<\body> There are three main kinds of objects for buffer management in : <\description> Every open document is stored in a unique editable buffer. Buffers typically admit aone to one correspondence to files on disk or elsewhere on the web. Some buffers are of amore auxiliary nature, such as automatically generated help buffers. All buffers admit a unique URL. In the case of auxiliary buffers, this URL is really a read-only ``placeholder'', so saving this kind of buffers is impossible (of course, it remains possible to save the buffer under a new name). It is possible to have multiple views on the same buffer. Every view is identified by aunique automatically generated URL, which again acts as a placeholder. Views (contrary to the buffers themselves) can be displayed in actual windows. Currently, any window contains a unique view and a view may only be displayed in one window at the same time (of course, it is possible to display different views on the same buffer in different windows). Windows are again represented by automatically generatedURLS. <\remark> In the future, views and windows should really be considered as documents themselves. Changes in the view will be automatically propagated (or not) to the corresponding buffer, and the other views. Windows will contain a document which specifies its layout (menus and toolbars). The corresponding view (or views) will be an active hyperlink (or active hyperlinks). The current APIs already reflect these future development intentions.