One the one hand, we have Moore's law. On the other hand, kernel compilation time. Since compilation time is monotonically increasing, do we observe exponential compilation complexity in the kernel?
I remember starting a 1.2 kernel compile on my 486 with 4 MB of RAM, going to bed, then going to school, and finding that it had finished when I came back home.
Back in pre-module days, Slackware shipped with "big" kernel with lots of drivers compiled in. The advantage was that this way the kernel could boot on a wide range of hardware. But it was very bloated (for the time) and the users were expected to recompile the kernel with unnecessary drivers removed. I remember compiling it on Pentium 60 with 16MB of RAM. Took 1-2 hours or so.
10 comments
[ 2.4 ms ] story [ 32.0 ms ] thread