| 12 Mar 2025 |
 Marie | do I decompile and diff the game now to see what they changed? I'm kinda interested | 17:57:52 |
 eyJhb | Yes. | 18:02:12 |
| eyJhb | :D | 18:02:13 |
| eyJhb | Now I want to play minecraft. | 18:02:33 |
| Marie | https://pastes.dev/lqiNayylk8
really small diff | 18:12:10 |
| Marie | but sadly i don't know shit about graphics programming | 18:12:42 |
| Marie | * https://pastes.dev/d4jkrlandK
really small diff | 18:14:54 |
 olivia | it's hard to tell without context, but
- the first hunk with glBindVertexBuffer looks like they're trying to invalidate some state when rebinding the same vertex buffer (possibly with a different size?). Doesn't seem like this should be necessary, perhaps it's a mesa bug.
- the second bit is moving from deferred initialization of a specific vertex buffer to initialized it up-front. No idea why
- Not sure about the third bit on L80.
- The fourth bit starting on L138 is switching from assuming a hardcoded minimum index buffer size when issuing a drawcall to reading the actual size of the buffer and using that. Possible the previous version was trying to read past the end? | 18:38:47 |
| olivia | https://gitlab.freedesktop.org/mesa/mesa/-/issues/12810 oh, looks like this is the relevant mesa bug | 18:39:31 |
| Marie | oh thanks for explaining :3 | 18:45:32 |
|  vidariondr joined the room. | 21:08:08 |
| 13 Mar 2025 |
|  @0rzklv:matrix.org joined the room. | 07:59:41 |
 K900 | OK, I've merged the cursed Mesa patch | 14:07:36 |
| K900 | That is significantly less cursed now | 14:07:57 |
 srestegosaurio (on tchncs.de) | In reply to @devintech:matrix.org
Sorry, I should have been more clear. Out of the options listed, I'm looking to understanding which is the best fit for this. From what I'm reading it sounds like either:
https://nixos.org/manual/nixpkgs/stable/#sec-fhs-environments. It says that's a last resort and I don't understand why? It appears to give you sandbox for the app so this sounds good. What am I missing here?
OR
nix-ld, which once this is enabled, am I just good unless there's a very specific dependency it needs? Then I just list for those packages to be installed. I guess i'd have to do a lot more extra to add bubblewrap to this? If it's a native game you can always write a derivation to manage it with Nix.
| 14:27:16 |
| srestegosaurio (on tchncs.de) | GoG installers are just selfextracting zips.
| 14:27:35 |
 AssertInequality | Heyy Folks
So, my long-standing issue still persists. Controller is not detected for non-steam games launched through steam.
- Xorg: controller detected only when steam-input is disabled
- Gamescope nested session (Wayland): controller not detected in non-steam games no matter what I do.
My plan is to bootstrap a minimal jovian-nix installation to a secondary SSD without any custom configurations/scripts and see what I need to implement in my system by diffing the two installations
Before I do that though, It'd be helpful if anyone can confirm that this actually works on a jovian install.
Steps to reproduce:
- Install Heroic Launcher
- Install a game with controller support
- Add game to steam
- In steam, enable steam input and test a random controller customization (I'm currently using an Xbox series controller, but should be the same for all natively supported controllers)
Even if this combination works in the dedicated gamescope session but not a nested session, I'll have something to go on. Currently, this is the only annoyance I have with gaming on nix.
As usual, any and all help is always appreciated
| 17:10:18 |
| K900 | Does Steam actually detect the controller? | 17:22:56 |
| AssertInequality | steam games work flawlessly with steam input | 17:23:27 |
| K900 | I wonder if this is some Heroic nested fhsenv nonsense | 17:23:42 |
| AssertInequality | AFAIK, adding gamescope in heroic's extrapkgs alleviates potential problems like that (?) | 17:24:38 |
| AssertInequality | The interesting thing is that launching steam in big picture (xorg) and disabling steam-input results in the controller being detected in non-steam heroic games | 17:26:05 |
| AssertInequality | If I can get steam-input to work in Xorg big picture, I'd be somewhat satisfied
The reason I need that is per-game controller config and gyro support with PS/DS controllers | 17:27:47 |
| AssertInequality | Another point of note is that the (gamescope native session + heroic games added as non-steam games) work flawlessly on steamdeck (Official SteamOS). There, I'm able to do all kinds of crazy mappings including adding gyro support to fps games that do not natively support that. On steamdeck, heroic is installed as a flatpak. | 17:33:46 |
| K900 | Actually hold on | 17:40:23 |
| K900 | It can't be nested fhsenv fuckery | 17:40:27 |
| K900 | Because Steam Input is just uinput | 17:40:34 |
| K900 | It's not doing LD_PRELOAD or anything | 17:40:43 |
| AssertInequality | Most probably why I can't put my finger on what's wrong. Even searching for the issue turn up people complaining and no solution or even hints
Example: https://github.com/ValveSoftware/steam-for-linux/issues/8992
Supposedly closed as complete but no discussion as to what exactly was/is wrong | 17:49:38 |
| K900 | It was fixed with a Steam client update | 17:50:05 |
