OSNews

Speaking of "Extras", when I clicked on the checkbox on my DVD installation, it crashed the installer. IMHO, that is pretty pathetic testing. I have never had a problem on my machine with a Linux installation since I first purchased it over 5 years ago. And yes, I made sure SHA1 matched and that there was nothing wrong with my DVD.

Extras was explicitly tested to work on all architectures

http://fedoraproject.org/wiki/QA/FC6TreeTesting

A reviewer even confirmed that it worked. Several users did too

http://www.phoronix.com/scan.php?page=article&item=573&num=1

Maybe you ran into the heavy traffic problem which made many mirrors slow or non working for the first few days after the release

You only require two CD's for the default set of packages or you can use the 6 MB boot image to do a network installation.

It is possible to use custom / third repositories or Livna if you want proprietary packages

http://rpm.livna.org/rlowiki/UseLivnaInAnaconda

"""Extras was explicitly tested to work on all architectures"""


You always seem to be going on about how this or that in FC is well tested, using cool methodologies that none of those other distros use. And yet, whenever I try to use FC for something, I end up with a show stopper bug that demonstrates that the testing, if there was any, was ineffectual.

This time it is that USB storage devices are not recognized if they are plugged in at boot time. Hotplug works. Cold plug consistently does not. (!)

Known problem. Bug report 2 months old. (At least this one got responded to after a month, though there has been no activity for the last 2 weeks.)

Oops.

And a bad deal if that usb storage device is a samba share that people depend upon.

Every so often I hear enough from people who are involved in FC talking about how great their testing framework is, and who vow that they are out to prove that FC is not just RedHat's beta, that I forget why I switched my clients from Fedora to CentOS in the first place, and try to use FC for something.

And I am always reminded of why I stopped using it before, and why this recent quote from Brian Stevens, RedHat's CTO, is so relevant:

"""We're convinced that there is a better way to develop software, so what we did is we blew up the notion of an Alpha and we use Fedora as an alpha. The engineers are goaled on not just producing enterprise quality software, but driving it through upstream in terms of the community."""

Source:

http://www.internetnews.com/dev-news/article.php/3628476

Edited 2006-10-28 02:15

The CTO quote was a unfortunate choice of words in relationship to RHEL usually taken out of context. So you get a reply here in

http://www.internetnews.com/dev-news/article.php/3639616

Anyway I have added the static ip+ extras repo configuration checks to the default QA matrix now.

http://fedoraproject.org/wiki/QA/TreeTestingTemplate

Would you point to the USB storage bug report?

Edited 2006-10-28 03:22

"""Would you point to the USB storage bug report?"""

Of course:

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=204396

After reading your link:

What I see there is that RedHat, the one paying the salaries, sees Fedora as their alpha. (It's hard to imagine "we use Fedora as an alpha" being taken "out of context", as you put it.

Spevack, the Fedora project lead, says it's not. (Surprise!)

My experience is that it's about beta quality... perpetually. Well, to be fair, only for about 9 months. After that, it's just unsupported.

Edited 2006-10-28 03:51

Max Spevack is a Red Hat guy leading the Fedora Project.

http://fedoraproject.org/wiki/MaxSpevack

So its not a question of Red Hat vs non Red Hat. Basically from the perspective of RHEL development, there is no alpha currently since all work happens in a public fashion in the development tree of Fedora. That would be evident if you see we blew up the notion of an Alpha .

When you get right down to it, what really matters is the performance in the field. I have users that are still on FC4. (On machines on firewalled networks only, since FC4 no longer has any support whatsoever: http://lwn.net/Articles/204722/ )

I have moved most of my clients to CentOS. (Most are on 4.4.) But I'm currently evaluating FC6. So I have a pretty good comparison view of the two distros.

Putting all that together with what I've read of the positions of RH and the Fedora guys, I would say that RH is willing to fund an alpha. (Think: "Eats your Brane!

The developers want to be able to call it production quality, so they go an extra mile. (Kudos!)

And the end result is about beta quality.

My gripe is not so much with the beta quality, as it is with people trying to claim that it is production quality when it isn't.

This really goes beyond just the number of bugs upon release. It extends to how those bugs are treated. How many bugs on bugzilla ever get a response? How many are resolved? How many end up "fixed in rawhide"?

