Waiting for Godot.

Because execution is controlled by the user, no software, anywhere, can know whether a process is actually finished. Has a process been terminated, or has the user left a bookmark, with the intent to return? Will he or she return? What if the user is hit by a truck?

Normally, on a single computer, processes have well-defined beginnings and endings. On the web, you never know if the next click will actually happen.

If the information is stored in the server, you need to keep a lot of state information, most of which will be useless. For instance, a popular web page gets 100k hits per day. If you assume that someone might want to keep URLs on their hotlist that they only access once per year, you should probably store state information for three years. If there is 300 bytes of state information, you should store 100k * (3 * 365) * 300, or 30Gb of state information. While that's technically feasible, it's inelegant. Storing just 1Gb would satisfy most users, but is still inelegant. The natural place for the state information is in the browser.

NEXT Lab1153 Homepage