Please note: the below theory is all conjecture from briefly studying the code. However I think it's still useful to get a bit into the 'why' of the normals being weird.
What that patch is…
Ah, I see. Cycles gets the uv layers via the python interface. So that was already fixed.
For the python api, yes: (#104783)
For cycles: I was wondering about that. I tested cycles expecting it to need extra changes. But it seemed to work as expected.
But now that you say the…
is is the stacktrace I get when trying to open untitled248.blend
Thread 1 "blender" received signal SIGSEGV, Segmentation fault.
0x00000000013c3987 in WM_window_get_active_scene (win=0x0) at /home/martijn/progs/blender-git/blender/source/blender/windowmana…Is this still relevant? I'm sorry I kinda forgot about it.
I think this should maybe still be useful to add after #105192, maybe after some reworking.
I could maybe add the patch to set the first created named to the default render layer. Because when somehow now a named layer is created first, before an anonymous layer, it will be set as the…
A better solution is maybe to remove anonymous layers from evaluated objects. (and update the 'active/render' layers as needed.
I'm struggling a bit with the right commit message. I have no idea about the customary terminology concerning the depsgraph.