
"Joshua Hutchinson" <joshua@hutchinson.net> writes:
That does work, but why note use the already existing <note place="margin"> markup? Then, you don't have to do the extra work of segmenting your paragraph for the same result. For an HTML edition, it already works in our transform. The text version we just have to decide HOW to handle it, then code up the transform.
I'm not very happy with segmenting either... but I am trying to fit the concept of a running analysis into a larger framework for scaling texts. To understand what I am talking about it helps to think of a book in terms of the 3D Modeling concept of LOD (Level of Detail). A 3D computer model is made up of polygons. The more polygons you have, the more detailed the model. Models used in big Hollywood films may have millions of polygons in a model. This allows you to create believable virtual characters like Gollum, or (sadly) Jar Jar Binks. But all of those polygons are expensive to render. And if have a shot with Gollum in the distance, you will be spending enormous amounts of resources to draw polygons that can't be seen. To deal with this, LOD is used to reduce the number of polygons in a model the farther away it gets and then increase them again as the model gets closer. When you see a book from far away, you may only see the title on the spine on the shelf. When you get closer you take the book off the shelf and read a synopsis of the contents of the book on the dust jacket. Get closer still and you see a table of contents. Closer again you turn to a chapter and there might be a summary of the chapter at the beginning. Then, in the case of works like the Merlin, there is a running analysis which provides a paragraph by paragraph summary. Then, finally you get as close as you can and are confronted by the body of text itself. So, in this way you can see the running analysis as a way of zooming in or scaling the text. In the XML tests I was doing two years ago on the Merlin I used this concept to progressively zoom in on the book from a single title in a list, to a brief synopsis to the detailed synopsis to a table of contents to a chapter summary to the running analysis. With this kind of a structural approach to summaries and descriptions it was easy to create some very powerful browsing interfaces and indexing mechanisms. I haven't been happy with what I'd done before with the running analysis so my last post were working notes towards finding a way of incorporating summaries at different scales into a text rather than a proposal for PG. Your note approach is fine for providing a presentational means of adding in a running analysis, but it doesn't tell us the span of text that each note describes. This is why TEI offers the <seg> and <span> approach. I _DO_ agree that this as overkill for PG texts. When I went to market today, I was struck by the fact that the egg stalls recieve eggs from the farms in plastic flats. The eggs arrive in an organized structured way. The eggs sellers then proceed to pile them into piles in bins by price. This helps to muddle the difference between the eggs when you, the customer are picking through them. The structure and organization provided by the flats was counter productive for the egg sellers because it made it more difficult to unload bum eggs. In the same way, excessive structure in a marked up text makes it more difficult to transform into simpler formats. It would be very awkward to map the notes in the interp tags to the segs in the text. Your notes approach is a lot easier. But I like the idea of having a base reference text which others can use to overlay their own annotations. The `resp' attribute is good at indicating who has annotated what in a text, so that you could easily toggle between annotations from different sources, or strip them out all together. The EETS version of the Merlin is a base text which Wheatley has overlaid all sorts of information. It's a good idea to keep the markup mechanisms for overlayed annotations separate from the base text that is being annotated. This is a larger issue and goal than simply providing electronic editions of books, and is beyond what PG is about. But it is worth keeping these ideas in the back of your mind, if for no other reason than to remember that reading a book from cover to cover is not the only or even the most common way that books are used. b/ -- Brad Collins <brad@chenla.org>, Bangkok, Thailand