Redesigning Frankston.com
The first major redesign of this site since 1998.06-Jul-2002

I created this site in 1998 and have made few style improvements since then. The frustration has been building as I've tried to maintain pointers to new essays published on SATN as well as this site. I also liked some of the style that Dan Bricklin used on that site but didn't want to build the style in.

Looking Back

This project also gave me a chance to revisit writings from 1998 as well as earlier writings, such as my Master's Thesis which was written in 1974. Back then there were no personal computers. In fact, there weren't even calculators. When I tried to tell people that computers were fundamental and useful I met, at best, incomprehension. Even in the 1980's the whole idea that normal people would use email, let alone depend on it, was strange.

Much of the writing on the site since 1998 has focused on the IP Infrastructure and related issues. While this may seem to be a narrow set of topics, it reflects underlying themes that have far larger implications. The Internet does serve as a very useful casebook study in how complex systems evolve and produce stable systems without any central authority guiding them. I do intend to write about these from a social and philosophical perspective but, it has seemed to be more effective to focus on public policy issues related to connectivity. But without the reader having the larger perspective the claims I see as obvious seem arbitrary.

The Site Itself

Since I'm the entire staff for this site any approach has to be low maintenance and also make it easy to publish documents and update the site. One reason for using blogger for SATN is that it provides me with a nice publishing capability but it doesn't provide me with the document management facilities I need.

I don't know of any tools that meet my needs and doing it myself is a great learning experience though often problematic. Whatever approach I take must leave the site operational at any point. The site itself is maintained on my own computers and then posted to the public site so I have one level of buffering but I don't have the luxury of being being able to run two sites in parallel. If things do go wrong I can fix it though not necessarily right away. The site itself is not pure -- older documents will continue to function and URLs reference the site should continue to work though some internal (#) pointers may not stay valid.

For Those Interested in the Site Technology

Aside from various stylistic improvements (such as the sidebar), I also wanted to assure that the site could be viewed by most browsers so the contents are generated server side with simple HTML for the clients.

The actual page is assembled out of simple pieces:

  • The document itself. While I've been freely using scripts in some of the documents, the new approach requires that the documents use XHTML. XHTML is the same as HTML but with more stringent requirements. The good news is that there is some support in FrontPage but it is far from perfect.
  • A file with a standard header and style information. Again, using XHTML.
  • A data file, in XML, with a list of items, categories and other information used to generate the sidebar and cross reference tables.
  • The JavaScript program that assembles the elements. It processes the XHTML files to adjust them and produce the actual documents.

I'll write about this more later. But first I want to make sure that the site is ready for wider use.

Just some quick notes:

  • Future plans
    • Store descriptive information in documents so I can generate the items database automatically
    • Per-user presentation preferences
    • Blog RPC stuff?
    • Use the CLS (.NET) server support instead of JavaScript and ASP
  • Technical issues
    • FrontPage "Apply XML style" doesn't always do a good job
      • .XHTML isn't a suffix so I need to manually convert HTML to XHTML using "Apply XML style"
      • Doesn't handle attributes without values like "nowrap"
      • " " isn't handled well in XML. I should be able to figure out how to handle this better
      • A list element often generates an unpaired "<p>"
      • It insists on adding a header to ASP pages which frustrates generating clean documents. I need to make sure I don't let FrontPage help me with these.
    • Since I am using the documents as data I do some processing that I would normally do in ASP:
      • Generate per-view mailTo entries (for spam control)
      • Remove the header information that FrontPage generates but retain the style sheet
      • Extract the document title.
    • I use XMLSpy to work with XML and it is very helpful.
      • But it is not good at handling entities like converting <> to &lt; That should transparent.
      • It insists on retaining attribute order which makes it difficult to work with a set of entries as a data table as well as I'd like.
    • XML
      • The MS XML DOM doesn't allow me to insert XML. This makes it difficult to do the equivalent of setting innerHTML.

For now ...

These redesign comments represent another work in progress ... maybe I'll write more later.