some dismay has been expressed
Posted Jun 16, 2006 21:23 UTC (Fri) by gdamjan
In reply to: some dismay has been expressed
Parent article: What's not going into 2.6.18
The chicken or egg problem with new drivers in kernel code (drivers don't get enough testing when outside of the tree, kernel devs don't trust code that's not been used) can be solved by easying the process of appling outside patches for early adopters.
For example, for kernels I compile for desktop/laptop use I don't care if I integrate some exeperimental wifi drivers. But the manual proces of searching for patches, applying them, recompilling, not forgetting to compile for each kernel version etc.. is really hard.
What would be great if there's a repository of 'for testing' patches, that would be liberal about who&how can put patches there. And an integrated method in the kernel-build-proccess to pull and apply those patches. I think git should be pretty good at cherry-picking patches? The info about which patches were added should be saved in .config, so I would only need that one file in any kernel-source, run 'make update' - get's the testing but also -stable patches, make oldconfig and make to have the latest stable kernel with the latest testing patches.
This should really be part of the kernel(-build-process).
to post comments)