[XSL-LIST Mailing List Archive Home] [By Thread] [By Date]

Re: [xsl] Are there any free, fully-compliant XSLT/XPath 3.0 processors?

Subject: Re: [xsl] Are there any free, fully-compliant XSLT/XPath 3.0 processors?
From: Michael Kay <mike@xxxxxxxxxxxx>
Date: Sun, 27 Jan 2013 09:56:55 +0000

On 27/01/2013 08:09, James Fuller wrote:
Hello Roger,

Yes I see your situation ... when non techies are confronted by
choosing between XYZ technology and one has a cost and another
apparently has no cost ... they classically drift towards the zero
cost solution. I sometimes wonder how much 'impl' detail to give to
people making cost decisions ... e.g. there is always a cost
associated with adopting any software.

Indeed this is true, and not just for non-techies. My business is essentially built on providing a zero-cost technology that people can use successfully to build some pretty significant systems, and then when they can see they are getting value from it, providing extra things they can buy to stretch its capability a little further (*). It's much easier to justify spending a couple of thousand at that point in the cycle than at the point where you are deciding what programming language to use, and the net effect is that 90% of the user base are getting a free lunch because the free stuff meets their requirements adequately.

Inevitably, though, the further we go down the track in developing the standards and the implementations, the more we are catering for the minority who need more than the current baseline offers. Generally, when we add small usability improvements to the language, users will thank us but will not reach for their wallets. When we add major features like streaming or packaging, we're catering for a small minority of the user base, but the users who need it need it badly enough to be prepared to invest. If this means that the new versions of the standards will have slow uptake because most people don't need them enough to fork out for them, that doesn't worry me greatly, so long as the minority who do need the new features generate enough revenue to recoup the (substantial) investment.

The investment has to come from somewhere. It's past the stage where it can come from hobbyists (look at the attempts to create a 2.0 version of libxslt). There are limited opportunities for using the software as a lost leader to get revenue from hardware or from services. To me, the most natural and simplest business model is that the investment comes from the people who are getting most value from using the technology.

There is another problem, of course: W3C process requires that you can't create a Recommendation without having two independent implementations. There have been plenty of programming languages that became hugely successful despite only having one implementation; and with a mature language, many implementors want to hold back until there is a Recommendation. So this process could yet prove an obstacle, or at any rate a delaying factor. But we're more likely to get multiple implementations if implementors see some prospect of revenue.

(*) I think that accurately describes the average buyer. There's another class of buyer who takes the commercial product because they don't trust open source, but I think they're in a minority these days.

Michael Kay

Current Thread