| Re: [xtext-dev] Persisting xtext document information |
| Hi Peter, yes, it's possible to store Xtext as ASTs in a model repository, but I wouldn't recommend it, because the AST, as the name Abstract Syntax Tree suggests, does not contain any information about whitespaces, line wraps and comments in your model. Therefore, you'd have to persist them separately. You might want to look at the OCLinEcore Editor from Eclipse OCL. It can open Ecore files (XMI) and make the content available as text in an Xtext editor. Are you aware that Xtext ships with two serializers? The old one "ParseTreeConstructorFragment" has been superseded by the "SerializerFragment". The new serializer will not produce a different syntax as long as it still has the old whitespace. And even with lost whitespace, the serializer will only do changes that your grammar explicitly allows: For example, if you use unordered groups, the order of the elements is usually not stored in your AST, and therefore the serializer can not recreate the old order if no whitespace information is available. If you want to go further in this direction, you should also think about how you want to persist documents with syntax errors. Since these documents can not be parsed completely you won't have a complete AST. Consequently, if you only persist the AST, the user will loose all parts of the document that are involved in syntax errors. If Syntax errors are so severe that the parser can't recover, the user will use the complete tail of the document. regards, Moritz On Jan 18, 2013, at 10:44 AM, Peter Pedron <Peter.Pedron@xxxxxxxxxxxx> wrote:
|