I don't know where you get your facts
If you did know, you would not challenge them so often with your odd arguments 
I was talking about playback of DSD, which is the simplest bit stream to process (and understand). DSD is just a stream of 1s and 0s which nudge the sound level up or down one level, but at a very high rate. If a dac does not handle DSD natively, there must be a lossy conversion to PCM (unless the PCM operates at 2.8-MHz which I've never ever seen).
Even interpolation algorithms are lossy, as they try to guess the missing samples.
I agree with you that DSD is a pig of a format for editing and mixing. It is hard for editing because there is no absolute sound level associated with each one-bit sample, unless you count back to the start of the stream.
It is hard for mixing because there is no way to represent, for example, taking one-third of a bit and adding two-thirds of another bit and getting a sensible result while rounding to the nearest one or zero.
Maybe half my SACDs were recorded in high resolution PCM (for example Chandos with notionally 24-bit 192 kHz). Others claim to be pure DSD (for example, LSO Live) while the very best from 2L.no are recorded in DXD at about 385-kHz and sometimes as 32-bit floating point numbers.
Floating point is a real game changer in my opinion because each sample can be represented by fractional numbers rather than being rounded to a whole number.
Even the extraordinary DXD, as used by 2L.no, samples eight times less frequently than the lowest rate DSD. Can you hear the difference? Not if your dac has to be fed down-converted PCM