Frazer Melton

Have had time to think about the design of the CMS after working with Drupal for a couple of weeks.

What has struck me now is that the idea of a single CMS is limiting. Just as there are different types of websites, there needs also to be different types of CMS to handle the content of those websites. Also, the CMS itself can be on the server machine or on the local machine - or possibly on both.

From my point of view - and in terms of the sites that I manage - there are the following classes of websites:

# Validation

Validation is one of the trickiest areas to manage in any computing application. The idea of validation, in principle at least, seems fine, yet when you come down to putting into place the procedures that are required to perform it, you come to realise that it is more involved than it appears on the surface.

# Permissions

This is an easy one to decide: just borrow the permissions system from UNIX/Linux. This is reasonably simple, straightforward, tried and tested. The basic idea of three groups of users - the owner, the group, the rest of the world - is applicable to most situations. On a typical business site employees can be divided into logical groups - accounts, sales, production etc. - with access to their own area of the site. The rest of the world - i.e. anonymous visitors - can see anything which is not company-confidential.

This may seem an odd one, but it strikes me as a good thing for the CMS not to make any use of HTML. "No HTML", you say, "What is the point of a CMS if it doesn't output HTML??".

On the surface a CMS is all about managing page content and formatting the content using HTML and CSS so that it can be presented to the visitor. However, the way I see it, is that the CMS's true task is to manage information flows and stores.

Boiled down to its essentials, a CMS tasks can be seen as:
1.Input some data
2.Process that data in some way
3.Output some data

# Processing & parameters

Websites consist of separate pages which are loosely linked through the logic of the menu system. The art is to make the look and feel of all of the pages similar enough to make the visitor perceive the site as an integrated whole. Unlike a conventional computer system, with a website the flow is per-page.

Recently I have been trying out the Drupal content management system for a number of my own and my clients' websites. I have been looking for an easy way for the customers to be able to update their own sites. They would like to be able to change product details, update prices and upload pictures of their products and also put some news about new products, promotions etc.

I have been using open source software now for what must be the best part of eight years. Firstly with things like the Apache web server, PHP etc. A couple of years back I went totally open source as well with my computing equipment - and haven't regretted it once.

Now, I've come to the conclusion that it is time to give something back. The open source software has been very useful for me and now I've decided to release as open source some or all of the software I have developed for my own purposes.

Have just read an interesting piece on computable.nl about a fascinating new way of finding data. The piece is about the rather excellently named Stratos Idreos - a researcher at the CWI centre in Amsterdam.

His new method, which he calls "database cracking" attempts to make searches faster by sorting and storing data anew each time a search is performed. This is in contrast to the more usual method of using index tables to store data.

So, England is out of the World Cup. This is a huge shame as the squad has a number of extremely talented players. I used the term "squad" rather than "team", because - clearly - the English side weren't functioning as a team.