:)

*Slightly* longer than the original URL. I'm guessing it has to do with using the key from the SHA-256/SHA-512 hash.

Original: http://www.dailymail.co.uk/news/article-1127093/Brown-pips-European-rivals-phone-President-Obama.html

Barney's: http://www.barneyb.com/applications/urlshrink/app/shorturl/3BBDE5E5539B479F341D60ED74F5E9B655BE06798E689AD87392EE910A65992F

SnipURL: http://snipurl.com/an586

Metamark: http://xrl.us/bednsq

I appreciate Barney's domain and app path adds a lot to the url's length, but I'm not sure it uses a full 64 chars for the identifier? This is a lot of padding. Instead of using 16 base consider using 36 base so your encoded urls become a lot shorter. If it logs the full urls in memory or even a db (so the short url identifier points to a record or struct) then start counting from 100 rather than a trillion trillion (or whatever a 64 char hex value is). That way the short urls will always be considerably shorter, or just a handful of chars long, even if there's a billion of them.

The advantage of your code is that each URL is almost "private" making it nearly impossible to guess working short urls by randomly typing them in. But private information on the web should be protected by a proper login process rather than security by obscurity which leads me back to the thought that a short url should be short.

I just listed you on my TinyURL Alternatives list.