On Fri, Aug 12, 2011 at 1:10 PM, Greg Dritschler <greg.dritsch...@gmail.com> wrote: > I thought this over again, and based on everything Mike has said, I think my > proposed fix is wrong. pushDownEndpointReferences() is adding targets from > a component reference to its "leaf" component reference. This should always > be an add operation, never a replace. That's why ASM_5023 is now failing. > > nonOverridable only applies between the composite ref and its promoted > component ref within the same composite. So far I haven't found the code > where the targets are resolved between a composite ref and its promoted > component ref. > Greg > On Mon, Aug 8, 2011 at 10:46 AM, Simon Laws <simonsl...@googlemail.com> > wrote: >> >> i've created https://issues.apache.org/jira/browse/TUSCANY-3912 to >> cover this. From the set of four failures I originally saw on a full >> build three are some bogus HTTP endpoint failure. The remaining is >> TUSCNAY-5023 which we need to look at. >> >> Simon >> >> -- >> Apache Tuscany committer: tuscany.apache.org >> Co-author of a book about Tuscany and SCA: tuscanyinaction.com > >
Greg, I think you're having the same doubts that I was having when I suggested that the previous example was wrong. I concluded that they were correct after all and for me it hinged on what you say here "nonOverridable only applies between the composite ref and its promoted component ref within the same composite". I read the following part of the assembly spec twice: 862 • nonOverridable : boolean (0..1) - a boolean value, "false" by default, which indicates whether this 863 component reference can have its targets overridden by a composite reference which promotes the 864 component reference. First time I read it I thought it was being very specific in talking about "composite reference". i.e the composite reference declaration in the same composite. That's when I started to worry. The second time I read it, along with what Mike had said, I came to the conclusion that it meant the composite reference in the wider sense, i.e. as configured by the top level component. Does that make any sense? Simon -- Apache Tuscany committer: tuscany.apache.org Co-author of a book about Tuscany and SCA: tuscanyinaction.com
