Implementation

Directory Structure

The directory structure is a direct translation of the linking diagram which makes maintaining the web a relatively straight-forward process. Each major topic being provided with its own directory, and I am confident about doing this because directories take up very little memory space, and aid structure enormously. Within the directory, two files are found;

• index.html
  Which is the 'main' page of the directory.
• bullet.gif
  Which is the icon used to symbolise that directory, by the list in the parent directory.

For example, information about the Erasmus exchange scheme is given its own directory, /department/erasmus. In this directory, is the department/erasmus/index.html file, which is the main page for the Erasmus information, and a department/erasmus/bullet.gif icon which is the icon placed next to 'Erasmus Information' in the /department/index.html page.

I have decided to use this method, because it means that the index file or the icons representing topics can be replaced or modified very easily, by simply locating that topics home directory.

Page Generation

The pages were created semi-automatically using a PC and Windows 95. I refined the generic look and feel page, and replaced 'Document Title' and 'sample text…' with control strings, "DOCUMENT_TITLE" and "TEXT_1". (View the control version). I then wrote 2 macros using Microsoft Word 7;

1. To replace the control strings with my strings of my own choices.
   (View the WordBasic source.)
2. To insert the major links, written in HTML, to subdirectories of my own choice, drawing on the two files embedded within that directory.
   (View the WordBasic source.)

I also found that I needed to supplement some links with explanatory text (for example, the Internet page). This posed a problem due to the fact that when a <H2> heading is ended using </H2>, there is a large gap before the following text. This made the page look 'messy', and so I implemented the explanatory text using tables. View the HTML. I considered replacing every instance of a bulleted link within the entire web with a table like this, so that the HTML is consistent, and maintainers can easily insert any explanatory text, but I decided this would make the HTML too complicated, and so table-based links would only be inserted where necessary.

Porting

At this stage the web was being developed on my home computer, running Windows 95, and the 'root' directory for the final ICS web was "file://c:/windows/desktop/ics/ics_web". When the web is transferred to the public web server, I will convert every occurrence of the old root to the appropriate root, which will be "http://www.cs.man.ac.uk/~buckleya/ics/ics_web". To do this requires the following steps;

• Transfer the entire ics directory structure to my public_html directory.
• Compile a list of every HTML file used within the final ICS web by using the UNIX command,
  find ~buckleya/public_html/ics/ics_web -name *.html -print > dir_list.txt
• Write a UNIX shell script which searches a file for every occurrence of the old root, and replaces it with the new root. This executable file is called fix_html.
• Compose a batch file based on the dir_list.txt which executes the script on each file. This is done by the UNIX command,
  cat dir_list.txt | sed s/"~buckleya"/"fix_html ~buckleya"/g > fix_list
• Convert the fix_list to a recognised UNIX executable script, by adding the line #!/bin/sh to the beginning, and changing its attributes to executable.

Following these steps makes sure the finished ICS web is correctly configured, and all graphics and links work correctly. I also used the above methods to create the two .tar files, which should make porting the ICS web to www.ics.lu.se a trivial process (hopefully!).

Note: This method replaced by using relative links 16-May-2002

Web Index

I created the web index manually, because I lacked sufficient skills and time to write a function to perform this task automatically. This function would have to operate within a UNIX environment, which added another element of complexity, as much of the development of this web was done on a PC.

I created the index by dragging in a list of all HTML files within the finished ICS web (Win95 DOS command, dir *.html /b /s > ..\dir_list.txt), and then performed a global replace on "C:\" and ".html" to add HTML tags. I manually typed in each link name.

Whoops

At this point, I was importing existing information from the current ICS web site, and I noticed the "student organisation" section. This part of the web has been placed under 'inside the department', but does not appear in any of the previous planning or design processes. This is an excellent example of the continuous development process, as described by John December. :-)

Browsers

The Research and Development of the Informatics Web

The Development of the ICS web

Next Page