To the OP:
Hello François.
Short answer:
You need to do your own testing on your own camera. My workflow has certainly changed.
Long answer:
Six years ago when I was using a Canon camera I would religiously ETTR because I had to. The noise in the shadows could get very ugly. And the OOC jpegs weren't that useful.
Times have changed.
Because it is not an optical viewfinder, with a Fuji camera I use the viewfinder to gauge the exposure. When I can see detail in the highlights I press the shutter and let the shadows fall where they may. I don't refer to the histogram that much.
The OOC jpegs are fine for medium size prints. This is a great time saver when there is rush to get some initial work out.
Most of the files I send to Photoshop have not had the exposure, highlight recovery, saturation and clarity sliders touched in the raw converter. And sharpening is set to minimum. With this sensor, mostly I want to remove sharpness, not add it.
To my eye the result is better tonality and colour. I don't find raw files to be as malleable as many make out. Oh, and I almost never need to use noise reduction on photos shot under 1600 ISO (Standard Output Sensitivity). The shadow detail is just fine.
I'm not talking about those who post here, but in general most people lose detail and fine tonality in the three quarter tones and above due to their desire to avoid shadow noise. Think white horses or wedding dresses. You see it regularly in photographs. Those highlights really matter to the eye in a full tonal range image.
The saturation and colour in the highlights is often off as well due to the exposure and highlight recovery sliders altering their balance. In PV 2012 Lightroom in particular can do unpleasant things to the three quarter tones and above, sometimes without even moving the highlight recovery slider.
Try this for yourself to see if it matters for your camera and your processing method. Render out a file in PV 2012 where the highlight recovery slider has been used. Now go to Camera Calibration _ Process and change it to 2010. Adjust the highlights using only the exposure slider and render.
Compare the highlight tonality in both images.
David
Edit: Some time ago when I tried the above I could see a marked improvement in the rendering of clouds. I tried it last night on an image of a white horse and the highlight recovery slider in the latest process version gave a better rendering. Who knows, maybe it's camera/image dependent or maybe I've gone mad.
Google brought up a link to Mike Johnston's article on tonal values in digital B&W
here. Although I wouldn't go so far as to agree that "almost all digital B&W is like drinking rotten pond scum".