I've never even toyed with VNAV alf, so I can't answer your question directly. Best I can offer is that, as part of our GNSS Implementation program, I'll be investigating the viability of Baro-VNAV procedures. These procedures will undoubtedly incorporate coded waypoints in the database that have been derived from survey data and verified by flight checks. Thus, the resultant procedure will provide a tabulation of altitude versus distance that can be used for continuously monitoring the descent.
As for NPA procedures where DME is not available, I'll go out on a bit of a limb here and say that - subject to a few considerations - it should be possible to fly the NPA as a CANPA. Had we not decided to go with GNSS implementation here, I would've gone to CANPA using a GPS reference point. I justify this on the basis of a thorough RAIM study conducted here, some years ago, by the Volpe Institute and a study done in Australia (and a few other places), which showed conclusively that the fix tolerance for a GPS-derived distance fits well within the final approach protection area for a Pans Ops designed NPA.
Therefore, a FAF could be fixed at a distance from the reference waypoint and, again, a "altitude versus distance" tabulation could be provided for pilots to monitor the descent.
I think it would be most unwise for pilots to try their own version of CANPA on a NPA without DME, unless the specific State's regulator has done their homework in the same way as we have. And then, they'd have to provide a formal approval process of some sort. Without any of that, pilots have little choice other than to use "dive and drive" on a single navaid NPA because they will never really know exactly how many track miles are available for descent on final approach.