On Tue, Apr 21, 2020 at 3:01 PM Consus <con...@ftml.net> wrote: > > On Tue, Apr 21, 2020 at 07:47:44PM +0100, Peter Humphrey wrote: > > On Tuesday, 21 April 2020 17:58:03 BST Consus wrote: > > > > > ... and even distribution kernel is not an official thing, but a desperate > > > attempt of someone to fix things. > > > > Eh? Desperate? > > Yeah, mgorny likes to do some provocative stuff like forking Portage. >
Much of the progress in Gentoo comes from things like this. Indeed it is practically encouraged if you read GLEP 39. While a lot of teamwork/consensus is required to keep the lights on and generally maintain good QA, most of the work of really advancing Gentoo comes from small groups of one or more devs just independently forking stuff and moving it forward. In a fork you can make more radical changes without worrying about breaking things, and then eventually the improvements either make their way back into the original project, or the fork replaces the original. Those who still care for the original do the community a service by providing incremental improvements without instability. The beauty of FOSS is that the code is all free to anybody to use as they wish. There is no such thing as a "bad fork." It is just more free code that anybody can use as they wish. Obviously people don't always work on the things we want them to work on, but it doesn't cost us anything really when they do so, and we're always free to do the same ourselves. There are some QA/CI tools out there that have substantially improved the quality of the distro, and most of them have started out as one dev just creating a tinderbox or whatever and filing bugs when they see problems. The only real downside to this is if somebody quits we might lose these tools - but there are efforts to host them on infra once we start to treat them as part of the core experience. When they start out they're just one dev's random contributions and they may or may not persist. On the topic of portage mgorny is not the only one to talk about forking it. Not too long ago there was talk about a fork to work on an improved DAG solver and other speed/quality improvements. And of course there was paludis and pkgcore much further back. Those projects in part lead to PMS which has improved the quality of our repos substantially. -- Rich