JSR 330 support

[GoogleCodeExporter]

I'm going to add this today.

The first controversial issue I've run into: should we coalesce @Named and 
@Named? It would be 
quite frustrating to bind one @Named and inject another; and the error message 
wouldn't be very 
easy to parse.

Original issue reported on code.google.com by limpbizkit on 7 Sep 2009 at 5:57

[GoogleCodeExporter]

I've made Guice's Provider extend the JSR 330 class. This should make it easier 
to interoperate with the new API. 
We could also change the code where we accept a com.google.inject.Provider to 
instead accept a 
javax.inject.Provider, but that would break binary compatibility.

As a compromise, I'll make it so that Guice:
 - always returns com.google.inject.Provider
 - accepts javax.inject.Provider, wherever it does not break binary compatibility.

Original comment by limpbizkit on 7 Sep 2009 at 7:52

• Changed state: Started

[GoogleCodeExporter]

From the commit log of r1077:

This checkin supports:
   @javax.inject.Inject to mark classes
   @javax.inject.Qualifier to identify binding annotations
   javax.inject.Provider injections
   A new Jsr330 utility class

Still outstanding:
   JSR 330 is more strict about where its annotations are applied. These
restrictions are not yet enforced.
   @Scope
   @Singleton

Notable change:
   When the user injects a javax.inject.Provider<T>, we canonicalize the
dependency, and treat it exactly as if they injected a
com.google.inject.Provider<T>. This is slightly weird! For example, these two
keys are equal:
Key.get(new TypeLiteral<javax.inject.Provider<String>>() {})
Key.get(new TypeLiteral<com.google.inject.Provider<String>>() {})
   This is convenient, particularly for SPI users. And it means we don't need a
linked binding between the Providers. But it could be quite surprising. I'm
still trying to decide the right long term solution here...

Original comment by limpbizkit on 7 Sep 2009 at 9:02

[GoogleCodeExporter]

r1078, r1079 and r1080 complete the implementation. I've opened issue 425 to 
address the @Named problem; I 
believe everything else is in place.

Original comment by limpbizkit on 8 Sep 2009 at 12:36

• Changed state: Fixed

[GoogleCodeExporter]

Why has this not officially been released ? Only 2.0 and a 3.0 snapshot are 
available...tehre is no 2.1.

Original comment by miroslav...@gmail.com on 6 Dec 2010 at 8:49