I'm curious about this. I've never found a situation where ETTR (defined as the RAW data for non-specular highlights within 1/3-stop of clipping, but not actually clipped) was not the best digital exposure strategy. Exposing in this way always results in the lowest noise, greatest captured DR, most accurate colors, and greatest processing flexibility. What exception have you discovered?
Choosing an ETTR exposure over a shorter exposure won't ever do harm (that I can think of), but sometimes the ETTR exposure may not be long enough to get the best result. Sometimes exposing to the right will leave your subject under-exposed. If the highlights are relatively unimportant and you need to capture the image in one exposure, exposing for the subject might be preferable to ETTR.