b3 status [message #538542] |
Mon, 07 June 2010 21:50 |
Henrik Lindberg![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=henrik.lindberg%40puppet.com) Messages: 2509 Registered: July 2009 |
Senior Member |
|
|
Hi,
I have been a bit quiet with status updates for a while - things have
been a bit busy around Eclipse release, and I have also been working on
some other things intermixed with b3 development.
I just checked in a new version with the following changes:
- syntax is now more consistent (some statements invented early had a
different style). Some statment, while working in simple cases were very
difficult to read when the same keywords showed up with different
meaning in what looked like a never ending list of words... - hard to
know when one things ends and the next starts, etc.
- some constructs made it difficult to provide good proposals for code
completion (if anything is possible, the list of choices is long, and
gives no clue about what to select).
- the updated syntax uses more and longer keywords in the declarations
of things (will follow up with examples)
- b3 DSL is now based on Xtext 1.0, and I have started to make use of
the various services (in many cases, just adding support for one or two
'trailblazing' features to make it easy for others (and myself), to
build out "full support for everything"...
- More validation and sanity checks have been added
- Quick fixes available for some statements
- Formatting source implemented for some statements
- More functionality at edit-time (now variable references are handled
when editing (as you expect it to work) instead of just catching things
at runtime. (Implemented with a declarative scope provider using xtext
polymorphic dispatch - I am glad I did not implemented this until I
learned how to use this great mechanism).
- Type inference not yet moved (up next).
- Property values are now referenced with ${name...}, this to make it
possible to know where the property name ends without requiring white
space after the property.
- I have also added a mechanism that allows scope providers and type
inference to find elements that are known by the runtime/engine (this is
done by having a special read-only EMF resource that reflects the same
things in a model as what the engine introduces in runtime.
Regards
- henrik
|
|
|
Powered by
FUDForum. Page generated in 0.03045 seconds