Debian May Be Leaning Towards Systemd Over Upstart

[Delgarde]

Systemd as Debian's new init-system of choice has probably just been dealt its fatal blow as upstart is preliminarily working with kFreeBSD. Given the immature refusal of Poettering to even accept any patches allowing systemd to run on other platforms, which is vulgarly defiant of Debian's stated aims, one assumes they will be leaning towards upstart once more.

Speaking as someone who's been following those discussions from the start - not likely. It's new that the BSD port of Upstart is working, but the fact that such a port existed has been known since the beginning, and it's had very little impact on discussions.

Basically, all the committee are accepting of the idea that while the continued existence of the BSD and Hurd variants is important, it's of secondary importance to doing the right thing on Linux, even if that means having different default init systems on Linux and other. It's increasingly looking like their decision will be to go with Systemd on Linux, and Upstart on the other two...

[oleid]

systemd is killing upstart!

That is actually the plan, I guess, as Lennard&Co. used to work on upstart, however, they eventually realised that upstart contains unfixable fundamental design flaws.

[bkor]

You can rip out the logind requirement, the developers specifically not (see the other user's post like 1 or 2 above me). MORE specifically though... Gnome doesn't actually require -LOGIND- what it requires are the dbus interfaces that currently only logind provides.

With the note that forking is not entirely free. On Ubuntu a logind fork is used to provide an alternative. Such is also mentioned in the ctte discussions. However, the dbus API will receive new items as development goes on. Things we need/want for user sessions as well as tty switching (running Wayland/Xorg not as root).

During development you need something to be aligned with. Systemd maintainers you can work with. You know the roadmap, plans, objections, etc. So you can explain your needs and they'll work on it. For Upstart option, the answer basically is "we maintain the fork". So what'll happen in practice is that GNOME will continue to work with systemd developers. Then we have no other option than to assume that Upstart option duplicates whatever has been done within the systemd project.

Above sounds like a totally insane way to work IMO. Way easier for Debian to use systemd. Choosing Upstart means consciously choosing to be behind in the latest developments. And this is not due to forcing, it is due to the lack of cooperation and development in Upstart. Further, it is not really needed, loads of people (distributions, systemd developers, developers from other projects, low level stack developers) are already working together under the systemd name.

[bkor]

The subshell is just an example to show that even something as dumb as a bash shell can capture the standard error of a daemon and save it to a file. So saying that this is a "real problem that can only be solved by putting code in PID 1" is simply not true. Which does not mean that putting code in PID 1 is wrong. What is wrong is stating it as inevitable.

That's not the same functionality. If you redirect the output to a file, it will forever go to that file. So good luck with logrotate and so on. Obviously you could add yet another program inbetween to solve that. But at one point it is better to just see that the way to solve it is by handling this in the init system and all other solutions might work, but are just hacks/unreliable.

[gens]

That's not the same functionality. If you redirect the output to a file, it will forever go to that file. So good luck with logrotate and so on. Obviously you could add yet another program inbetween to solve that. But at one point it is better to just see that the way to solve it is by handling this in the init system and all other solutions might work, but are just hacks/unreliable.

using standard POSIX

http://stackoverflow.com/questions/5789526/log-rotating-with-a-bash-script

dosen't seem unreliable to me
you see something that could break ?

i think maybe it could be done without cat

maybe "it is better to just see"
thinking hurts

[gens]

sry if i came on too hard in last post

just i don't see a big problem with using standard tools and kernel provided things

edit: i see a problem
cat can be killed with a couple char's in buffer
solution, dd if=/that/fifo of=/that/log bs=1

[lcomputer]

Yes the first thing is you must learn actually what is the problem behind the reason the find solution or need to support from other.