Monday, November 22 2004

It's a terrible name, I know. It betrays everything that is good to neglect a project a compelling name. Naming things, after all, is Important.

This is an XSLT 1.0 stylesheet for converting a DocBook article into a draft-quality Simple Standards-Based Slide Show System (S5) XHTML presentation.

I gave a talk last Friday. When it came time to make the slides, I decided to give Eric Meyer's magic HTML-CSS-JavaScript presentation tool a whirl. It's a fantastic piece of work : easy to set up and, aside from the poorly commented stylesheets, understand. Creating the slides consisted of my copying headers from a PDF file into h1 tags and then boiling down the contents of that particular section in to suitable buzzword-bingo.

There is nothing wrong with this approach; it works just fine, thank you. The only real criticism you can make is that it's boring and boring tasks are good candidates for automating. One of the reasons that S5 is so nice is that all your slides are contained in a single file, which means you can generate a presentation using XSLT. (Single-document output is a limitation of XSLT 1.0 that has been addressed in version 2.0 of the specification; the spec, however, is still being tinkered with and application support is limited at the moment.)

I write anything that looks, or smells, like a paper in DocBook and then convert it to some other more palatable format for viewing. It's easy to imagine using S5 for any future speaking engagements so over coffee, this weekend, I wrote an XSLT stylesheet to convert DocBook articles in to bare-bones S5 presentations. The tool takes care of generating the various HTML headers, a couple of default slides for abstract and copyright information, and a new slide for every section element it finds, mapping titles along the way.

That's it. The meat of the presentation will still need to be written by hand. I'm sure that there are some people out there who would like a program to analyze the content of an essay and generate talking points from it but I'm not one of them. This is, to be clear, an exercise in leg-work not artificial intelligence. Poorly named, at that.