I ran into some problems when QA sent back a bug regarding links that only had “http”, “http:”, “http:/” and “http://” entered. Odd results would come back. After some researching, I found the following.
http – attempts to goto a directory within the current page location. (same as “http://domain.com/http” if you are looking at “http://domain.com/page.htm”)
http: – goes to the current page that you are looking at. (same as “.”)
http:/ – goes to the root domain you are linked to (same as “/”)
http:// – theoretically goes to the root of the internet – but since a website is not located for the root of the internet, the browser does nothing.
All links are relative to the internet, yes? It was just interesting to see how it worked out. The classic problem came up where users enter “www.google.com”, but the webserver just returned a message that said the page wasn’t found. Of course, they forgot to enter the protocol, so the browser would interpret it as “http://mydomain.com/www.google.com”.