Most enterprise knowledge is glass. It holds its shape and makes sense up close, and nothing connects it. The definition of a customer lives in a database schema, a Java class, a wiki table and three spreadsheets, and each one is a projection of a model nobody ever wrote down. Ask who owns the definition and you get silence. An AI assistant can't be fed it either.

People don't skip structure because it is useless. They skip it because it costs more to keep than it pays, and a spreadsheet never asks anyone to adopt anything.

This book argues that the cost has changed, and shows a runtime built on that argument. The Nasdanika SDK Runtime is a set of modules on three foundations: the JDK, EMF core and the OpenTelemetry API.

• Models as the API. Data and operations are written as text in Xcore and generated by a Maven build, with no Eclipse IDE. Models can be saved as XMI, JSON, YAML or binary, and their operations can be called by a command line or an agent.
• Capabilities. Code says what it needs and the framework finds it among the modules present, by type, down to the generics.
• Telemetry as part of the code. Watch work live with no OpenTelemetry SDK and no backend, or save it as a model next to the models it changed.
• Volumes. A model, a running JVM or a remote system can be presented as files, served over HTTP or mounted as a drive over WebDAV.
• A command line assembled from contributions. Adding a dependency adds commands.
• Models that ship with it. Meta, the JVM, HTML, Bootstrap, TypeScript, React, and diagrams with Draw.io, ELK and SVG.

A solution built on it runs in one JVM, on a laptop or a single virtual machine, with no cluster, database server or message broker to ask permission for.

Who it is for

• Architects who want more structure with less effort. You don't need to write Java to get value from the first part.
• Java developers who want an observable, testable foundation with few dependencies.
• Researchers and R&D teams who need to try an idea before anyone will provision infrastructure for it.
• AI assistants. The terms are used consistently, the rules of thumb are collected in one place, and every listing compiles.

What it is not

It is not an introduction to modeling or to Java. My sister, who knows modeling a bit, says of this material that someone walking in from the cold will not get it. She is right. If the problems are not yet yours, the book will keep until they are.

How it keeps you honest

Every listing is included from example modules that are built and tested before the book is rendered, so a listing that no longer compiles stops the build. Status labels say what is AVAILABLE, what is a SKETCH and what is PLANNED.

This book is the menu: the why, the what and a high-level how. The recipes are the Nasdanika Zephyr Books, in-depth manuals for the modules that need more than a chapter. Draw.io and Groovy DSL are on Leanpub in a bundle with this book, and all of them are free on the runtime site and Maven Central.

The book follows the runtime's quarterly releases. The text is licensed under CC BY 4.0 and the examples under EPL-2.0.