On 05/28/2016 09:55 PM, Rainer Weikusat wrote:
> This means the leading sentence would be more appropriately worded as
> "In my opinion, it's actually quite strange that UNIX(*) enables users
> of the system to run background jobs".

Well put!


On 05/29/2016 02:48 AM, Steve Litt wrote:
> I see poettering's point, but it's just not a problem for me. I don't
> use a desktop environment that, without my permission or knowledge,
> starts tens of processes in my behalf. For the most part, when a
> process is started in my behalf, I personally started it from a command
> prompt, Dmenu or UMENU. So I can choose whether or not to close it
> before logging out.

I think this is Lennart's biggest failing. He is trying to make a
traditionally server OS into an entertainment system. Face it, most
people these days use their devices for entertainment - not work. So
it's not even fair to say that systemd is useful for workstations.

> Every bit of this was predictable from the moment we learned about
> systemd's architecture. Gratuitous component intercommunication leads to
> ever worsening problems. A system with gratuitous component
> intercommunication is so complex that it's difficult to predict exactly
> what those problems will be, but it's a certainty those problems will
> occur.
>

It's also scope creep, which is diametrically opposed to "one tool for
one job". Some of the comments on that list were interesting: "...who in
turn will finally get annoyed at systemd." It's like they understand the
problems associated with systemd but are committed to it. So now they
have to protect their decision - making their relationship with systemd
more emotional than logical.

I am not afraid of change, yet, experience has shown that consensus is
wonderful. Lennart doesn't seem to understand that. I am so glad that
there are people with, shall we call them, traditional (unix) values
still on this planet.