On Wed, Jun 16, 2021 at 08:11:23PM +0200, Miguel Ojeda wrote: > So, in a way, I am trying to go a bit further by bringing the kernel > development workflow into GitHub (and similar services), rather than > keeping the ML as the main place for code review etc. I know it can be > controversial, but hey, I want to try. :) That's pretty cool, but I'm opposed to this on theological grounds. :) I think there's intrinsic value to development happening outside of any proprietary environments, especially if this becomes cornerstone to how a subsystem is maintained. I'm not saying Github is going to go away any time, but on the other hand we just watched Freenode completely implode within the span of 3 weeks, so I'm extra wary of introducing any single points of value into the development workflow. If my pr bot goes away, it'll at most reduce the convenience of the submission process, but not hobble it in any way, so I intend to just treat Github as a developer frontend tool. > By the way, the gccrs project (also in GitHub, but also following > GCC's workflow which involves MLs) also wanted something like ML-to-PR > direction, and I suggested to them taking a look at `b4` too. Sure, but just be wary that b4 internal structure is still under heavy development, so I wouldn't recommend treating it as a stable library until it reaches 1.0. Best, -K
