I had someone ask me the above question about "Dialogue" by Eugene Bozza:
The piece ends in octaves, written Ab in the 1st part and G# in the second part. It's definitely a little strange at first glance, but I pointed out to them how (especially in the last two lines) the first part uses almost exclusively flat accidentals, while the second part uses sharps. I talked about how, for the ease of reading, it makes more sense to have that last note match the 'mindset' that the player is in (sharps mode vs. flats mode) than to have the last note be written in the same way.
I think I generally got my point across, but I never formally studied part writing, so I figured I'd also ask here -- is my answer generally right? Are there general principles for when ease of reading overrules things like writing enharmonic notes in different parts?