4 ms·

I know this conversation keeps coming up again and again, but this is firmly nvidia's issue. Nvidia has been antagonistic to the Linux desktop for forever. Bef

I know this conversation keeps coming up again and again, but this is firmly nvidia's issue. Nvidia has been antagonistic to the Linux desktop for forever. Before Wayland, it was not sunshine and rainbows. For a long time, a lot of shit just did not work and there was no way to make it work. Nvidia did not care. It took Linux Trovalds shitting on Nvidia to make them care a little. It took YEARS for Nvidia on the Linux desktop to be decent. And, to this day, a lot of Nvidia features just do not work, even on X. You get a MUCH better experience on Windows. Wayland cannot see the Nvidia source code because nobody can. They're trying their best, but they're not going to regress to old ass versions of stuff to get it to work and they're not gonna throw spaghetti at the wall. Nvidia needs to step up because they decided they're the only ones who can develop the driver. There's nothing anyone can do.

charcircuit 1y ago
>Nvidia has been antagonistic to the Linux desktop for forever. Have they? Or have others refused to work with them, refused to make compromises, and refused to offer stable interfaces for them to target. Don't pretend like none of this "antagonism" wasn't self inflicted.
- const_cast 1y ago
  1. The kernel never provides stable interfaces in kernel space. That's just how it is. 2. Nobody can work with them anyway because their drivers are closed sourced. They decided they should be the only ones developing them, so be it. If they upstreamed them they would get thousands of contributor's worth of help.
  - charcircuit 1y ago
    1. It doesn't have to be that way. Windows provides a stable interface for drivers for example. 2. Directly contributing code is not the only way to work with someone.
    - boudin 1y ago
      You're talking about Linux here, that has nothing to do with wayland, it's always been like this and unlikely to change.
      - charcircuit 1y ago
        The other person was talking about Linux. For a Wayland example look at the tight coupling Wayland compositors have with Mesa. It wasn't until 2021 where Nvidia had to be the ones to contribute their own code for being able to dynamically load new gbm backends that weren't already compiled in.
        boudin 1y ago
        That's a really weird way to look at it. In reality nvidia just chose to not invest in it. This sums it up: https://blog.martin-graesslin.com/blog/2016/09/to-eglstream-or-not/ https://blog.martin-graesslin.com/blog/2016/09/to-eglstream-... Gnome did land eglstream support by the way, xwayland as well. I think kde too at some point (not sure about that). Nvidia had a lot of options on the table: - allow nouveau to reclock gpus or, better, collaborate with them so that at least there's one usable driver. - follow the community consensus and go with GBM straight away. Didn't seem to be a problem in the first place as it's what they are doing now. - open source their driver like other manufacturer and be part of mesa. Community had one option: dedicate a lot of time and effort to create a whole code path for nvidia only and having to debug it against a black box driver
        charcircuit 1y ago
        >In reality nvidia just chose to not invest in it. This sums it up This article is proving my point. People were not willing to work with Nvidia and would prefer to stick to their unstandardized Mesa specific approach with gbm. As a path forward Nvidia tried proposing a cross platform open standard that would solve not only wayland's use case, which to me appears as Nvidia going above in investment. >allow nouveau to reclock gpus Nouveau able to if they load the proper firmware onto the GPU. >at least there's one usable driver It's quite a stretch to call it usable and would not make sense to be the default experience for the average user. >follow the community consensus and go with GBM straight away As I mentioned GBM was tightly coupled with Mesa, so that wasn't directly possible. >open source their driver like other manufacturer and be part of mesa. This is an unreasonable demand, that would never happen. Saying that Wayland won't work on Nvidia hardware because they won't open source the driver and port it to Mesa just shows how stubborn and uncooperative one is as opposed to trying to find solutions to deliver user value. >dedicate a lot of time and effort to create a whole code path for nvidia only and having to debug it against a black box driver There was never anything nvidia only in this situation. The only "only" was gbm being a proprietary API that was only a part of Mesa.

charcircuit 1y ago

>Nvidia has been antagonistic to the Linux desktop for forever. Have they? Or have others refused to work with them, refused to make compromises, and refused to offer stable interfaces for them to target. Don't pretend like none of this "antagonism" wasn't self inflicted.

const_cast 1y ago
1. The kernel never provides stable interfaces in kernel space. That's just how it is. 2. Nobody can work with them anyway because their drivers are closed sourced. They decided they should be the only ones developing them, so be it. If they upstreamed them they would get thousands of contributor's worth of help.
- charcircuit 1y ago
  1. It doesn't have to be that way. Windows provides a stable interface for drivers for example. 2. Directly contributing code is not the only way to work with someone.
  - boudin 1y ago
    You're talking about Linux here, that has nothing to do with wayland, it's always been like this and unlikely to change.
    - charcircuit 1y ago
      The other person was talking about Linux. For a Wayland example look at the tight coupling Wayland compositors have with Mesa. It wasn't until 2021 where Nvidia had to be the ones to contribute their own code for being able to dynamically load new gbm backends that weren't already compiled in.
      - boudin 1y ago
        That's a really weird way to look at it. In reality nvidia just chose to not invest in it. This sums it up: https://blog.martin-graesslin.com/blog/2016/09/to-eglstream-or-not/ https://blog.martin-graesslin.com/blog/2016/09/to-eglstream-... Gnome did land eglstream support by the way, xwayland as well. I think kde too at some point (not sure about that). Nvidia had a lot of options on the table: - allow nouveau to reclock gpus or, better, collaborate with them so that at least there's one usable driver. - follow the community consensus and go with GBM straight away. Didn't seem to be a problem in the first place as it's what they are doing now. - open source their driver like other manufacturer and be part of mesa. Community had one option: dedicate a lot of time and effort to create a whole code path for nvidia only and having to debug it against a black box driver
        charcircuit 1y ago
        >In reality nvidia just chose to not invest in it. This sums it up This article is proving my point. People were not willing to work with Nvidia and would prefer to stick to their unstandardized Mesa specific approach with gbm. As a path forward Nvidia tried proposing a cross platform open standard that would solve not only wayland's use case, which to me appears as Nvidia going above in investment. >allow nouveau to reclock gpus Nouveau able to if they load the proper firmware onto the GPU. >at least there's one usable driver It's quite a stretch to call it usable and would not make sense to be the default experience for the average user. >follow the community consensus and go with GBM straight away As I mentioned GBM was tightly coupled with Mesa, so that wasn't directly possible. >open source their driver like other manufacturer and be part of mesa. This is an unreasonable demand, that would never happen. Saying that Wayland won't work on Nvidia hardware because they won't open source the driver and port it to Mesa just shows how stubborn and uncooperative one is as opposed to trying to find solutions to deliver user value. >dedicate a lot of time and effort to create a whole code path for nvidia only and having to debug it against a black box driver There was never anything nvidia only in this situation. The only "only" was gbm being a proprietary API that was only a part of Mesa.