- From: <Albertfine@aol.com>
- Date: Fri, 5 Sep 1997 13:42:55 -0400 (EDT)
- To: www-html@w3.org
- cc: sugalsd@lbcc.cc.or.us
sugalsd@lbcc.cc.or.us (Dan Sugalski) wrote: >If you're going to maintain any sort of backward compatibility with current >HTML, I think you'll find that any error-free description of the data on a >page won't be much, if any, smaller than the actual data itself. Also, A text description could be compressed. It would probably be a series of numbers. If for example, a pattern repeats such as a row of three ones you can represent it with a single number. Their are other ways to compress the description. Remember, this is being done by the HTML editor. >since you're going to need to use font metrics to accurately make those >descriptions, you're going to end up with an inaccurate description for a >significant segment of the clients viewing your page. Font degradability is a big problem. It is almost a printing problem. This is really not my field. I think panose is the best solution. A base font could solve other problems. A database with a basic description of fonts would also help. These solutions would not be complete but not as inaccurate as you say. Of course, this only describes regular text. What about weird things like HTML math? I am working on a global description of a text. The main problem seems to be that it breaks at a variety of points. It really becomes a description of a series of squares and rectangles. The descriptions is not meant to only describe text. So I am considering other ways of describing a variety of data. I am sorry if it seems incomplete but I am working on it right now. Albert Fine
Received on Friday, 5 September 1997 13:43:00 UTC