 | To deal with broken links, I propose that URLs have a date field that describes when the URL was created, and that web servers keep track of the history of their filesystems (the publicly accessible part of it). When a request comes in, the server can look at the history to map the location of the file when the URL was created to its current location, and send the file.
For instance, if a link to /foo.html on a server example.com was created on 10 jan 2005, the URL might be: http://example.com/foo.html[10-jan-2005] (I haven't given much thought to the syntax yet) If the resource moves to /bar.html on 1 feb 2005 and the server records this, the old URL would still work because the server knows that what is now /bar.html was /foo.html on 10 jan 2005.
What do you think?
-Kartik
|
|