Operating system: Arch Linux
Graphics card: GTX Titan
Broken: 2.80.75 (f6cb5f54494e)
**Short description of error**
Behaviour with respect to keeping Cycles/Evee shader nodes around seems to be inconsistent with the documentation, as the shader editor is either not available or shows an incomplete list of shader nodes.
**Exact steps for others to reproduce the error**
On https://docs.blender.org/api/current/bpy.types.RenderEngine.html the property `bl_use_shading_nodes_custom` is described as
> Don't expose Cycles and Eevee shading nodes in the node editor user interface, so own nodes can be used instead
Using the attached RenderEngine test class, which is based on the code in the above doc page:
1. Start `blender -P blender_render_engine.test.py`
2. Switch to `Custom` renderer
I see the following depending on `bl_use_shading_nodes_custom`:
- When `bl_use_shading_nodes_custom` is `False` (i.e. keep Cycles/Eevee nodes
around) the Shader editor's add node menu (`Shift+A`) under `Shader` only
has a `Principled Volume` entry, other Cycles nodes are gone. But when creating a new material in
the editor the initial graph is `Principled BSDF -> Material Output`, i.e.
using the principled BSDF node that cannot be created from the menu.
- When `bl_use_shading_nodes_custom` is `True` the Shader editor is
no longer available. Is this new behaviour, where a custom RE
needs to provide its own editor page, instead of adding to the existing shader editor?
There's also a corner case when having the Shader editor visible (for example when Cycles is active) and then switching to a render engine that has `bl_use_shading_nodes_custom=True`: the icon for the editor will become blank, but the Shader editor remains visible. In the console I warnings in this case:
`WARN (bpy.rna): ../source/blender/python/intern/bpy_rna.c:1451 pyrna_enum_to_py: current value '3' matches no enum in 'SpaceNodeEditor', '(null)', 'tree_type'`