dvd+rw-tools were teached to obtain sense information from kernel manually, so they fully work on 5.2.1. cdrecord relies on working autosense and thus don't work on 5.2.1.
I would call this behavior of dvd+rw-tools a bug as the CAM standard requires auto-sense to work. If FreeBSD is fixed, the modified dvd+rw-tools will stop working in case it correctly evaluates error codes.
I can assure FreeBSD users that dvd+rw-tools won't stop working the moment the kernel bug in question is fixed. Nor would you have to recompile dvd+rw-tools after it's fixed. dvd+rw-tools detect the kernel deficiency at run-time and are designed to work in either particular situation.
Can one criticize dvd+rw-tools for so to say sweeping the bug under the carpet, which might have indirectly complicated life for other applications and their developers? Well, one could *if* it was done silently. But it wasn't. Thorough problem analysis was presented to the public on a FreeBSD list (see http://lists.freebsd.org/pipermail/freebsd-ports/2003-December/007424.html) the moment this [non-intrusive] workaround was introduced in dvd+rw-tools 5.15.x update. The problem description explicitly mentions that *all* recording applications are affected, thus depicting how serious problem is. A.