alt Hacker News> I'm betting the Linux kernel actually tests their config combinations.
What makes you bet that?
As far as I know, the kernel doesn't have any official CI. People test as they feel like it. Some people/companies build some CI initiatives, others don't use them.
https://lwn.net/Articles/972713/
Also the kernel has > 10000 Kconfig options. Somebody told me that 2^64 is a lot, and that 2^10000 is a lot more.
all config combinations aren't valid (because of dependencies), and when there isn't they are mostly independent (e.g. 2 builds with CONFIG_FOO=y/CONFIG_BAR=y and CONFIG_FOO=n/CONFIG_BAR=n cover 100% of the cases for unrelated FOO/BAR), so it's not that bad.
And folks do build stuff, including with the awesome "randconfig" mode which generates a.. random config, so while it's still possible for some specific config to not actually work at runtime it's actually quite unlikely for build itself to fail (for releases at least)