[foxfluff]

Well, ps on linux shows kernel threads whereas on OpenBSD it doesn't.

The list is pretty short on my alpine install if you exclude kernel threads:

    # ps ax | grep -v ']$'
    PID   USER     TIME  COMMAND
        1 root      0:01 /sbin/init
     1985 root      0:00 /sbin/udevd
     2361 root      0:00 /sbin/syslogd -t
     2387 root      0:00 supervise-daemon wpa_supplicant --start /sbin/wpa_supplicant -- -i wlan0 -c/etc/wpa_supplicant/wpa_supplicant.conf
     2388 root      0:00 /sbin/wpa_supplicant -i wlan0 -c/etc/wpa_supplicant/wpa_supplicant.conf
     2473 root      0:00 /sbin/udhcpc -b -R -p /var/run/udhcpc.wlan0.pid -i wlan0 -x hostname:pi4
     2567 chrony    0:00 /usr/sbin/chronyd -f /etc/chrony/chrony.conf
     2592 root      0:00 /usr/sbin/crond -c /etc/crontabs
     2621 root      0:00 sshd: /usr/sbin/sshd [listener] 0 of 10-100 startups
     2652 root      0:00 /sbin/getty 38400 tty1
     2653 root      0:00 /sbin/getty 38400 tty2
     2654 root      0:00 /sbin/getty 38400 tty3
     2655 root      0:00 /sbin/getty 38400 tty4
     2656 root      0:00 /sbin/getty 38400 tty5
     2657 root      0:00 /sbin/getty 38400 tty6
     3099 root      0:00 sshd: root@pts/0
     3102 root      0:00 -ash
     3109 root      0:00 ps ax
     3110 root      0:00 grep -v ]$

I don't think kernel threads are a good measure of bloat. And a base install of Alpine is smaller than OpenBSD.

But that's a fair point about presentation. OpenBSD seems to be more 'quiet' by default, and tends to only show relevant information. dmesg on OpenBSD seems much cleaner compared to the mess that Linux outputs.

[enriquto]

Thanks for the clarification! It is indeed true that alpine is very clean an not bloated at all; just that openbsd seems more intentionally polished.

As a matter of fact, I don't think this has anything to do with linux itself, just with the large distributions. If you use alpine, slackware or void you get a similar streamlined experience.

[flatiron]

what are you trying to measure?

openbsd will definitely feel from 10000 feet more "bloated" because its simply not as performant as linux. that's not a bad thing, it purposely does things the "right" way for security purposes and doesn't take any shortcuts.

alpine is a lot smaller than openbsd but it really was created for an entirely different purpose. i always take alpine as "a muscl distro that makes a good docker container, oh and it runs on bare metal too, i guess". i've never seen alpine on metal in prod and i've been around the block a whole bunch. ive seen it in a metric ton of docker containers though.

a chatty dmesg also isn't really bloat as well. although dmesg is a bit of a mess (and only recently default restricted to privileged accounts at least on arch).

[foxfluff]

I don't consider poor performance to be bloat (though poor performance can be a symptom of excess bloat). Bloat is more about obesity, and it shows in disk usage (upgrades take forever? yeah two gigabytes of data across 2000 packages; kernel doesn't fit in flash? keep unticking those kbuild options..), excessive memory usage, ridiculously long man pages, etcetra. Bloat doesn't necessarily impact performance in normal use, e.g. those 2000 packages I got on Fedora mostly sit on my disk untouched. But it's still there and it shows when it's time to update (and sometimes while doing other stuff).

OpenBSD might not scale well to a large number of cores, and the program running on a 8-bit microcontroller (with 512 bytes of ram) on my breadboard isn't fast, but neither are particularly obese.

I don't think the purpose between Alpine and OpenBSD are that far apart. Alpine aims to be a simple, small, and secure general purpose OS. OpenBSD is very similar, even if Theo has been pushing the "research OS" angle. There's obviously a big difference in how much software include in the base install.

[formerly_proven]

> I don't think kernel threads are a good measure of bloat.

It kinda is. Totally unrelated, but there is nothing quite like seeing hundreds of [nfsd] tasks hard-blocking in D on each server in a cluster. Time to reboot... well... everything.