New subject: Sorbet (Sherbet) no longer on Fedora's future but on bios merits

Saturday, 20 March 2010

For a while, I've been arguing with very knowledgeable people here
that there are way too many bugs in Fedora, bugs that either hinder a
pleasant user experience or plainly break systems to the point that
one wonders if he's not being hacked. And, for a non-geek like me, get
rid of them before new ones add to the heap, is just impossible.

Developers might not be aware of some bugs I'm experiencing because
they're manifestly hardware related(1), while some others can't have
escaped their attention(2).

(1) The only option available in my Gigabyte MA770T-UD3P's BIOS
offering only options for entering passwords, for exemple.

(2) For instance, "New File" entering the clipboard every time a new
file is created.

Some bugs reports, even filed by Red Hat employees, have been
outstanding for so long that most users certainly feel it's no use
filling reports and following the outcome... unless one wants to make
a full time job arguing with geeks on what is worthy bug and what is
not.

In Linux Weekly News, Mr Sorbet... err, make this Corbet, has written
a nonetheless delightful article on the matter of what is causing this
avalanche of bugs in so-called "stable" Fedora releases. To me, the
sorbet of the whole article pretty much freezes down to this:

"(...) the system which Fedora has in place for the review of proposed
updates - Bodhi - is often circumvented by updates which go straight
out to users. The testing and voting which is supposed to happen in
Bodhi is, in fact, not happening much of the time, and the quality of
the distribution is suffering as a result. So some Fedora developers
are looking for ways to beef up the system."

https://lwn.net/Articles/377389/

And rightly so, since not breaking stable releases is the most
fundamental Fedora rule, as expressed here in the Stable release
update vision:

"The update repositories for stable releases of the Fedora
distribution should provide our users with a consistent and high
quality stream of updates."

This, and more very important stuff, under "Vision Statement" at this URL:

http://fedoraproject.org/wiki/Stable_release_updates_vision

If this is the Fedora's game, I'm wishing to play. Otherwise, I'll
move to Ubuntu or, as security is important to me, CentOS or
Scientific Linux, soon as RHEL 6 is released.

So, one might ask, what will the contribution of non-geeks to Fedora
be? Well, as I said, I have a problem with my mobo. I also can't get
sound through HDMI to my TV. A recent update has made playing DVDs
impossible... except with Kplayer! (not KMplayer) Etc.

So, if there was a place where I could report those bugs without
registering to 10,000 different bugzillas and dealing with
don't-give-a-shit geeks, I certainly would fill them and would be more
than interested in trying packages in update-testing to see if the fix
works. But I'm certainly not interested in enabling update-testing
just to see if new stuff i don't need works, and possibly break my
system.

If my problems do not concern Fedora/Red Hat developers directly, they
can address the bug to software developers upstream. If bugs take
years to be fixed, maybe they can suggest another software... or
desktop environment be used by default on Fedora. You know, Fedora/Red
Hat certainly has the clout to wake up developers. OTOH, if Red Hat
relies on disgruntled users to fill reports on bugs that never get
fixed, users won't be the only ones to suffer.

As for users/developers who feel more "adventurous", Rawhide does
provide enough of a stampede experience, I would think. That's the
rolling distribution that some are asking for, though, even in this
case. I wonder if it wouldn't be a good idea to permit new updates
only every Sunday. But I don't have a solid opinion on this.
Developers are better placed to make an informed decision on this
matter.

Then, as suggested by Matthew Garrett, before a package is moved from
Bohdi to update testing, it should receive the signatures, or "karma",
or whatever, of 3 developers. Developers know each other. If somebody
doesn't do a good job, nobody will want to sign for him. If somebody
always output a clean job, others will almost sign eyes closed.
Signatures put pressure on developers: they know that if their
software always has problems, nobody will sign for them.

Of course, certain projects are more obscure than others, their
software is not as common as, say, a word processor. But the same goes
for the kernel development and, as far as I know, everything bears 3
signatures.

What I wrote here might be in part ill founded. When you're not a
developer, you can't comment with insight. It's an outsider's view,
but a very clear fact remains: whether it's only a rant or a
fullfledged case study, users must be allowed to express their POV
freely and it should be taken into consideration.

Chasing users away with "Why don't you fill (no-use) bug reports?" or
"You don't like it? The code is there, modify it!", the way it is
typically done on Debian and Slackware groups, leads to disaster.

If flame wars wouldn't have been so common in the community, if user
needs had been better taken care of, Debian could have achieved what
Mark Shuttlewort did, which is build a community, the largest user
base in the Linux world. So that, if you speak to Windows users
contemplating a move to Linux, the first distro that comes to their
mind is Ubuntu. That's because it's pretty much the only distribution
the generalist press talks about.

Now, Shuttlweworth is planning to offer an iTune look-alike service
for his users. He's going to bring some money in to pay his
developers. It's not the financial clout that Google gives to
ChromeOS, but it's a move in the right direction.

