I think anyone interested in implementing, not just testing, RDA should be paying close attention to what Deborah Tomaras is saying. The difficulty of implementing RDA across our large and diverse cataloging community goes up considerably without some kind of compatibility convention to preserve the LCNAF as a single, consistent, integrated file for authorizing name headings. Such a convention eased the earlier transition to AACR2, and would serve us well in this case.
Specifically, the convention would state that name and title headings on bib records coded "LDR/18=a" (AACR2) or "040 $e rda" are in principle authorized by the LCNAF. Over time, authorities in the LCNAF will be revised from AACR2 to RDF, differing AACR2 forms can be retained in 4XXs as pre-AACR2 forms were, and the task of managing this transition will be a lot simpler.
I accept that things will be confusing during testing, and acknowledge that LC cannot publish decisions about the test until the test is completed. But perhaps the Coordinating Committee could communicate that a compatibility option has not been ruled out if the decision to implement RDA is made.