Why this book¶
I came back to XML in 2026. That it was still everywhere was no surprise — Office documents are zipped XML, SVG is XML, plenty of config is XML, and I'd always known it. What did surprise me was XSLT. I'd filed it under "the 2000s," a thing you used once and moved on from. Instead I found it alive and genuinely useful — driving e-invoicing pipelines and transforming documents at scale. Official European standards bodies ship XSLT: the EN16931 validation artefacts come as stylesheets, and you are expected to run and handle them. And the material for someone returning to all this is rough going. Most of what's online is either a tedious from-zero tutorial that spends three screens on what a tag is, a spec page written for compiler authors, or a 2004 blog post half of whose advice is now obsolete — verbose, dated, and aimed at someone who's never seen any of this before. And that's before the presentation: the pages are often annoyingly formatted, buried in interstitials, or plastered with advertisements you have to read around.
There's a reputation to address, too. XML, the popular imagination holds, is the domain of fifty-year-old Java developers who listen to Dire Straits and remember when SOAP was the future. And — fine — that's largely true. But it turns out the door is unlocked and you can still join; the taste in music is optional. No one will check your age at the angle bracket.
This is the manual I wanted instead. It assumes you can program and you've seen angle brackets before, so it skips the throat-clearing and shows you what's actually going on. I don't expect you to follow along at a keyboard, copy-pasting as you go. I expect you to crash on the couch with a tablet and read it — every concept carries a small, complete example precisely so you can see how a thing works without having to set anything up to believe it.
The book is built to be jumped into. Each section stands on its own: start with the XSLT tutorial if you want a guided path, or go straight to real-world e-invoicing if that's the problem in front of you. XPath, XQuery, and XSD are there when you need the underlying machinery, and Schematron when you need rule-based validation on top of it.
The real-world tour deserves a word of warning. Some of what it covers — SOAP, XSL-FO, a few others — is more of historical interest than state of the art, and I don't pretend otherwise; you may never write a line of it. They earn their place because they're where the culture of XML shows itself most clearly: how the people who built these things thought about namespaces, extension, validation, and versioning. Read them less as "tools you need today" and more as field notes on a way of thinking that still runs under a lot of what you do.
A note on how it was made: I wrote it with Claude Code. That's part of why it can be both broad and current — covering the whole processing stack and the places XML actually shows up today — while keeping every example real rather than illustrative. It's built with Zensical and published on GitHub Pages. It's meant to be read, not consulted — written by someone who was standing exactly where you are.
This is a living book. I'm still finding my way back, so I'll keep adding material as I go — new corners of the stack, new vocabularies, more old things I only just learned. Check back; it grows. And if there's a corner of the XML world you find interesting — or wish someone would explain — tell me: suggestions are welcome on GitHub.