GoogleCodeExporter
commented
9 years ago I think is time to move away from Suco to the more efficient, less verbose,
google powered gin library for dependency injection. I want to start a
discussion of how can be done in a good backwards compatible way.
Suco provides two different areas of services: dependency injection and events.
The change will replace dependency injection with gin and the event-related
classes with the HandlerRegistration gwt's built-in class (and implementing an
event bus instead of lot of little listener collections).
1. General approach
The suggested approach will be deprecate all Suco related classes and promote
the use of new methods. This has big impact over the listeners. In order to
provide backward compatibility a copy of modified Suco classes will be shiped
inside the emite-library so the suco library will be no longer necessary. This
modified classes will have the same functionality than before, but using the
new gin/HandlerRegistrarion infrastructure.
2. Dependency injection
The module system of gin is completely different from suco, so I think it wont
be easy (or probably impossible) to keep backwards compatibility when creating
new modules (so modules should have to be refactored to the gin way of IoC)
Also, other big difference from Suco is that gin doesn't use container
explicitly. That means that you won't be able to add new instances for
injection at runtime class (I think its a minor problem).
The principal methods of the Suco facade will remain the same (Suco.getInstance
and Suco.getProvider) but a automatic-gin-injection will be promoted.
3. Event and Listeners
Since version 1.5 (I think) gwt comes with HandleRegistration class that makes
easy to implement EventBuses and ListenerCollections. Is the same class that
uses the Widgets to fire its events.
The impact of move to HandlerRegistration is BIG, because we will have to
change EVERY listener in our code. The good news is that is quite easy to
maintain backward compatibility wrapping the new classes with the old interface.
4. Conclusion
I think gin is much better IoC library than Suco and move to it worth the
effort. Probably generated code will be smaller and for sure will be faster. I
think it will be a future investment.
What do you think?
Original comment by danigb on 9 Jun 2010 at 3:52
- Changed state: Accepted
- Added labels: Priority-High
- Removed labels: Priority-Low
Original issue reported on code.google.com by
danigbon 20 Jan 2010 at 6:12