Search Issue Tracker
Won't Fix
Votes
1
Found in [Package]
2.1.0-preview.1
Issue ID
1309684
Regression
No
Scene visibility toggles visible when entering Play Mode for TMP SubMeshUI that is in a Prefab and is invisible
How to reproduce:
1. Open the user's attached project("BugProject.zip")
2. Load "SampleScene"
3. Make sure "Canvas" is toggled invisible in the Hierarchy
4. Enter Play mode
5. Notice "TMP SubMeshUI" has been toggled visible again
Expected results: The TMP SubMeshUI stays invisible after entering Play Mode
Actual results: The TMP SubMeshUI becomes visible after entering Play Mode
Reproducible with: TMP 2.1.0-preview.1, 2.1.3 (2019.4.20f1), 3.0.0-preview.1, 3.0.3 (2020.2.3f1, 2021.1.0b5, 2021.2.0a4)
Not reproducible with: TMP 2.0.0 - 2.0.1 (2019.4.20f1)
Could not test with: TMP 1.5.3 (2018.4.32f1) (No scene visibility option yet)
Note: Scene visibility stays toggled visible again for the TMP SubMeshUI when exiting Play Mode
Add comment
All about bugs
View bugs we have successfully reproduced, and vote for the bugs you want to see fixed most urgently.
Latest issues
- Spring Joint shows only one anchor gizmo in Scene view when "Auto Configure Connected Anchor" is enabled
- Crash on _platform_memmove after entering large value in Graphics settings Preloaded Shaders field
- Disproportionally large impact on CPU frame time when writing to a rendering entity's LocalToWorld
- "Constant Force" Component numeric fields drift out of view while entering a really big value in the Inspector
- Scene view camera speed pop-up appears empty or cut off when Scene view is very narrow
Resolution Note:
This bug comes from the fact that temporary non-interact-able objects are both visible and select-able in the Hierarchy. When entering play mode, the temporary child objects are destroyed and recreated, losing any user-set values. This isn't specific to visibility flags. Static flags, tags, icons, etc will be lost as well. Adding special handling the scene visibility rendering code to handle this case isn't clear cut, as there are valid arguments that doing so would be a regression to expected behavior (eg, a user can select and modify this gameobject, therefore the systems should work as designed).
Given the low user pain and ambiguity of a "correct" fix, will resolve this as WontFix.