Hi, I hope to work on this stuff (including support for RFC 0172) pretty soon.
Regards Carsten 2012/4/13 Robert Munteanu <[email protected]>: > Hi Caspar, > > Thanks for the idea, I did not consider ASM. Since it reads bytecode it will > be able to access those annotations and that is confirmed by the docs. > > Thanks. > > Robert > >> -----Original Message----- >> From: Caspar MacRae [mailto:[email protected]] >> Sent: Friday, April 13, 2012 5:32 PM >> To: [email protected] >> Subject: Re: Proposal to make org.apache.felix.scr.annotations >> available at Runtime >> >> Hi Robert, >> >> Apologies, I didn't read the spec in detail. As it's for unit tests >> you >> could access the runtime annotations with ASM? >> >> cheers, >> Caspar >> >> >> On 13 April 2012 14:03, Robert Munteanu <[email protected]> wrote: >> >> > Hi Caspar, >> > >> > Thanks for your reply. I've glanced over >> > http://www.osgi.org/download/osgi-early-draft-2011-09.pdf and noticed >> > that the annotations are defined with >> > @Retention(value=RetentionPolicy.CLASS) . I've also found that the >> same >> > will happen with the SCR annotations from Felix in the next release ( >> > https://issues.apache.org/jira/browse/FELIX-3247 ) . >> > >> > That means that the annotations will not be available at runtime, so >> I >> > will not be able to configure components by myself in unit tests. I >> assume >> > that Felix annotations will follow the official ones, so it makes >> little >> > sense to argue for making them available at runtime. >> > >> > Robert >> > >> > > -----Original Message----- >> > > From: Caspar MacRae [mailto:[email protected]] >> > > Sent: Friday, April 13, 2012 3:48 PM >> > > To: [email protected] >> > > Subject: Re: Proposal to make org.apache.felix.scr.annotations >> > > available at Runtime >> > > >> > > Hi Robert, >> > > >> > > In a broader scope this will happen for RFC 0172 Declarative >> Services >> > > Annotations (runtime processing of SCR components), however this >> brings >> > > with it standardised annotations. >> > > >> > > cheers, >> > > Caspar >> > > >> > > >> > > On 13 April 2012 13:33, Robert Munteanu <[email protected]> wrote: >> > > >> > > > Hi, >> > > > >> > > > I've noticed that the annotations from >> > > org.apache.felix.scr.annotations >> > > > are declared with @Retention(RetentionPolicy.SOURCE) . I think >> the >> > > > following use cases would benefit from the annotations being >> > > available at >> > > > runtime . >> > > > >> > > > 1. Configuring component instances outside a OSGi container >> > > > >> > > > I often develop components which I want to test in plain JUnit >> tests >> > > > rather than in a container. These components have references >> which I >> > > need >> > > > to satisfy in order for the unit test to run. Since the >> annotations >> > > are >> > > > available only in the source file, I am unable to inspect the >> > > declared >> > > > references and inject them in a generic manner. >> > > > >> > > > The alternative is to manually configure each reference, but that >> > > gets >> > > > tedious pretty fast. >> > > > >> > > > 2. Processing SCR annotations in a JSR-269 annotation processor >> > > > >> > > > Having the annotations available to the processor ( strictly >> speaking >> > > only >> > > > RetentionPolicy.CLASS is required ) would allow much simpler >> > > integration of >> > > > SCR processing since the processor would only need to be >> available as >> > > a JAR >> > > > on the compile classpath, rather than a separate maven plugin >> > > invocation or >> > > > explicit ant task call. >> > > > >> > > > Above the other two use cases, having annotations at runtime can >> > > generally >> > > > ease integration with other systems which can inspect the >> component >> > > > configuration at runtime. >> > > > >> > > > The single downside that I see is that the annotations will >> increase >> > > the >> > > > file size of the deployed bundles ( although I don't think it >> will be >> > > a >> > > > huge increase ). >> > > > >> > > > I'm willing to prepare a patch, but would like your opinion >> first. >> > > > >> > > > Thanks, >> > > > >> > > > Robert >> > > > >> > -- Carsten Ziegeler [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
