A First Grammar

Here we highlight some information when walking through (Xtext 15 Minutes Tutorial).


In this section you will learn

  • How to model hierarchical data structures (things containing things).
  • How to model references (things referencing things).
  • How to identify model elements (by name).
  • How to model specialization.

Note: How to identify model elements in other hierarchical elements (e.g. packages) will be handled later.

Step 1: Create and run an Xtext project

Take your time for this tutorial now and continue reading afterwards. You may stop before "Second Iteration" where "packages" are added to the model.

Step 2: Questions

Check if you understood thw following points:

  • What is the role of the start rule of a grammar in a meta model?
  • What is the meaning of "Type: DataType | Entity;" in terms of inheritance? Explore the generated Java Interfaces for the classes Type, DataType, Entity in src-gen/org/example/domainmodel/domainmodel in your main project.
  • How do you typically add keywords to your language (like "entity")?
  • What is the difference between "x=Rule1" and "x=[Rule1]"?
  • How can you model the following?
    • "a named University aggregates named Students".
    • "a House is composed of Rooms".
  • Explain the role of "?" in the following examples:
    • (x=INT)?
    • enabled?='enabled'


  • Optional attributes can be defined using "?". Note: some types, like references, are null if not set. Others have default values (like an empty String for STRING, "0" for INT or the first enum value defined for enums).
  • Rules without attributes does not yield an object (e.g.: "Thing: INT"). You can force the object creation with the following syntax using curly brackets: "Thing: {Thing} INT".

Step 3: More examples

Can you interpret the following snippet:

// ...
Model: customers+=Customer* computers+=Computer* owns+=Own*;
Computer: 'computer' name=ID;
Customer: 'customer' name=ID;
Own: 'the' computer=[Computer] 
     'is' 'owned' 'by' customer=[Customer] 
     'since' date=STRING;

What changes, when we define the 'Model' differently:

Model: (customers+=Customer computers+=Computer owns+=Owns)*;

Step 4: Editor

In addition to the text editor, you can also edit the model with a tree editor: open the file with "Open With..." - "Sample Ecore Model Editor".

When editing the model graph, the model text is changed accordingly: see the Xtext "Formatter" in the online help of Xtext (in the internet or within eclipse):

Step 5: Visualize the meta model

The ecore model deduced from the grammar can be visualized: see (Mooij et al, 2017a), section "Optional: Ecore diagram" („Initialize Ecore Diagram ...“).

Moreover a syntax tree can be rendered from the grammar: see (Mooij et al, 2017a), section "Optional: View Diagram of Xtext Grammar" (Window/Show View/Other.../Xtext/Xtext Syntax Graph).

Optional Step 6: Adapt you unittests

Adapt your unittests.

  • Check some individual model entries (e.g., a name or a list size).
  • You can explore the model structure with the debugger in your unittest (just place a breakpoint and run the test).