Trends - how to save Fedora ?

[夜神 岩男]

>> Red Hat as a company is poised to be a billion dollar company this year
>> (FY12). The FY 2006 earnings were $278.3 million.[1] That's a 4X
>> increase in just 6 years. That's *amazing* growth.
>
> Yes, it is. But it is also a reflection of economic decline, financial crash,
> IT crash that make "free" software attractive, even necessary for survival.

Capitalism mimics nature: chaotic, violent, cannibalistic, and promotes 
progressive adaptation above all other things. This is not a sign of 
economic decline, but a shift to a better mode of operation. When the 
web is recognized to be something other than the OS/development platform 
it has been mistaken for of late it will not longer be a fad absorbing 
gobs of trash funding -- and that will not represent economic decline, 
but rather a structural correction within the market.

You won't be lamenting the decline of internal combustion engine makers 
when a new "cleaner" method of energy conversion is developed to replace 
current automobile engine -- because it is politically and socially 
unacceptable to lament such "dirty" things. IT is no different, just 
less politically charged in the eyes of the general population because 
they understand that they don't understand it well enough to have strong 
opinions on most points (whereas everyone is an expert in climatology 
and planetary cosmology). The IT market is massively overweight, 
overvalued and engages in enormously wasteful development practices 
right now. Open source development for the most common of software 
system elements + a revenue stream based on hardware sales and computing 
services provision (a very broad category worth huge money on its own) = 
a better model for the customer. IBM knows this. Intel isn't too happy 
about this. RedHat has placed itself at the most critical part of the 
process as the servicer. Microsoft is done creating success and is 
scrambling to now not create failure -- which is a really bad operating 
mode for a business (IBM was there once itself). That's just good 
business on RedHat's part and indicates a mature market understanding on 
the part of IBM.

>> Look at things like http://fedoraproject.org/wiki/Statistics, which
>> indicate that downloads and torrents are going up with each release, not
>> down.
>
> That does not mean much - what sticks, counts.
> I downloaded F16, it gave me a big kernel dump with other errors - it is good
> for my dev machine as a reference of what is going on, but not good beyond
> that.

Fedora was never intended to be useful in any other way to you. That it 
in fact is far more useful than that in most use cases is a testament to 
how coherent the Fedora project really is, despite its pace of 
development. And that is pretty amazing considering we lack a common 
architectural goal or vision.

> Fedora is unstable, release by release, progressively worse.
> It is becoming a dump place for projects that are pushed by RH and
> automatically sanctioned by its subordinates here at Fedora (some of them
> admit to be torn between job loyalty and doubts), without consideration for
> their sometimes questionable goals, quality, effects on system stability,
> adherence to UNIX principles, lacking adequate testing, in short too
> disruptive even to pre-conditioned Fedora community.
>
> There is a lack of independent users representation in Fedora project's
> governing bodies who should and would be able to be more critical and stop
> some of this damage even before it enters the actual development, not to
> mention implementation stages.

I'm not on the board, but I'm an independent outsider. I don't like the 
state of systemd. I liked SysV because I know it well. Spending time 
reading the systemd-dev list has convinced me that systemd is actually a 
good idea, just not one that is fully implemented yet. Very importantly, 
it is also not suffering from the problems that Hurd only recently 
overcame on the project level. I expect great things from systemd -- 
within a year or so. Until then, RHEL or SL are fantastic stability 
options -- and Vine fits my wife's needs perfectly without being too 
different for me to manage for her.

> SELinux is a static, straightjacket-like security control system, badly
> designed with its requirement for off-line system re-labeling, ineffective and
> inflexible for ad-hoc installed packages, with incomprehensible/non-intuitive
> psedo-scientific naming convention for control labels, difficult to use and
> judge by an average sysadmin and user (which mostly results in accepting
> problem cases as valid exceptions, or filing Bugzilla reports which makes
> the maintainer and RH services unavoidable).

