Automated Web PublishingStage Three
StageThree is an automated publishing system for the World Wide Web. For information on the path that led me to StageThree (including my intellectual debt to AutoWeb, check out PreFab's About This Site.StageThree is many things:
- an architecture or publishing model
- a sitewide page design and site navigation strategy
- a set of scripts that implement the above
Target User
One of the benefits of HTML is that it is a relatively simple markup language. Anyone with a modest technical background and a modest time investment can learn it. I doubt the Web would have taken off at breathtaking speed without such a simple -- accessible -- foundation. I hope that HTML sticks around as the foundation for Web publishing; but I'm convinced that the next generation of authors will focus on content and presentation, learning exactly zero HTML. These authors are my target.
Defining the Issues
At present, it appears that most Web publishing tools are focussed on creating a single page. Translators to HTML abound; HTML editors spring up from nowhere, new add-ons for editing HTML in another favorite editor are announced at every trade show. Although it is not yet clear which tools will thrive, most are adequate for the current "pioneer" generation to create a home page and perhaps a few linked pages of extras.What about creating a whole site? How many editors provide a simple way to create links? How are multiple documents organized? Must all links be maintained by hand, as documents are rearranged and new ones are added? Is there a simple way to have consistent headers, footers and other page elements? How are documents from multiple sources integrated? What tasks can be automated? These are the issues addressed by Stage Three.
StageThree Architecture
I believe it is useful to view Web publishing as three stages:- Raw documents, in any format (any file from any word processor or publishing software, any database, any e-mail system, etc. on any platform). These documents are created and modified using their "native" editor.
- Self-contained HTML documents. Each HTML page is likely to focus on a single subject (or set of related subjects). Each can generally stand alone, and could appear quite naturally in many different contexts. For a large web site, these documents may be created by many different authors.
- Site-linked HTML documents. The documents are assembled into a cohesive whole, with standard headers & footers (if appropriate), navigation links between documents, contents pages showing which documents are available on the site, etc.
My No-Tags Markup and WebDepot scripts turn Stage1 documents (text files and InfoDepot files, respectively) into Stage2 HTML documents; my StageThree scripts build the site.
Perhaps I will expand this discussion as questions raise issues to be addressed.
Sitewide Page Design, Site Navigation
Hierarchy is a natural way to organize information; books are divided into chapters, chapters (in technical books) into sections and subsections. Papers are stored in file cabinets, divided into drawers by subject or year, further divided into hanging folders then into manilla folders. Although the Web itself is unstructured, many individual sites will be organized hierarchically.I believe that StageThree's sitewide page design and site navigation falls naturally out of an understanding of the strengths and limitations of hierarchy. Starting at the top, the home page includes an outline of every document on the site (or, for deeply nested sites, at outline of the top levels). In fact, every page includes an outline of all descendents. I believe it is important to show more than just the immediate "children"; hierarchy can otherwise get in the way, imposing too many layers between here and there.
Equally important, every page (except the home page) shows its context, a complete list of ancestors leading back to the home page. It is easy to get lost in a conventional hierarchy; the problem is compounded since there are likely to be links across the hierarchy. The context provides both a map in and a way out.
Every page has links to its immediate "siblings", the Next and Previous item at the same level in the hierarchy. Note that an alternate strategy is to flatten the hierarchy so that Next/Previous can be used to go back and forth through the entire site ("depth first") with no dead ends or backtracking. Here's the crucial distinction for a "generation" in the middle: with StageThree, Next/Previous moves within the generation and the outline of descendents is the path down. With flattened links, Next would move down to the first "child". Neither approach is inherently superior; the former preserves structure and navigational flexibility, the latter is somewhat simpler. The structured approach is better where people are likely to stick to the structure, the flat approach is better where people are likely to continue reading independent of the structural boundary (e.g. in threaded discussions).
StageThree Scripts
This site and the PreFab site were built with StageThree scripts, written in Frontier. The context (ancestors), outline (descendents) and Next/Previous links are all built automatically. At present, the StageThree scripts are only available as part of my consulting services.
No-Tags Markup