Ugh... I think I finally ironed some of the basics of the code involving interpretation of the information in FRX files (now that I've got the basic scafollding working right, I can start "fleshing out" the details). Spent a couple wracking my brains trying to figure out anything but trivial cases would yield consistent results. Then it dawned on me -- do what is likely happening internally with report editor -- do everything in terms of pixels. Ended up getting flashback of time working out weird screen positioning problems back in FPW (again, solution was to convert everything to pixels first, perform all the computation to get location in pixels, then convert that result to foxels. Working in foxels generally yields inconsistent or incorrect results).

Thank to all of you for the suggestions!