3 ms·
In general, the goal was to manage as little state as possible, so anything that was a one-off change, such as changing a single field, changing a node parent,
by fatnoah 3y ago
In general, the goal was to manage as little state as possible, so anything that was a one-off change, such as changing a single field, changing a node parent, etc. resulted in an async server round trip. This was done to reduce JS complexity (no jQuery or equivalent, so we had to maintain separate code for IE and Netscape) and to ensure validation was consistent. For more form-oriented activities, we used a more traditional form post & page refresh.
At the time, XML was the hotness, so all of the server back end APIs accepted and responded with XML, so when we did actually want to manage state, we did it through modification of the XML representation of the data the user was working with.