Many thanks for your feedback! The main reason for this is that Epsilon
is agnostic regarding the types of models it handles, and for some
technologies (e.g. schema-less plain XML: http://eclipse.org/gmt/epsilon/doc/articles/plain-xml/) a program
doesn't even get to know the types of expressions involving model
elements (i.e. XML nodes in this example) until it's actually executed
against a specific model (i.e. XML file here).
Our medium-long term aim is to develop a hybrid solution where code
completion is provided where possible (e.g. for EMF models) but
metamodel-less technologies such as XML (and CSV, support for which is
coming up) are still supported. Needless to say, any contributions from
the community in this direction would be more than welcome!
Michael Zeising wrote:
> Has anyone considered building an Xtext-based editor for the E*L
> languages, yet?
> I think the framework could benefit a lot from a real code completion,
> hyperlinking and all the stuff you know from the JDT...