B> I definitely consider it a bug for the reason you mention. All of the B> Trinidad components do push and pop the EL context during B> invokeOnComponent. I believe that tree visiting does establish the B> context correctly, which also shows where the intent lies. I agree this is probably a bug. I have made the change and am re-running our automated test suite. If there are no errors, I'll take the liberty of modifying the spec and impl, attaching a patch, seeking review, and commiting it. I think this can reasonably be grandfathered in under existing JSF 2.1 changelog issue 868. [1] Ed [1] https://javaserverfaces-spec-public.dev.java.net/issues/show_bug.cgi?id=868 -- | edward.burns(a)oracle.com | office: +1 407 458 0017 | homepage: | http://ridingthecrest.com/ | 16 work days until German Oracle User's Group Conference