Plus, as the above lwn.net link shows, Fedory Legacy... isn't relevant. There is no tangible support for Fedora Releases after they are "handed off" to Fedora Legacy.

Fedora has its strengths. But it would be better if everyone would just let Fedora be Fedora and not try to claim that it is something that it is not.

Edited 2006-10-28 04:54

If you go with the default install you inly need 2. But then again you can do a netinstall with a 7MB boot iso and only use one disc.

Part of me is beginning to wonder though if this philosophy of keeping out proprietary options from the default installation is what is putting Fedora behind Ubuntu. When it comes down to it most people just want an OS that works right out of the box.

It is not really an issue anymore because Zod's Anaconda allows user to install their favorite third party repositories so their can grab their favorite closed source format. It will be a matter of time that method will be quickly adopted by other developers.

So it crashes while using iSCSI and a static IP. Thats not a very regular combination. Maybe you can try using the updated image in that case.

Edited 2006-10-28 01:11

You did not read the filing correct. It says when select Extras or iSCSI. Selecting Static IP with Extras is very common. I know because I ran into the bug. And thanks everyone for "modding me down."

Edited 2006-10-28 01:13

I have had that bug too.Quite annoying.

... You are calling the installer broken just because it doesn't work quite right under a closed source Microsoft VM?
Most likely the virtualized VGA supplied by VPC support in X.org 7.1 (used by Anaconda) is broken.
Try filing a Bugzilla entry and in the mean type, switch to text installation. (boot: linux text)

- Gilboa

I found out what the problem is.

The problem is that X.Org doesn't query the ROM for what modes are actually supported by the virtual S3 Trio 64.

You can blame MS for not implementing 24-bit support in the virtual graphics card if you like, but fact remains that while other operating systems DO query for what is actually supported, systems using X.Org or XFree86 DON'T do this.

This has apparently been a known problem for at least 1.5 years, and yet the X developers don't seem to give a shit that their assumptions about what modes the card can handle are incorrect, and Linux distributions don't seem to give a shit about it either.

This has apparently been a known problem for at least 1.5 years, and yet the X developers don't seem to give a shit that their assumptions about what modes the card can handle are incorrect, and Linux distributions don't seem to give a shit about it either.

I'll ignore for a second the obvious troll mode to which you have entered (and the offensive language which has no place in here), and ask you this:
A. Did you report this bug to bugs.freedesktop.org? If so, what was the outcome?
B. Did you report this bug to bugzilla.redhat.com? If you did, what was the outcome?

- Gilboa

Strike my previous answer.
I did some reading about the subject and this is -indeed- (and only) Microsoft's fault.
The Trio 64 chip is a fully 24bit GPU and as such, the Anaconda may (or should) assume that it's fully capable of running in 24bits.
Cursing the X.org developers due to their reluctance to modify a fully working (and stable) driver just so they'll fix a bug within Microsoft's own virtualized implementation of the GPU is one of the worst displays of hypocrisy I ever seen in this site.
Before calling X.org and Fedora 'shit', have you even tried to reporting the problem to Microsoft? Or was it just easier to point the finger at X.org and at Fedora?

As I said, in the mean time, switch to text based installation. (linux text)

- Gilboa

I disagree with your opinion that Anaconda or X should make such assumptions, and I also disagree that it is a bug in VPC.

Like I said in a previous post, you can blame either MS or X.org for it, but fact remains that systems that don't make assumptions handle the virtual S3 without problems.

FYI, it was from Microsoft that I found out what caused the problem to begin with.

.... Sigh.

You do understand that neither Anaconda nor X-org have any way of knowing if the GPU is virtualized or not. (Nor should they)
-You- are asking X.org (or Anaconda) to assume that each S3 Trio GPU is broken, just because Microsoft was too -lazy- to fix their own bug.

As I said before, if you had a -real- Trio GPU and Anaconda malfunctioned, your rant would have been justifiable.
As it stand, your rant is pure hypocrisy.

- Gilboa

I'm NOT asking X.org to assume ANYTHING about the S3 Trio. I'm simply asking for them to actually ASK what the GPU supports. If non-X based systems can do it, why can't X-based?

http://blogs.msdn.com/virtual_pc_guy/archive/2005/05/09/415814.aspx
This does not cause a problem on other platforms as we have removed the 24-bit entries from our ROM - which most operating systems probe to determine possible displays.

If the ROM claimed that it supported 24-bit, but didn't, it would be a bug. Apparently they did have such a bug once, but that's been fixed already.

Edited 2006-10-30 16:29

Repeating the same point over and over again doesn't really serve your purpose - far from it.
As long as there are no real S3-Trio64 cards that have no 24bit support, X.org shouldn't waste one second fixing Microsoft's own bugs.
(And -again-, Linux isn't officially supported by VPC 2004... Which makes your point even weaker.)

- Gilboa

As an additional reply...

What do you think happens when Average Windows User decides to try out a Linux distribution, and because he/she don't want to go through the hassle of reinstalling in case he/she don't like Linux he/she decides to try it in Virtual PC?

Virtual PC because that's what AWU found before.

Average Windows User tries Linux only because he/she heard you can do everything in a GUI like with Windows. From installing to running programs.

Those pre-install options look scary so AWU just presses enter or F1 or whatever it is for the distro chosen. And the next thing AWU sees is screwed up graphics.

AWU wonders what is wrong, and tries to install Windows in Virtual PC instead to see if it's VPC or the distro that makes life hard for him/her.

AWU sees that Windows installs without problem in this program called Virtual PC.

Ok, then it's just this distro that is broken, AWU thinks next and uses Google to find another distro. Only to encounter the same problem during install.

AWU's conclusion:
Linux don't work, I stay with Windows.

------

If you subscribe to the notion that it's the virtual hardware that is buggy, would it be too much to ask for to get a workaround for this "bug"? There exist many workarounds for physical hardware bugs, so why not workaround for virtual as well? Is it because it's virtual or because it's MicroSoft?

As an additional reply...
...
AWU's conclusion:
Linux don't work, I stay with Windows.

Microsoft doesn't officially support Linux in VPC 2004. If AWU decided to try it never the less (instead of say, download the free VMWare Player or Server which do support Linux) he shouldn't blame Linux (or even Microsoft) when things break.

If you subscribe to the notion that it's the virtual hardware that is buggy, would it be too much to ask for to get a workaround for this "bug"?

Did you ask Microsoft to fix their bug? If so, what was -their- response?
Oh... If Microsoft really wants to support Linux, nothing stops them from adding their own virtualized driver to the X.org tree. (E.g. vmware-vga driver)

There exist many workarounds for physical hardware bugs, so why not workaround for virtual as well? Is it because it's virtual or because it's MicroSoft?

A driver should fix hardware bugs.
The virtual machine software should fix virtualized hardware bugs.
It's as simple as that.

- Gilboa

A driver should fix hardware bugs.
The virtual machine software should fix virtualized hardware bugs.
It's as simple as that.

Great. Then I just have to make a "buggy" S3 Trio 64 in real hardware, and they will fix this issue. Why didn't you say so right away?

Let me get it strait, you are using an unsupported product under another product that has a buggy implemntation of the original specs and you are asking X.org to clean up -your- mess?
Somehow hypocrisy sound rather like an understatement when I read your last post.

- Gilboa

check out http://rpm.livna.org instructions on installing drivers.

Totem-xine deliberately conflicts with totem since the Fedora Core totem uses the gstreamer engine and Totem xine engine cannot work in parallel with that.

#yum remove totem
#yum install totem-xine would work if you want to use xine.

Totem-xine deliberately conflicts with totem since the Fedora Core totem uses the gstreamer engine and Totem xine engine cannot work in parallel with that.

The problem is, you get that message even if you don't have totem-xine installed. Totem-xine acts as an update for totem; it isn't, it shouldn't.

That's because fedora uses aiglx and not xgl. And aiglx requires driver support. So you can´t do anything about it until ati releases better drivers.

Yes a bit bumpy is the ride.I don´t mind perse otherwise i would have installed CentOS.

Anaconda seems to have the same lvm bug FC5 also had.Am i the only one who uses raid0 lvm's?

... I have two machines with LVM on SWRAID0. Didn't really have any problem with this configuration.
(Same goes for LVM over SWRAID1 and SWRAID5)

Can you elaborate?

- Gilboa

[/i]Can you elaborate?[/i]

Yes,i have stored the exception on floppy.I can send it to you if you wish.

Though it's not happening all the time that's rather awkward.

Please create a ticket under Anaconda in bugzilla.redhat.com and attach the exception log.

- Gilboa

Edited 2006-10-30 06:44