Re: [Geotools-devel] FactoryRegistry Refactoring for Java 9 Compatibility

2017-09-25 Thread Nicolai Parlog
Hi Jody. > We should also be able to detect implementations that still implement > ServiceProvider, log a warning, and ignore - rather than fail. Yes, I think this makes a lot of sense. I updated the proposal accordingly. I'm particularly interested in your opinion about returning streams and k

Re: [Geotools-devel] FactoryRegistry Refactoring for Java 9 Compatibility

2017-09-22 Thread Jody Garnett
Thanks this looks great. The key design question is the API break from Java ServiceProvider to GeoTools Factory. I think this change is important enough to Break API; in part because we were broken by Java 9 restrictions - by changing our API we are really clear to downstream applications that the

Re: [Geotools-devel] FactoryRegistry Refactoring for Java 9 Compatibility

2017-09-22 Thread Jody Garnett
Wow thanks Nicolai, I will look forward to reviewing this today. -- Jody Garnett On 22 September 2017 at 00:21, Nicolai Parlog wrote: > Hi! > > I finally found some time to accompany my pull request[1] containing the > refactored FactoryRegistry with a formal proposal[2]. These are the main >

[Geotools-devel] FactoryRegistry Refactoring for Java 9 Compatibility

2017-09-22 Thread Nicolai Parlog
Hi! I finally found some time to accompany my pull request[1] containing the refactored FactoryRegistry with a formal proposal[2]. These are the main points: * FactoryRegistry extends javax.imageio.spi.ServiceRegistry, but on Java 9 the latter throws an exception if extended by non-JDK class