looks OK Joe On Nov 26, 2013, at 5:27 PM, huizhe wang wrote: > > On 11/26/2013 1:59 PM, roger riggs wrote: >> Hi, >> >> I looked at that twice also. java.time had a similar situation. >> >> To get to the TCCL you need to call ServiceLoader.load(type). >> >> The FactoryFinder:348 findServiceProvider method should test if cl == null >> and call the original ServiceLoader function. > > Yes. I forgot that part. load(service, loader) treats null as the system > class loader was one of the topics in the discussion of the spec. > > Joe > >> >> Roger >> >> On 11/26/2013 4:37 PM, Alan Bateman wrote: >>> On 26/11/2013 18:38, huizhe wang wrote: >>>> Hi all, >>>> >>>> Here's revised webrev, as Alan suggested, including the implementation. >>>> >>>> http://cr.openjdk.java.net/~joehw/jdk8/jaxp16MR/8028822/webrev/ >>> So is this handling the null case correctly? It is spec'ed to use the TCCL >>> but it looks like it's going through to ServiceLoader.load where null means >>> the system class loader. Maybe this needs to wait for tests as that it >>> really the only way to check this. >>> >>> -Alan >> >
Lance Andersen| Principal Member of Technical Staff | +1.781.442.2037 Oracle Java Engineering 1 Network Drive Burlington, MA 01803 lance.ander...@oracle.com