Except the vast majority of the kernel is in driver modules.
So for an individual machine, the attack surface is not really any bigger than it needs to be.
The OS will only load modules it needs for your hardware, so the “bloat” only exists at the source code and binary size level. You are free to compile an optimized binary for your hardware. The complete kernel binary should fit in a 200MB boot partition.
As for maintenance, that’s a fair point, but the effort is at least somewhat distributed if hardware devs provide the drivers.
More lines = more attack vectors, more maintenance, more bloat.
Time to refactor the whole kernel into a one-liner!
A single line of lambda calculus.
With about forty million parentheses.
That’s cheating. It won’t actually make the code smaller.
source?
Common sense. Just making the same code into a single line won’t change much.
that doesn’t sound right, it’s so much smaller if it’s just one line
The line will be like several quadrillion characters long though.
but if there’s a bug I’ll know exactly which line to look at
Easy fix. Just remove newline characters.
Except the vast majority of the kernel is in driver modules.
So for an individual machine, the attack surface is not really any bigger than it needs to be.
The OS will only load modules it needs for your hardware, so the “bloat” only exists at the source code and binary size level. You are free to compile an optimized binary for your hardware. The complete kernel binary should fit in a 200MB boot partition.
As for maintenance, that’s a fair point, but the effort is at least somewhat distributed if hardware devs provide the drivers.