More Detail:

• NRL Rule Editor
• Java Generator
• Component Overview

NRL Mapping Editor

The NRL mapping editor lets users map between UML models and XML Schema. In practice, this is used for common transformation tasks, for example:

• XML to XML
• CSV/Fixed Width/SWIFT to or from XML
• Java objects to other formats

Features provided by the editor include:

• Control of individual attribute value translation using NRL
• Control of deep structure generation / repetition using NRL for "generation rules"
• Model-sensitive: auto-complete is available at all times
• Incremental validation when the file is saved
• Model caching: mappings are revalidated when the models change
• Mapping repair: when a model element is deleted or changed, repair functions are activated to synchronise the mapping. This helps with upgrades and greatly simplifies regression testing.
• Works at a semantic level, not at the level of syntax - analysts will not have to deal with date or value formatting
• Export of mappings to Excel for review purposes
• Trace tables between source and target models, for end-to-end completeness review purposes

Requirements and deployment:

• The editor plugs into Eclipse version 3.2 or higher
• IBM Rational® EMX model files are supported
• Java 1.5+ is required for the user interface

What is Different About This?

There are plenty of mapping tools on the market, and developer tools like XSLT or plain code are also frequently used to perform transformations. Here are some of the things the NRL mapping editor does differently to some or all of them:

• It is model-driven; the analyst always works against a model, never against syntax
• Its rule language is NRL: you do not fall back to code or XPath whenever you have an expression to write
• It shows the target to be generated on the screen, and avoids "spaghetti-junction" source/target diagrams
• It is fully object-oriented and deals well with XML Schema.
• There are no relational techniques involved that would flatten the model and complicate the mapping
• It handles deep structure and repetition well, as that is its native representation
• No runtime requirement: the Java generator produces plain Java code - no engine is required, and you are not dependent on the editor after deployment
• It is not "positional", hence insertions into your model do not make your mapping obsolete.
• Advanced XML Schema: handles substitution groups, namespaces, model groups, anonymous type derivation and other advanced structures transparently, and shields the analyst

And here is something else that may matter to you in the future, if not now:

• It can be analysed, as it is based on a language with a published, formal semantics. This can include future proof of correctness and completeness