sonic2kk
commented
4 months ago Nice catch, thanks!
The issue and solution you described makes sense after looking at the code too, I didn't realise how __GLX_VENDOR_LIBRARY_NAME was being set at first.
The solution should be as straightforward as moving the check for Zink to be after Primerun (as you described) in the following place (looks like GitHub doesn't embed linked code snippets for files with many lines): https://github.com/sonic2kk/steamtinkerlaunch/blob/b106b3304dd3c4fd6ad8d4364cb79003f2ca5cc3/steamtinkerlaunch#L12422-L12426
I would normally open this up to ask if you would like to make a PR for this, since you took the time to understand and cleanly report the problem, but I actually want to do a slight refactor here to break some of the code into separate "checker" functions, so I'll take a look at fixing this. Otherwise I would've given an opportunity to contribute the fix and get all the credit, but you'll just have to settle for most of it I think ;-)
I'll push a branch and open a PR. I'll let you know when it's ready for testing and you can check if the problem is fixed on your end with the branch. I think we tested some branches before but let me know if you need any help checking out the branch.
Rabcor
To ensure compatibility between prime-run and zink, the zink variables must be loaded after the prime-run ones (this is because both of them set
__GLX_VENDOR_LIBRARY_NAMEand for Zink to activate this must be set to mesa.Currently it seems the prime-run variables are loaded after the zink ones thus leading to no zink if prime-run is also enabled.
To illustrate:
zinkrun prime-run mangohud glxgearsprime-run zinkrun mangohud glxgears