I just watched the FOSDEM presentation "systemd in Debian". A couple of issues that occur to me -
1) The continued use of "/lib/systemd/" instead of "/usr/lib/systemd" and "/usr/" generally - The default installation of the native systemd source is so focused upon "/usr/" that this continued Debian modification to avoid using "/usr/" would seem to be a waste of time. What is the point? In general, creating a Debian package which deviates far from the native source package without some very practical motivation seems a waste. 2) While I appreciate the desire to provide a smooth introduction to systemd for those people still using sysv-init, at the same time, my own interest in systemd is relief from the ridiculous "cruft" of Debian sysv init scripts, or from maintaining my own custom init scripts. In this case, I am more interested in deleting the sysv init scripts and having provided native Debian systemd unit configuration files. I would consider this a much more important task, providing Debian unit configuration files, than having the Debian systemd maintainers focusing their attention upon repairing the sysv init system. And after all, what good is a systemd-to-initscript converter if there is no systemd unit configuration file to convert? Of course, I'm also hoping that a native set of Debian unit files would allow me to move from systemd-196 to systemd-197 with less hassle. ;) But importantly, the boot-time "speed-up" claimed for systemd is not possible when running a lot of "do-nothing" sysv init scripts, including scripts for services which are simply not enabled, but especially services which check and re-check configuration files and administrative directories using the init script instead of the package installation scripts, and re-run every single time the system boots. 3) No mention was made of Debian systemd interaction with "dracut", the systemd-based alternative to the "initramfs-tools" package. 4) systemd-44 is dated 2012 Mar 16, so is not "six months" obsolete, but more like six weeks shy of one year obsolete. What portion of all Debian users are running "unstable" as compared to "stable"? I'm only just realizing that you plan to release systemd-44 as part of the next Debian "stable". And then, it cannot be changed for two years? Considering the rate of systemd development, and the impact upon Ubuntu, that seems like a mess, imposing some kind of bastard "systemd-sysv" hybrid on people. That is not even a fair representation of the "systemd" being designed by the systemd developers. Perhaps better to have _no_ systemd in the next Debian "stable". Really. James -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
