/usrmove?

[Lennart Poettering]

On Fri, 10.02.12 11:49, Ralf Corsepius (rc040203 at freenet.de) wrote:

> On 02/10/2012 10:06 AM, "Jóhann B. Guðmundsson" wrote:
> >On 02/10/2012 04:45 AM, Ralf Corsepius wrote:
> 
> >>In this spirit, I eg. would propose to table usrmove for F17 and to
> >>concentrate on systemd integration and anaconda/grub2 improvements,
> >>both topics, I perceived as the "hall of shame of F16".
> >
> >Better systemd integration of services is not going to happen I can just
> >tell you that here and now
> Why not? Users are supposed to struggle with the swamp/mess the
> systemd integration currently is in? Could it be systemd reached its
> design limitations (== is a failure)?

I am not sure why you think things are so bad.

Yes, not all services are converted, but progress is being made (and JBG
is doing a great job at it!), and we do have SysV compat support that is
quite comprehensive (and where it isn't this is extensively documented:
http://www.freedesktop.org/wiki/Software/systemd/Incompatibilities).

It is hard getting the whole distribution converted, but I am quite sure
the way we handled it is pretty much the only way this could ever work:
if you don't push it in the distribution people don't see the need to
convert their stuff, and the change never happens. What is key is that
we convert the core and then pull the rest with us.

Fedora is much more bleeding edge than any other distribution. It's in
our motto even, the fourth F stands for "First". Being the ones who push
the envelope means that we have a rougher ride then the ones who just
follow. Fedora is the pioneer in Linux development, and that's great
that way, but this comes at the cost that sometimes things don't work
as smoothly as they might if people would always only follow what other
people already did.

The IT industry is quite competitive and it advances faster than almost
any other industry. The idea of "Hey, let's stop all progress for now,
and spend a year on nothing" is not how you get to the top, and I do
believe that that's where Fedora and Linux in general belongs. In fact,
we probably need to move even quicker rather than slower if we want to
outclass the competition comprehensively.

Technology generally suffers by something like the uncertainty
principle: you can have everything well tested and ultra-stable, or you
can have all the features and support for new hardware. But you cannot
have both. While Fedora is something different to all of us, I think one
thing is clear: that Fedora should have things First, that's our motto,
and that means that not everything is as well tested as it is in RHEL or
CentOS. Accepting this means that we will always have a few hiccups
before things have stabilized perfectly. It's a necessary part of the
game, and it won't change in the future either. As long as we decide
that Fedora is where things should be available first this will stay the
same.

You can't have a pony and eat it too.

Or in other words: if you want to make big changes harder or impossible
and drown them in even more politics and bureaucracy then you change
massively what Fedora is. It will just become another Debian or
Slackware then, which just gets everything after everybody else had it
for 5 or 10 years.

And coming back to the /usr move in particular: I am not sure you are
aware of the amount of politics such a move involves and how exhaustive
dealing with that is. There are only very few people who are willing to
go through this, and who believe enough in Fedora to go for it. I am not
aware of anybody who was willing to deal with all these politics to
accomplish such a massive change before Kay and Harald stepped up.

I am sure there are many folks who'd prefer if Linux would just stop
developing at all and would continue to look exactly like an OS from the
1970's with sysvinit, and shell glue. But honestly, that's 40 years
ago. And this is the computer industry. It doesn't wait for you. It just
ignores you eventually if you stay behind.

Lennart

-- 
Lennart Poettering - Red Hat, Inc.