Doug McIlroy

% fortune -ae paul murphy

Nichievo: re-visiting a lost Linux opportunity

When I did my Nichievo series for in mid 2002 my intent was to use a fictional business scenario to illustrate the political considerations that come up when rebellious users want to give Linux a chance.

Since the incumbent systems people always fight user directed change and usually prevail, it sometimes makes sense to adopt a kind scorched earth policy under which they lose by winning. In other words, you sell the users a strategic vision built around a technology that isn't quite there yet knowing that the other guy will get a short term win out of denouncing your choices but then find himself competing with a vision he can't deliver.

In the Nichievo case the challenge was to create a reasonably secure means of processing acceptance orders - agreements by Nichievo to insure client receivables against defaults. The illustrative solution adopted applied Apache Cocoon and XML running on Linux -not because those technologies offered the best or simplest option, but because the choice combined what the users wanted to hear with the opportunity to build a grand vision for unified document processing while getting the immediate job done.

Imagine, therefore, that three years and lots dollars later the internal guys still haven't delivered something the users are happy with -imagine, in other words, that the strategy worked. Some of the key players have moved up the organizational ladder, user unhappiness has increased, and the incumbent's credibility has fallen further. The opportunity, as they say, is now -and the first question is whether the technology selection needs to be revisited.

I have one overwhelming reaction to the difference between Cocoon then, and Cocoon now. Fundamentally it seems to have evolved from science experiment to professional product - there is now a clarity of purpose and design simplicity that promises to make it a joy to work with.

The contrast with both Microsoft and Domino couldn't be stronger. In 2002 both looked like candidates, today they seem more confused and amorphous than ever. Please correct me if I'm wrong, but most of Microsoft's more interesting promises for XML, Jupiter, and the XML content management server seem to have survived only in press releases and Longhorn retractions. What's particularly interesting, however, is that the pieces that do exist, like XML structured word processing file locking and distribution control, seem to have suffered a form of regression to the mean - meaning that what works is not just a pretty basic subset of what was promised, but the particular subset requiring the least change from what came before.

I haven't downloaded and started up the new Cocoon technologies yet, but it's clear from the documentation, demonstration sites, and sample code that they've gone in the opposite direction: from a scattergun agenda including ritual bowing to market realities (like com) to a clear focus on doing one job as well as possible at each stage in a pipelined processing framework. Looking at it, I'd guess Doug McIlroy in particular, but also Thompson, Ritchie, and the other original Unix creators, would be proud to see one of their core ideas carried forward and expressed so well in a working application.

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.