Persistence of Location

The solution I came up with was to assign permanent numerical IDs to each of the relevant objects (essays, works, and actors) and to locate the objects within the site solely on the basis of these IDs. For example, this essay has the value 115 as its permanent ID, and will remain at the following location regardless of changes in title or hierarchy.

http://www.urticator.net/essay/1/115.html

As a side benefit, this method completely avoids the problem of having to invent file names based on the object titles.

In case you're curious about the extra directory level, what's going on is that in order to limit the width of the individual directory nodes, I've divided the objects into groups of 100. This scheme is efficient for up to 10,000 objects of each type, and so should keep me busy for quite a while.

As usual, there's an exception to the rule, but here it's because of technical difficulties, not for any good reason. The problem is with the indices: the number and character of the pages in each index is data-driven, and so is subject to change. Thus, if, for example, you wish to return to the list of good books, I recommend that you place the bookmark on the top page of the appropriate index, not on any of its children.