GLTF Mesh Import Error: Material of model is not a subset of reference model

[AtlasLinden]

Environment

Second Life Project Gltf-mesh-import 7.1.14.15119648860

Description

This error appears when first attempting to upload a GLB file.
Error: Material of model is not a subset of reference model
It can be bypassed by selecting the "Clear settings & reset form" button
https://gyazo.com/f67fcc1eec6aeda04c1ef316113ff64f

Reproduction steps

Upload a GLB file

[AtlasLinden]

Here is a model where selecting the "Clear settings & reset form" button doesn't appear to bypass the error
https://sketchfab.com/3d-models/kpa-fire-hydrant-ae381eeb6eda49d5b3de33653cae4aa3

[aiaustin]

Here is a very simple two part model (Vane-Covers-X-2.glb) where hitting the "Clear settings & reset form" button does not clear the error, even after multiple button presses. The two parts on their own both upload fine. Vane-Cover-1.glb and Vane-Cover-2.glb. The two part model looks fine in the Khronos glTF Sample Viewer. See Vane-Covers-X-2.jpg.

Vane-Covers-X-2.zip

Environment: Blender 4.4.3 with built in glTF exporter, Second Life Project Gltf-mesh-import 7.1.14.15122149339.

[marchcat]

The fix from PR #4121 addresses the main error, which can be bypassed using the "Clear Settings & Reset Form" button.

However, the error related to the mentioned models is different and will require further investigation.

[marchcat]

Vane-Covers-X-2.glb looks correct after #4109,

but there are still problems with https://sketchfab.com/3d-models/kpa-fire-hydrant-ae381eeb6eda49d5b3de33653cae4aa3

[aiaustin]

Using the Test viewer gltf-4019 Second Life Test 7.1.14.15231514054 Vane-Covers-X-2.glb looks correct in preview now, but still, for me, continues to show the "Clear Settings & Reset Form" button over and over without end, so cannot be uploaded.

I see a regression on glb rigged avatar uploads which are scrunched up in preview and if uploaded and worn, but work in latest gltf-mesh-import (Second Life Project Gltf-mesh-import 7.1.14.15205102881 on Windows), but possibly that is because all the changes that were merged into gltf-mesh-import are not in gltf-4019?

[marchcat]

@aiaustin, thank you for checking!

The rigged model regressions are likely caused by recent changes introduced in #4109. They will be addressed separately under #4080 or a related new issue.