@Sebastian Parborg (zeddb), I felt that is something wrong before manual. Then I read it. And it is still not working as in manual or as expected(intuitive way).

The same problem if I want to drive property by bpy.context.scene.render.engine
It is strange that info-editor recieves almost all events (changing render engine or colors) but drivers can`t…

@Clément Foucault (fclem) I mean crashes appear in similar situations: when render is on and shader is compiling.
Not only with this particular shader.

No, my win is x64

As I found it works now back way: rays shooting from selected to active and hits inner cage.
Ray distance is not working at all.
Ofcourse cage extrusion is not working with backray: it extrudes only outside as it should.

@Antonio Vazquez (antoniov)
Thanks a lot for fixing that!
I noticed that now it works (when my addon said «object_depth_order not exist»)

Ok. It seems, that ray distance do not work for backface and inside of mesh. So cap and border baked on the bottom. And side was baked on the top:

Yup, it acts weird:

Tried. True.

@William Reynish (billreynish) yes, it`s priority can be low.
But, perhaps, it can be bonded with bug, and i am trying to reproduce it.

I agree, that it is not a bug. But loss of efficiency since 2.79.
How and where can I report it?

@zeddbb Crashes was with GPU.
Now I can`t reproduce. It seems problem was with video-drivers and with GPU frequency/voltage jumps (I decreased maximum limit a little with afterburner).
Thanks for help. And what about regression? 2.79 do not render each time I add node as you can see on the video. But 2.8 does this useless work.

@Sebastian Parborg (zeddb) With this library inside folder Blender do not start at all.
Also my system dir includes such library already.
I refreshed drivers. May be it'll help

@Sebastian Parborg (zeddb)
Can I dump crashlog somehow? I tried to find logs and found nothing.
Crashes are often, so may be I can collect info

@Sebastian Parborg (zeddb) it is not about crashing, it is about regression (ppl call it regression bug)

@Brecht Van Lommel (brecht)
Yeah, bugs everyday :)
I am trying to not overhaul with unnecessary info.

Another bit of info. If i add driver, it starts to move but interminent

@Philipp Oeser (lichtwerk) Yes, it helped, THANKS!
So it looks like some animation was created and it pinches value?

Oh, here it is!

Sadly can`t reproduce, but such stucking looks like the driven behavior.
And yes, it is look totally the same as I reported before: https://developer.blender.org/T61729

Oh, i remember another one thing, that can be bonded.

Pole target acts like a side force. But there is no friction forces inside hinges.
As I said, I know about pole-tartget, but it fixes symptoms, not the reason.

@Sebastian Parborg (zeddb)
Yes, you are right. New metas should born with settings, inherited by family. It will prevent such problem in 2.8 and 2.79 too

Oh, I see. s far as we can`t make separate options for strokes or layers inside one GP-object, I feel too, that this options should bonded.

Additional info

Worked with pencil. It seems selection exist but invisible

Also default cursor`s rotation mode is Euler now. It causes problems with gimbal lock very fast when dragging it here and there with snapping.
Please make Quaternion default.

I looked into info-console and saw this.
Still view-oriented, ignoring 3d-cursor settings and snap-settings partially

It seems, there is wrong conversion from normal axes to cursor`s and it inherits rotation from first cube
p.s sorry for bad enunciation.

Here it is

Anything else i can do?

Can i help anything else?
I see status, that info required. What kind of info do you need?

Thank you!

You guys are awesome!

Did you found the difference between startup file and nomov.blend? May be that difference can lead to the root of bug…

Wireframe option is inactive. Is it just not implemented? It is Confusing me…

Yes, you are right. I temporary disabled it, until bug will fixed.

First file have no such bug at all. It is initial file for comparation with bugged.

I mined more
2.80 (sub 45), 893fa598319e, 2019-02-24 22:02

May be something indirect: AV-sync affects to playing frames, frames switching during baking, so pace of frame switching affects baking somehow…
As I said, sometimes it looks like skipped frames and interrupted simulation.

Yup, it works good, if baked without AV-sync.
But AV-sync playback works good with this proper bake.