Future work will include some investigations into further expansions to this method, the introduction of hierarchical attribute values for example, where the apparence of a parent attribute could cause the whole child structure to be within the current context. This could improve the handling of folding structures.
Also worth investigating would be the introduction of elements that allow the actual manipulation of the current context, turning XHTML into a very simple programming language. Applications for this extension would have to be found though.
This extension was originally to be used exclusively at our institute as an additional service, but while working on it it came out that it was more flexible than originally thought. The possibility of running the parser locally as a preprocessor further enhanced its usability. This mode of operation even though not suitable for all such applications should be looked at by any who plans to improved HTML by some way.
Also we think that our extensions stay within the "spirit" of HTML by not interfering with the way the information should look on the side of the user, but introducing a new way of structuring HTML and introducing new semantical information into the raw document. The groups involved with working with HTML3 have not been approached but maybe this extension might make it as a valuable addition to the HTML set of elements.