Have you ever configured Sendmail or tried to write a common coding 
specification for a web application which is supposed to run in Chrome, 
Firefox, Safari and IE? (Or just for starters tried to make sense of the 
new Firefox cycle or figure out what is going on in Chrome's hacked-up 
bundled libraries before the next version is already released?) SELinux 
is a snap compared to either experience.

SELinux is designed precisely to be a straightjacket -- the prisoner can 
get air, and the exact amount of allocated food and water to his mouth, 
and mumble just loudly enough to ask to poo; and that's all its supposed 
to be able to do. Ad-hoc installation of packages is the root of many 
security evils, and SELinux is specifically designed to play mean and 
unfair games with unknown quanities. Somehow I haven't had trouble 
installing software from source, developing on non-root accounts, or 
really doing anything once I learned just a tiny bit about SELinux (and 
I really mean a *tiny* bit).

> GNOME 3 is an example of how not to do it, also influenced by RH devs.

I don't like Gnome 3, either. When it came out it made me unhappy, so I 
tried the whole XFCE thing out. Its OK. But my SL and Vine systems are 
so comfortable with Gnome2... but then I actually gave KDE a real look. 
And... I'm really impressed. There are alternatives, and some of them 
are pretty amazingly good if you set your preconceptions aside. The 
reason I'd never really known about KDE4 before was I had followed the 
general sentiment against it when it came out (it was trash early on 
anyway) and permitted that early prejudice to color my thinking since 
then. It was about time to take a re-look, and I'm glad that I did -- 
the paradigm is somewhere in between the Gnome3 and Gnome2 worlds and it 
is at least as full featured as Gnome2 was with a lot of subtle, slick 
features for developers thrown in there that take time to discover, 
which suits the way I used the system just fine.

Anyway, I'm not trying to plug KDE, SL or Vine, but rather point out 
that there are both distro alternatives and very featureful desktop 
alternatives for you within Fedora. But don't be surprised that 
something about Fedora changes. Everything about Fedora changes in very 
complete ways with regularity. And it should.

Someday when we live in a world when the walls of my house are gigantic 
tablets and laser detection of my eye and finger movements makes 
wish-clicking a normal thing, I'll give Gnome3 another look. Saving 
that, the project might fix some of the silly interface ideas they have 
implemented and draw me back before that. Either way, its a project 
worth watching, whether or not I'll ever be using it myself.

> With regard to Systemd, it is the most recent example of non-UNIX-like (or
> more like old M$-like) approach to software develoment. It is obvious by its
> goals, design, and reaction to criticism - they are not of UNIX mind ...
> Linux API to be a new standard, over POSIX. Screw up everybody else ...
> Integration of GNOME desktop and systemd in sight ... on the way to integrated
> old style M$-like desktop and system that ironically M$ is trying to get away
> from ? LOL !

Fortunately it isn't quite as bleak as all that. I was afraid of the 
same initially myself. If things do go that way then yeah, Fedora and 
RedHat will have not merely shot themselves in the foot but blown their 
lower half off -- and someone else will fill the gap there. That's fine 
with me. Someone smarter will just win and I'll be using their system 
instead -- no reason to be a loyalist if the project starts doing things 
that colossaly stupid.

As far as Linux API vs POSIX -- that is a valid point on the surface, 
but the POSIX specification is actually really vague. It is not 
unimaginable that the intent is for the Linux API to be a stricter 
definition that fits (at least mostly, if not entirely) within the POSIX 
specification. That's not unreasonable at all, actually. Consider 
Fedora's policies regarding FHS. Fedora-compliant means something a bit 
different than FHS compliant -- but the Fedora spec fits within the FHS 
spec by being more clear about things and disabiguating the things that 
flimsy committees built around the sweet dreams of consensus aren't 
competent to put their foot down about in the context of the bitter 
nightmares of reality.

Blah blah... I'm done blathering and will resume my net coma.

tl;dr: Economic progress is always creatively destructive by nature, and 
this is a good thing. Things aren't as bleak as they seem. Open source 
can still move in any direction that you are willing to move it in (but 
not by trolling mailing lists). There are alternatives if this isn't 
your cup of tea.

-Iwao