I tried to explain it. Form my point of view, blender 2.8 series has holes in its workflow.
Following sentences will be caricatural but with a little bit of truth.
Community focused on UI, 3DView editor, modeling and rendering improvements as always.
Improvements in animation side came from Open Movies.
And the rest is struggling, let to GSOC students and often postponed.
Existing users will not produce a ton of documentation, tutorials and courses that could be tagged obsolete after 3 months.
And, anyways, to produce pertinent learning material ; they need some distance with new features.
One main issue is that basics of some modules can be modified, quickly.
After one year, modifier stack will be completed different in 2.90. And nobody expected that.
After GSOC on outliner, its usability will be incredibly boosted. But discovery of this improved usability will not be instantaneous.
The feeling of speed of penetration of a piece of information is relative to the mass of people to inform.
There is always a shift between the reality of software and how it is perceived by community.
That introduces a bias in description by commentators.
So, developers should kept an eye on the dates while they are reading users feedback.
The overall picture will always be unfair with new stuff.
But when basics of all modules will be settled for upcoming decades, the feeling of not knowing where to start your learning of Blender will fade away.
To be less caricatural, I will be more precised :
Fundamentals of workflow are settled for UI, Modeling, Animation & Rigging, Grease Pencil.
But Sculpt, Paint & Texture module does not have its brush management task done.
Nobody knows how new paint modes will look like and when work will start.
Because of design document, we have a vague idea of how it should work but no certainty it would be the case.
It is supposed to be well articulated with Sculpt Vertex Colors. But work is not done on both simultaneously.
Data, Assets & I/O module should have its Assets task starting soon, after Overrides will be finished.
Cycles module is waiting its Baking task.
EEVEE & Viewport module is waiting its Viewport Compositor task.
Nodes & Physics module just had Mantaflow integrated at beginning of the year and Particle nodes will be marked as experimental during a long period.
With so many modules missing a core feature or having recent features, there should be no surprise on the fact that current UX sucks at some steps of workflow, or that nobody feels perfectly comfortable.
I think that a shade is necessary, here.
Yes. Anybody may complain that you broke his workflow. And by definition, any change marks evolution from one status to another one.
But there is a difference between :
_changes breaking ancient workflows to propose an innovative alternative one that covers same field of uses.
and
_changes breaking ancient workflows to propose no alternative one or workflows covering a narrower or simply another field of uses.
If the use is still relevant and no more covered, that is an undeniable regression.
I think that most of time, changes in workflow should be grouped to avoid updates damageable to users.
When changes are expected every 3 months in same area, you can’t expected somebody to dive in in order to make a precise global documentation of it. That only happens when hype and frenesy of novelty is cooling down.
And branches system is a good system for that.
Basics have been tested in branch. After development in branch, people have a basis to document and a basic perfectible workflow but that is already able to deliver something.
What is really damageable is the update that brings nothing to users.
The one where there is no gain in productivity, just lost of time in learning a new workflow that does not deliver something exploitable.
We can expect Blender developers to change their methodology. That may have an impact on the long run.
But for 2.9, 3.0 series, that is too late.
