Comment by lonjil
10 months ago
> The maintainers of core subsystems are the people he trusts, at least trusts as much as you can in this space. He'll take their opinions before anyone else, since they know best about the subsystems they maintain
But there were no technical arguments against the Rust wrapper. And in any case, the Rust wrapper isn't in that subsystem, it just uses that subsystem. Hellwig's argument was nothing more than "there shouldn't be a second language in the kernel". He had nothing specific about the DMA wrapper. And Linus has already approved Rust in the Linux kernel, so what's the problem? Why can't Linus put his foot down on an issue that he has already decided on?
> Hellwig's argument was nothing more than "there shouldn't be a second language in the kernel".
Which is a valid viewpoint. Let's not pretend that's not a technical argument.
Having different technical views from yours isn't a crime, legally or morally.
> And Linus has already approved Rust in the Linux kernel, so what's the problem?
As an experiment, as clearly stated in the kernel docs. It's still up to the whole community to figure out how exactly to proceed with it.
https://docs.kernel.org/rust/index.html
> > Hellwig's argument was nothing more than "there shouldn't be a second language in the kernel".
> Which is a valid viewpoint. Let's not pretend that's not a technical argument.
So if Linus shouldn't overrule his deputies, and one deputy can completely block something in a subsystem that isn't theirs and has explicitly stated to do "everything in their power" to stop it, what exactly can the community "figure out" to get it to proceed? His decision not to get involved literally makes it impossible for the situation to change, so the confusion is why it's being phrased as if it's anything other than that.
If Linus were to come out and say "I changed my mind, I no longer think it's worth pursing trying to integrate Rust in the kernel due given the issues at hand" or even "regardless of my personal views, I don't have any desire to change the current system we have for how code gets merged into subsystems or who has the ability to block it, so I'm not going to overrule this", it would make a lot more sense to me, but by making it sound like there's anything left to figure out when someone with veto power has a stated intent to stop things from moving forward is just going to let the issue fester and produce more frustration on both sides. It almost seems inevitable that any future discussions will spiral out of control; there's no other way for it to conclude than either Linus overruling his deputy or the experiment just ending as a failure at this point, so he might as well just make that decision now rather than later.
It is not a valid reason to reject a patch when the decision to include the second language in the kernel has already been made.
A decision to include Rust in the kernel isn't a decision to include any patches from any specific person with no questions asked.
It's valid for maintainers to reject a patch even if you disagree with the reason. Repeatedly causing social media storms to "shame" them for doing so, Marcan's own word BTW, isn't.
Rust is young and efforts to use it in kernel programming is even more so. It's completely understandable for wanting to take things slow and not be too quick to put it in places where there would be no going back. Everyone should at least be able to recognize that this concern exists, regardless of their opinions on Rust.
4 replies →