registered borgosoft.com and CNAMEed it to microsoft.com. In the days before HTTP 1.1, a Host: header was not required on a request, and it would happily serve up microsoft.com's content under borgosoft.com. Eventually they started requiring a host header, and the trick stopped working. The guy I worked for claimed credit for making them do this; I have my doubts.

Yes. The trailing dot is typically to tell the stub resolver not to try qualifying with the local domain. Rules vary by stub resolver though. On iOS I’m getting “A server with the specified hostname could not be found. NSURLErrorDomain” regardless of browser. I’ll have to debug tomorrow when I’m not too lazy to get up off the couch from my phone.

Both of your URLs work for me. A 404 on the first. I stay logged in to HN on the second. My guess is my trouble is with my local dnsmasq config.

The SNI spec dictates that trailing dots shall be stripped from the hostname.

The HTTP host header which is sometimes (ab)used to perform a kind of not-quite-right-SNI doesn't have the luxury of being quite so well specified.

Gs.com. is probably because the server doesn't expect a . At the end of the domain in the host header

Reply on news.ycombinator.com