Intel joins the club

By Michael Kay on December 09, 2009 at 02:18p.m.

I see from Twitter that Intel have now announced availability of a beta release of their XSLT 2.0 processor. Excellent news! Like IBM, they've chosen to bundle the product as part of a high-end web services offering (SOA Expressway in their case, Websphere in IBMs) rather than giving it away as open source or selling it at commodity prices. This is a good decision - it reflects the high value of the technology, the substantial investment that goes into it, and the benefits that it delivers to customer. It also happens to be a positioning that's well differentiated from Saxonica's.

I'm looking forward to hearing reports of how the products compare. Saxon is running in some pretty high performance sites (like Highwire, see, with 1m requests per day) with a very modest price tag, so Intel and IBM are going to have to deliver some superb performance to beat that: assuming that they achieve this (and I know both teams are very capable) this should leave Saxon well placed to meet the needs of the 90% of users whose performance requirements are less extreme, while knowing that there is top-cover if they need it.

Some users have been reluctant to move to XSLT 2.0, despite its immense productivity benefits, because of the lack of support from mainstream suppliers. The entry of IBM and Intel into the market should assuage any anxieties they have, and this can only be good for Saxonica as the market leader. Application vendors too will be more inclined to move forwards now they can offer their customers a choice of underlying technologies. Except in the browser space, anyone who is still using XSLT 1.0 really should be planning their move forward now.

What does all this say about open source? Clearly, XSLT 2.0 has been too big a project for the amateur weekend coders, and too expensive a project for the big players to justify giving the results away for nothing. I don't know how you make a business case to spend lots of money on an open source product if you're inside IBM, Oracle or Microsoft. Clearly, on this occasion, no-one in any of those companies succeeded in doing so. It's much easier of course to persuade the decision-makers that a version 1 product is going to make a killing: you can invent numbers showing the projected size of the market, the proportion of it that you expect to win, and the service and spin-off revenue that will result. It's much harder to do that when version 2 comes around, which is almost certainly the real reason why Microsoft are doing so little. Meanwhile the Saxonica business model, combining a commodity open-source product with an enterprise-level product costing a few hundred dollars, is proving very successful.