Despite what the #systemd #devs might think, "42% less #Unix philosophy" is an anti-selling-point.
"Replace #sudo with #run0, let systemd do it" - sure. Throw away a well-audited, widely-used codebase which has worked well for decades, and instead turn it into a request to a #PID 1 process that is a huge modular-but-#monolithic codebase full of constant churn which has barely been #compiled, much less #understood.
Dollars to doughnuts there are more root holes lurking in systemd than in sudo.
@shuLhan @cazabon Well, based on my experiences with #journald as a #syslog_ng guy, my expectation is that around 5-10 years of security nightmares are about to come with #run0:
https://www.syslog-ng.com/community/b/blog/posts/systemd-journald-vs-syslog-ng
Yes, a decade later after journald arrived, I have no problem recommending it. But the first 7-8 years were catastrophic both for users and developers.
FWIW, I still detest journald. Yes, there are problems with the way #syslog traditionally handled text files - but the solution to that is to fix that handling, not toss them out and replace it with a binary format.
text logging can absolutely be done safely and with no risk of message loss. `multilog` and friends proved that. And unlike journald, I can explain #multilog/daemontools to a novice in 5 minutes and they can still use all the regular tools on the logs.
C.
shulhan
Peter Czanik