I understand your answer, and I realy
don't have a problem with the actual development method.
I just wondered, why not use lets say gcc 4.8.1 ? For example my kernel is Linux version 3.9.7-1-ARCH (tobias@T-POWA-LX) (gcc version 4.8.1 (GCC)
) #1 SMP PREEMPT Thu Jun 20 22:45:32 CEST 2013, x86_64
GNU/Linux. Is an archlinux unchanged kernel. So the gcc 4.8 is tested to work OK by a wide community.
The idea is the same like in "Aboriginal Linux" to have a self hosting linux (using just kernel, libc and busybox etc).
So, first to have the gcc [cross?]compiler, then we compile the kernel, just once.
Lets take the worst scenario, that the kernel compilation will fail, very unlikely because any vanila kernel JUST compile, otherwise the kernel developers can go make sepuke themself
On my Core2Duo with 2 Mb RAM, a 5 years old notebook, the full tinycore kernel will compile in less than 20 minutes.
So do we discuss about 20 minutes eventual waste time in a full core life circle of maybe one year?
Anyway I can live with this, I just like tinycore concepts, but many times I wish to rebuild things.
Keep the good work, and sorry for my noise.