Comments by "Lawrence D’Oliveiro" (@lawrencedoliveiro9104) on "Debian Debates Systemd Alternatives" video.

  1. 8:57 That so-called “Unix philosophy of do one thing and do it well” is a half-truth. Look at any Unix/Linux system, and you will a mix of large pieces and small pieces. And the small pieces would not work without the help of the large pieces. Consider: * The Linux kernel, currently at over 20 million source lines and counting. Which “one thing” does it do well? It is successful because it does so many things well. * The X server, which was notorious in earlier times for practically being an OS unto itself, with its own drivers reaching deep into the hardware, running with root privileges. Which “one thing” does it do well? (To be fair, it should eventually be replaced with Wayland.) * Good old Bash. I counted over ¾ million lines in its source code. Its man page is over 5000 lines long. I use it every day, and I know lots of other people do, too. What “one thing” does it do well? It’s basically a large monolithic mish-mash of miscellaneous functionality, accreted over the years. But at least it is useful miscellaneous functionality. I could go on. Big pieces make little pieces possible. systemd makes possible the writing of very short config files to manage custom-developed services—unlike the dozens or hundreds of lines of boilerplate commonly needed with sysvinit. So your code can concentrate on “doing one thing, and doing it well”, thanks to systemd!
    4
  2. It’s not something that ordinary users will notice. It makes a difference to sysadmins and developers of more complex subsystems. sysvinit is basically an old collection of hacks from the early days of Unix. Most people agree that it is past its use-by date. Where they disagree is what to replace it with. systemd is simply the most sophisticated of the replacements.
    2