[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#727708: Fsck SystemD and its developers and its users. GR to override this please.

Hi John

On 10/02/2014 20:41, John Paul Adrian Glaubitz wrote:
> On 02/10/2014 06:47 PM, Clint Byrum wrote:
> Neglecting reliability and maintainability for the sake of being
> able to choose such a core component is a bad idea. I do not
> think it's really feasible to maintain several init systems, it
> just affects too many components of the system.
> We don't even manage to maintain two versions of ffmpeg (the original
> and the fork) even though many users actually prefer the original. How
> should this even work with the init system then?

I recommend reading all the previous discussions on the topic. Over the
last year, it's become clear that in the short-term at least, a very
minimum of 2 different init systems would need to be supported. At least
sysvinit, because there are so many packages that ship with
configuration and it won't be possible to convert them all in one
release, and also because the alternative operating systems that Debian
ship with won't work with Upstart or systemd.

sysvinit has hit its limit with it's dependency-based nature, and the
only way to fix many outstanding bugs is by switching to an event-based

So, supporting at least 2 init systems in the short-term will be
necessary. Upstart is pretty cheap to support, since it's Ubuntu's
default init system and a very large amount of packages have Upstart
scripts already. systemd has a very enthusiastic community and it's
likely that many packages will have systemd configuration files by the
time Jessie is released.

There are so many possibilities going forward. It's possible that
Canonical might give up on Upstart since the rest of the world has
pretty much gone with systemd on Linux. Or, it's possible that Upstart
would better support the niche systems / toy ports which would make it a
good candidate for replacing sysvinit on those systems. Or perhaps in
hindsight it might just turn out that Upstart was a better, more
logical, sane, level-headed, secure, unixy and unpoetered solution and
the technical committee will ask themselves "what were we thinking!?".

Historically though, it seems like it has consistently been beneficial
for Debian to support all the available options and let things evolve.
Maybe it's just better accepting the choice and move on even if you
don't like the choice.

When I saw the announcement this morning I thought "Oh well, at least we
can have debian-devel back now." I think if anything has been said
before it doesn't have to be re-hashed to infinity. This is one of those
things that you just have to learn to let it slide a bit.


Reply to: