The lack of backward compatibility for the timeout command was one thing, but what I was most concerned of was the size. My slacko 5.6 comes with busybox v1.21.0. At 689K it is not really lightweight compared to single GNU commands, which typically weigh only around 50K. But now this new v1.30.1 is a real monster with its 958K.
I always assumed that running a busybox command must be slower than a slim GNU command and running a fat busybox version must be slower than running a leaner version.
Seems I was wrong. I tested the timeout command on all 3 versions:
For busybox 1.21.0
Code: Select all
time busybox timeout -t 1 yes
Code: Select all
time busybox timeout 1 yes
Code: Select all
time timeout 1 yes
- busybox timeout (BusyBox v1.21.0; size 689K)
real 0m1.006s
user 0m0.037s
sys 0m0.383s
busybox timeout (BusyBox v1.30.1; size 958K)
real 0m1.002s
user 0m0.040s
sys 0m0.490s
timeout (GNU coreutils 8.24; size 47K)
real 0m1.006s
user 0m0.050s
sys 0m0.480s
I also compared other commands. Not all showed dramatic differences, but I could see a tendency that newer busyboxes are faster than older (more efficient code?) and busyboxes are not slower than the coreutils. Surprising, but good to know