Is it possible to improve the development process of the Linux kernel?
Currently to submit patches to the kernel, you have to go through the mailing list, an outdated way. Including review by the maintainer also happens on the mailing list. However, mailing lists have a number of significant drawbacks, such as patches being ignored and not being easy to view, and discussions on mailing lists are not structured enough, resulting in important information being drowned out in the mass of emails. Can current kernel development be migrated to a platform like Gerrit?
I think the fact it's so "archaic" or "outdated" is actually a good thing, by increasing friction you make sure only well thought patches make it to the maintainers. Take, for example, how hard it would be to keep patches high quality if they were using GitHub as an upstream instead of a mirror. Also, it's really cool because it's truly decentralized, email is one of the last protocols that are decentralized and well-supported. Overall, I like it and I think it shouldn't change.
It's certainly possible -fork the repo, start coding, ask others to contribute, start reviewing, use any review tools and processes that you consider aren't outdated.
as a device creator the main issue is that SoC vendors maintain their own trees and their work doesn’t seem to easily make it into the mainline tree. this causes a lot of work on my end especially because there is no way to manage patches and bring over all necessary changes related to one subsystem from their tree to the mainline or other way around. this means that small companies can forget about using linux on devices unless they get lucky because other people already did the work.
so the main issue i see is that it is too hard today to group changes related to a subsystem and too hard to manage patches.
on the other hand the maintainers are open to bizarre and very involved changes like allowing rust in the tree. the effort this is taking could have been spent solving the above problems.
While the technical divide between your mom and GKH is probably vast, their reasoning for sticking with old-fashioned methods is the same: "Works for me."
I have vague recollections of someone floating the idea of having a linux kernel mirror on github, where pull requests would interop with the mailing list.
Why ask on HN? You know who you need to talk to.