|
I'd like to put forward a controversial opinion, if I may: I do not think adding Rust to the Linux kernel is a good idea; and I suspect this project will struggle as soon as the financial backing is removed. First, a bit of background: I am mainly a C++ and C# user space developer, so presumably despised by all camps involved. Secondly, I have no experience of kernel development whatsoever. Thus, take my comments with a large grain of salt. My argument is as follows. Every decision you make on a software project is akin to an intersection on sets of programmers. When you choose C, you take a small slice on all possible programmers; when you choose kernel development, you take a slice on the set of competent C programmers, and so on. The intersection of programmers that are competent in Rust _and_ in C _and_ in kernel development - including the interaction between the two languages, at a very low-level (because kernel-space is special) - must be astonishingly small. And herein lies an important problem. Let us posit that the kernel does gain a significant amount of code in Rust; the interaction between these two languages (read: friction) will become significant. Either the Rust people will break things for the C people, or the C people will break things for the Rust people. Those who like C but not Rust - presumably a large subset of the Linux C programmers, else one would assume they would be developing a kernel in Rust - will become alienated. Similarly, those who like Rust will be forced to spend a lot of time doing non-rusty things just to get Rust to work, and they too will not enjoy the experience. Finally, many of the sweeping code clean-ups the kernel experiences periodically will apply differently to Rust, if they apply at all - meaning it will be a second class citizen, barring some major investment to compensate for this. In the end, I do not think the problems will come from the languages involved per se but due to the complex technical and especially social interactions that this will create. It is, however, an extremely interesting project technically and I suspect that Rust (and maybe even the kernel) will benefit from it, though not necessarily in ways one would expect. As far as research goes, it is pretty cool. |