The Zero Room

If you are planning a non-trivial website, this twenty-five minute presentation by Paul Albert, Digital Services Librarian at Weill Cornell Medical Library which compares Drupal, a free Web Content Management System and Application Framework, to FatWire, a proprietary Web CMS, should be required viewing.

Many of the arguments presented here in favour of Drupal also apply to any of a number of popular free systems, and most of the arguments against FatWire apply to any proprietary system. In this case the systems were evaluated for their suitability for use by a Medical Academic Library, but any situation where management of a great deal of information via the Web is required might be considered broadly comparable. The staff at Weill Cornell Medical Library found that:

• Drupal is and will be cheaper
• Drupal is robust, extensible, and enterprise ready
• Other departments in their college are already using Drupal, as are other colleges and libraries, and many are dropping equivalent proprietary systems
• Drupal supports "perpetual beta" (continuous improvement of a site)
• Drupal has a much more active support and development community, and "a culture of sharing solutions"
• Drupal has been paired with other technology in proven ways
• Drupal has a gentle learning curve (compared to other development environments)

It's a shade off a hundred minutes in length, but Lullabot Drupal podcast no. 38 is the most fun I've had since the last LugRadio podcast. A contentious litany of Drupal modules that have outlived their usefulness, or soon will, thanks to CCK and Views, I found myself almost punching the air in delight at how many of the more contentious declarations I agreed with. Just days earlier, I had been trying, and failing, to come up with a justification for the continued use for the taxonomy module, so it was gratifying to hear Jeff Eaton declare that in his opinion it's days are numbered. He pointed out that you can do most of what Taxonomy does using CCK text fields, but I'm surprised nobody talked about terms-as-nodes, using node reference fields, which to me seems an obviously good idea that allows you to retain a heirarchical taxonomy structure, "related terms" and what-not in a more flexible manner.

Moshe Weitzman has written a brilliant article about the site he put together for the New York Observer, eschewing the Scheduler, Scheduled Actions, and Workflow modules for a hand-rolled CCK and Views solution. He declares "CCK has arrived, and helps us build sites quickly, and change them easily". Hear, Hear!

On the topic of modules that should be led behind a screen and shot, I've just finished one site, and am painfully progressing through development of another using the Drupal E-Commerce modules, and I have vowed to never do so again. Which means that I either never do another E-Commerce site (which I wouldn't shed any tears over), use another set of modules which are bound to have similar problems due to the difficulty of devising a general solution to problems that are in the real world so very particular, or use CCK and Views. The last of these seems to me almost a viable solution. In theory, e-commerce modules should give you a convenient one-size-fits-all solution, but in practice I'm unlikely to find a customer whose needs can be met by anything out-of-the-box, so I'm either going to hack or extend somebody else's code, or spend the same amount of effort putting something together using CCK and Views. I'm willing to put money on the latter option being more maintainable in the long term.

Just got my copy of Pro Drupal Development in the post. Very exciting. The first chapter has just been published as an article on Dr Dobbs. I hope nobody minds if I quote at length from the forward by Dries Buytaert, because I think it says everything:

"... on January 15, 2001, I made Drupal available as Free Software.

Haven't blogged here for a while, so I thought I should mention for the record that I think this is a development of world-shaking importance, in order to in future gloat about how precient I am.

Earl Miles (aka. merlinofchaos on drupal.org), posted this demonstration of some JavaScript magic intended for the Drupal panels module.

We make a point of providing our customers with the tools to easily update their own websites, and encourage them to do so on a regular basis.  Web developers' magazine "A List Apart" has an article titled "10 Tips on Writing the Living Web", which is a goldmine of useful advice for anyone who publishes online.

A couple of irritating things fixed up:

TinyMCE has an entirely unnecessary "nowrap" attribute in it's toolbar, which can screw up your page layout. You should edit jscripts/tiny_mce/themes/advanced/editor_template.js to remove this.

Truckloads of gratitude, of the falling to one's knees and weeping kind, are due to Andrew Clover for his excellent JavaScript patch for Internet Eexplorer's buggy background image behaviour. I can't tell you how often this is going to save what's left of my sanity.

According to web designer Amber Simmons, keeping your readers' attention is a matter of earning their trust. Her tips for becoming a "faithful writer" are:

Do you know your web site has become dated and unmanageable, but are scared of a major overhaul? Perhaps you are right to be; even changes unquestionably for the better can alienate and anger some of your users.

In "the Quiet Death of the Major Re-Launch" usability consultant Jared M. Spool explains how even major web sites with millions of users can make radical changes over time that hardly anybody notices.

The VP leaned forward and asked me, "How do we orchestrate a re-launch on a site this big without upsetting our customers? Any change is going to be so dramatic that people are definitely going to complain. How do we do this?"

I leaned back in my chair, paused for a second, then dropped the bomb. "You don't, " I responded. "A re-launch is a very bad idea. I highly recommend against it."