4 ms·

This one bit looks exactly right, though: > As for how to move forward, (...) Either Linus takes the pull, and whatever Christoph says is irrelevant, or he doe

This one bit looks exactly right, though: > As for how to move forward, (...) Either Linus takes the pull, and whatever Christoph says is irrelevant, or he doesn't, and R4L dies. Everything else is a waste of everyone's time and energy. It does look like maintainers should have a "disagree and commit" mentality at some point, whatever decision they end up making. I thought Rust in Linux was evaluated, discussed and agreed upon years ago. The fact that there are people still trying to sabotage it shows that they don't follow the "disagree and commit" principle. They are more like "disagree and make the others lives a living hell until they bend to my will".

tptacek 2y ago
That's a much stronger argument when you aren't at the same time yelling about the email patch system. I want R4L to succeed!
geodel 2y ago
> disagree and commit It is not some million dollar RSUs getting vested by year end either way. A lot of them working for the love of craft and prestige. If they can just rollover on a technical disagreement then corporate office job is more suitable than open source OS kernel.
- gf000 2y ago
  The Linux kernel is most definitely not developed for free by some college student in their bedroom for the most part. It's Intel and Red Hat and similar. Edit: related email by Hector: https://lore.kernel.org/rust-for-linux/c5a49bcb-45cf-4295-80e0-c4b07083df00@marcan.st/ https://lore.kernel.org/rust-for-linux/c5a49bcb-45cf-4295-80...
justin66 2y ago
> I thought Rust in Linux was evaluated, discussed and agreed upon years ago. I've seen Linus talk about it in one of his public chats with Dirk Hohndel as an interesting experiment that might succeed or fail, or that's the impression I got. I'm not sure everyone else got that memo.
arp242 2y ago
"R4L dies" is hyperbolic nonsense. There are many ways/places to integrate Rust in the kernel, even for wrappers. It's not like "I think this code should be someone else" automatically means the death of Rust. Marcan goes full nuclear every other time someone disagrees with him. He's very much the "if it's not my preferred way then we might as well not do it at all"-type of engineer (many of us have worked with those people).
- steveklabnik 2y ago
  Rust is only allowed for drivers, because that sidesteps the platform support questions. DMA is needed for an overwhelming number of useful drivers. If you can’t use DMA from Rust, then you can’t really properly evaluate the usefulness of Rust, hence the effort is basically dead.
  - arp242 2y ago
    People can just include their own dma.rs in driver/.../mydriver/ Whether that's a good or bad approach is besides the point. It's what was suggested as an alternative, and clearly it's something that would work. "You can't use DMA from Rust" is just not true.
    - codys 2y ago
      If it was a good approach, the C folks would have copied around C files instead of having common core code. They did not do that though, because it'd be counter productive due to increasing maintenance burden. Everyone knows it would increase maintenance burden, decrease reliability, and increase the amount of apparent churn of rust code in the linux kernel.
      - Brian_K_White 2y ago
        Everyone knows you could start the crude way today and refactor the duplication later after it proves itself, if you actually wanted to proceed rather than throw fits.

tptacek 2y ago

That's a much stronger argument when you aren't at the same time yelling about the email patch system. I want R4L to succeed!

geodel 2y ago

> disagree and commit It is not some million dollar RSUs getting vested by year end either way. A lot of them working for the love of craft and prestige. If they can just rollover on a technical disagreement then corporate office job is more suitable than open source OS kernel.

gf000 2y ago
The Linux kernel is most definitely not developed for free by some college student in their bedroom for the most part. It's Intel and Red Hat and similar. Edit: related email by Hector: https://lore.kernel.org/rust-for-linux/c5a49bcb-45cf-4295-80e0-c4b07083df00@marcan.st/ https://lore.kernel.org/rust-for-linux/c5a49bcb-45cf-4295-80...

justin66 2y ago

> I thought Rust in Linux was evaluated, discussed and agreed upon years ago. I've seen Linus talk about it in one of his public chats with Dirk Hohndel as an interesting experiment that might succeed or fail, or that's the impression I got. I'm not sure everyone else got that memo.

arp242 2y ago

"R4L dies" is hyperbolic nonsense. There are many ways/places to integrate Rust in the kernel, even for wrappers. It's not like "I think this code should be someone else" automatically means the death of Rust. Marcan goes full nuclear every other time someone disagrees with him. He's very much the "if it's not my preferred way then we might as well not do it at all"-type of engineer (many of us have worked with those people).

steveklabnik 2y ago
Rust is only allowed for drivers, because that sidesteps the platform support questions. DMA is needed for an overwhelming number of useful drivers. If you can’t use DMA from Rust, then you can’t really properly evaluate the usefulness of Rust, hence the effort is basically dead.
- arp242 2y ago
  People can just include their own dma.rs in driver/.../mydriver/ Whether that's a good or bad approach is besides the point. It's what was suggested as an alternative, and clearly it's something that would work. "You can't use DMA from Rust" is just not true.
  - codys 2y ago
    If it was a good approach, the C folks would have copied around C files instead of having common core code. They did not do that though, because it'd be counter productive due to increasing maintenance burden. Everyone knows it would increase maintenance burden, decrease reliability, and increase the amount of apparent churn of rust code in the linux kernel.
    - Brian_K_White 2y ago
      Everyone knows you could start the crude way today and refactor the duplication later after it proves itself, if you actually wanted to proceed rather than throw fits.