Columns

A unique take on today's technologies

• By Gary Barnett
• December 1, 2001

While some of the most vocal proponents of Web services are projecting visions of B2B nirvana and driving the evolution of standards like UDDI, developers across the globe are starting to think about Web services as a handy way to link loosely coupled systems using technology that doesn't bind them to a particular component model, programming language or platform.

And there are some great visions. A glance at white papers by some vendors could lead one to believe that Web services will change the world. In this new world, a little box can be installed in the corner of an office to negotiate with suppliers, find customers and run the business with little intervention.

Users are excited by Web services, but are also deeply suspicious of the vision. They are uncertain about the resilience of the Web, nervous about security, and not convinced the pieces will be in place for some time. Many of them are still wrestling with EAI, once hyped as an easy solution to integration problems.

Ultimately, the future of Web services will be determined by pesky corporate developers, and all the PDF and PowerPoint in the world won't stop them.

Raising the quality of the Linux debate
I'm not actually interested in the intrinsic moral goodness of one solution over another. Indeed, though this will cause offense to some, the technical superiority of one solution over another can be a moot point.

I've been watching the Linux phenomenon for a couple of years now. My clients live in the real, complex, difficult world of end-user computing, and they're coming to me for help, not hype. I can't afford take a subjective stance. First, my clients are too smart to let me get away with it. And second, they'd laugh at me and then go somewhere else for advice. I say this because whenever I say anything about Linux I get flamed by some group of advocates.

Sadly, the majority of the flames come from pro-Linux folks. I only have to say one vaguely negative thing and my e-mail box is full. I'm accused of being stupid (which I can live with) or of being a Microsoft double-agent (which irritates me).

I'm not only a strong supporter of Linux, I use it. I've installed it, tinkered with it, even done some development on it. Linux is great, and it's going to be even greater. But the growth in its popularity, and the pace with which Linux improves technically, is being held back by the dreadfully low quality of debate.

I want to engage in reasoned debate without fear of obscenities filling my in-box. I want to be able to state my strong concerns about Linux's credentials as a desktop operating system. I want to engage in a sensible discussion about these concerns, and discuss what improvements could or should be made.

I want to state my firm belief that Linux is very close to offering a viable alternative to the "Big Unix" offerings from the likes of HP and Sun. I want to challenge claims about the Total Cost of Ownership that seem to forget that OS cost is usually a small percentage of the "Total" cost.

I want to be persuaded by reasonable people when I'm wrong. I want a chance to persuade reasonable people when I think they are wrong. Perhaps if more people felt able to state their misgivings and point out areas for improvement the Linux community might learn something. This isn't too much to ask.

XP: Powerful medicine
Extreme Programming (XP) is getting lot of attention. It's billed by some as the definitive solution to the need to implement business solutions quickly. Users who believe the ambitious claims made about XP are in for a nasty surprise. XP embodies some superb ideas and goals. Pairing developers can result in an astonishing boost to productivity. But it depends on the matched developers.

The healthy obsession with re-factoring results in amazingly slick and efficient code that can sometimes descend into a strange, never-ending quest for nirvana. A minimalist approach to documentation certainly gives it a chance to remain up to date. But strike the wrong balance and new team members will take an age to get up to speed.

XP is powerful medicine that needs to be administered in measured doses. It works best with an able project leader, a relatively small team that is committed and skillful, and a well-defined problem.

XP can be applied to larger teams, and to much less-defined problems—but in the same way that anyone can drive a car quickly in a straight line, the curves require considerable skill. My advice is simple. Create a sensible framework. Be clear about the documentation needed. Select a strong project manager with the people skills to keep the team motivated. Pair up the right developers. And don't forget that your goal is not the perfect subroutine, it's the business solution your customer needs.