It's very strange, but it seems that open-source developers like to
pay their bills just like anybody. As more and more major companies
are entering the "open-source market", Nokia and Intel, for example,
who's going to be left developing open-source for nothing in project
that more and more will look rather futile, compared to mainstream?

Because, believe it or not, open source is now becoming mainstream, As
I explained here:

http://lists.fedoraproject.org/pipermail/users/2010-March/368642.html

(Read from "As Wikipedia puts it". The rest is of no interest.)

Google is apt to turn competition into confetti.

Apple, which is certainly far from showing an open-source attitude(1),
makes billions with BSD-based OS X using "repackaged" standard
no-real-specs-available(2) hardware and still gains market share over
Microsoft.

(1) Please don't bring forward this nonsense about their contribution
to WebKit. WebKit was a fork from KHTML, a GPLed project. So the code
had to be opened. and whatever contribution Darwin makes to the open
source community pretty much amount to a drop of water in the sea.

(2) Try to get the specs of their mobos, for instance.

At 19 years old. Linux is certainly not a new kid on the block
anymore. How come, even with Ubuntu, it is still howering at around 1%
of the market share? How come all the brawlers who invade Linux
groups/forums/lists are still allowed to bash new users pretending
that market share is not important in order to be accepted in
standards definition, that they'll still be surfing the net with Lynx
ten years from now?

The benevolent dictator would not permit such nonsense to happen
repeatedly on this groups. Why does Red Hat, a company listed on the
NYSE, allow this? Do all the non Red Hat members on the Fedora board
agree with this?

Can't anybody notice that traditional little budget open-source is
right in the middle of the track where the large corporations'
open-source is riding full steam head? Brawlers -- and you know, it
might be two people with 5 email accounts -- ask that they'd be taken
at their face value as real Linux advocates, and they do provide a
useful technical hint once in a while, but who's interests, knowingly
or unknowingly, are they serving?

Anyways, that's more than enough on brawlers: some people might think
I have somethings against them :) Documentation, now.

When, I installed the NVIDIA drivers, I went, first place, to
fedorafaq.org. Since Fedora couldn't provide instructions on
installing proprietary drivers, it seemed like an appropriate place.
But the instructions didn't work. It took some time before I got to
rpmfusion, the provider of the kmod package. Still, though I asked the
maintainer to correct his instructions, the Fedorafaq page is still
unchanged.

How come anybody is allowed to use the Fedora's name to give wrong advice?

Now, I want to remove the kmod-nvidia driver and there's no
instructions on how to remove it, either at rpmfusion or
fedoraproject. (I suppose removing proprietary software is not against
the law!) Just as for installing, you receive different advice all
over the place but, if the process can't be automated, how come there
is no offical information on such a fundamental matter?

Anyways, I could go on like this for hours. If Fedora stagnates behind
Ubuntu for a total Linux market share of ~1% -- servers excluded, of
course -- it's not because God cursed Fedora, it's because there are
HUGE administrative problems. The project lacks direction, a
benevolent dictator (or an enlightened triumvirate?), somebody that
could be pointed at if everything turns to a mess like now.

All along this summary, I've been comparing Fedora and Ubuntu, but
both are in a very different situations. Ubuntu is based on the latest
version of Debian, which is already very stable. Then, some packages
from testing are added and tested. If some bugs in unstable are
related to hardware architectures that are not supported by Ubuntu,
the package might also be added.

Though, due to the wide variety of PC hardware, all kind of problems
pop up after an Ubuntu release, after, say, 8 months, inexperienced
users can feel confident that adopting the past release -- a new one
coming out every 6 months -- will be fairly trouble free. Every
release being maintained for 18 months, the user can, 10 months later,
choose the new release, the last or the second last. And it seems
Canonical can come up with fairly stable server releases this way.

With 13 month releases, Fedora users have to upgrade much sooner. And
the latest version being based on the previous version of Fedora
instead of Debian, some question the validity of the whole process.
Here's what Corbet writes about this;

"Fedora does indeed not hold back on updates; a quick look in the LWN
mailbox turns up over 600 package updates for the Fedora 11 release -
in just the last month. This is a release which is scheduled for
end-of-life in a few months. Many of these updates involve significant
changes, and others have been deemed "worthless". Regardless of worth,
there can be no doubt that all these updates represent a significant
degree of churn in a distribution which is in the latter part of its
short life. It is difficult to avoid breaking things when things are
changing at that rate."

So, when I hear some suggesting that Fedora moves to a rolling stable
release, this sounds to me like a recipe for disaster. Of course,
stricter control on updates is often suggested, but getting to a
non-rolling stable release certainly seems like an inescapable first
step.

IMO, Fedora releases will have to become much more stable and urgency
to get more market share will have to be established as a clear
priority. Very F-A-S-T. The Stable release update vision should be
followed "à la lettre".

P.s.: Many thanks to Jonathan Corbet for providing me with insight on
Fedora's release process.
To all: Brawlers might find my answers have a sarcastic tone... if I
give any at all.

Sorbet on Fedora's future