3 ms·

I think the anti-systemd crowd hurts its arguments (which I tend to agree with at least in spirit) when they throw out stuff like accusing the authors of system

I think the anti-systemd crowd hurts its arguments (which I tend to agree with at least in spirit) when they throw out stuff like accusing the authors of systemd of being "Ignorant of fundamental operating system concepts" (https://without-systemd.org/wiki/index_php/Arguments_against_systemd/#Ignorance_of_fundamental_operating_system_concepts https://without-systemd.org/wiki/index_php/Arguments_against...). You can dislike the author and the project and the design, but that's just a foolish thing to say that makes it difficult to take any of the other (potentially valid) arguments seriously.

gnulinux 4y ago
I checked both of those arguments and wasn't convinced. First link demonstrates an email by Poettering claiming checksums won't fix anything useful. Clearly he's talking about a specific usecase of checksums and not talking about the whole idea of using checksums to recover data. He was asked about a specific feature in systemd, there is no reason to believe he'd want to bring up anything bigger. So implying he doesn't know RAID... is simply absurd and childish. He could have referred that person to RAID; but that'd be completely off-topic. Second link is about a feauture request where someone wants a complete machine login functionality from machinectl unlike su. Then, Poettering explains why this situation exists (because su doesn't inherit cgroups) and then they go ahead and implement the desired feauture. Original user says "it works perfectly". Now, the criticism seems like Poettering calling "su" broken since cgroups were introduced after su. This doesn't make any sense. Of course this is a problem of su. If there are two tools: ssh and su, and they perform a certain task differently and one of the behaviors causing problem, clearly it's relevant to mention su's method is flawed and a more complete implementation can be written (and then they went ahead an wrote it). Bizarre. Another day of reading tons of anti-systemd material for me, and another day of realizing systemd is a perfectly ok piece of software yet there is an odd internet obsession against it.

gnulinux 4y ago

I checked both of those arguments and wasn't convinced. First link demonstrates an email by Poettering claiming checksums won't fix anything useful. Clearly he's talking about a specific usecase of checksums and not talking about the whole idea of using checksums to recover data. He was asked about a specific feature in systemd, there is no reason to believe he'd want to bring up anything bigger. So implying he doesn't know RAID... is simply absurd and childish. He could have referred that person to RAID; but that'd be completely off-topic. Second link is about a feauture request where someone wants a complete machine login functionality from machinectl unlike su. Then, Poettering explains why this situation exists (because su doesn't inherit cgroups) and then they go ahead and implement the desired feauture. Original user says "it works perfectly". Now, the criticism seems like Poettering calling "su" broken since cgroups were introduced after su. This doesn't make any sense. Of course this is a problem of su. If there are two tools: ssh and su, and they perform a certain task differently and one of the behaviors causing problem, clearly it's relevant to mention su's method is flawed and a more complete implementation can be written (and then they went ahead an wrote it). Bizarre. Another day of reading tons of anti-systemd material for me, and another day of realizing systemd is a perfectly ok piece of software yet there is an odd internet obsession against it.