My experience is related to multi-channel recording and reproduction of classical music using 64fs DSD as my format of interest, at this time. I use five full range SoundLab electrostatic speakers in an ITU alignment for my home listening, with a Sonoma DAW as the primary source. I rip every SACD I purchase and play them out of the Sonoma, so I rarely listen to the actual SACD. I also have the edit masters of projects I was involved with, along with the actual session tracks.
Comparing all of those gets me to these conclusions, again, all by listening:
The largest change, by far from the session tracks occurs in mastering. Mixing and editing are pretty benign for DSD classical music projects, involving cutting/splicing of elements, and track level adjustments. The majority of those tasks are either performed in DXD on Pyramix workstations, or in analog. The mastering process however shapes the sound to the producer/conductor's wishes. This may be no adjustment, like is typical of Channel Classics and some other labels, to considerable sweetening with EQ, and reverb, and IMO, the subsequent loss of spaciousness and detail cues.
Comparing either the ripped SACD .dff file, or the final edit master (the file that is then authored by encoding to DST for the SACD manufacturing process) to the then decoded DSD stream prior to any player digital manipulation or DAC conversion, shows no difference to my ears. This is an easy test to do on a Sonoma, since the SACD transport and stored ripped or edit master file can play simultaneously, and be switched between. The same is not true when comparing two stored files. Only one can play/advance at a time with a Sonoma. It's therefore for me, much more difficult and inaccurate to compare a ripped SACD with the edit master.
But I do like what Barry said in that thread for CD. Makes sense, I just can't reproduce it with an SACD.
I'm all for computer audio, which I think is the future for the High-Res recording business.