Alan Cox Calls Fedora 18 "The Worst Red Hat Distro"

[Xake]

Oh yes, it is.

I want to see your bugreports to prove it.

You are somewhat right in that Fedora may be a bit more flakey then Ubuntu, but my experience is that that is because Fedora actually fixes teir bugs in between releases. Sometimes that introduces new regressions. But they are pretty fast on resolving those IF you report them.
That said, Fedora may not be what you want to run at your company productivity 24/7 server unless you know what you are doing. But most Fedora people I meet tend to run CentOS there if they do not have a reason to pay for support.
Fedora is the more "set up a test server and see how you may handle that migration lurking on the horizon".

[BSD SUCKS DICKS]

What ever happens to Fedora (and for all Linuxes), it's still 1000000000000000000000 times better then BSD

[GreatEmerald]

So a couple of reasons...

1. the oldUI code was not cleanly separated from the backend, they were pretty much tied together.
2. the old UI kinda sucked. I mean, it wasn't good. Re-doing it has been on the table for a while. Just because the new UI has some teething problems doesn't mean the old one exactly cured cancer - tell me http://media.if-not-true-then-false....ation-type.png is a great screen, for instance (lots of people got *used* to it, but take a step back and tell me it's an awesome screen). Or the screen you got when you clicked on 'Change device' on http://media.if-not-true-then-false....oot-loader.png (sadly not pictured in that walkthrough), that was awesome. Or my favourite giant ball of win, and pretty much where Mo started seriously looking at the installer design: https://fedoraproject.org/w/uploads/...enshot_f14.png .

If you read through the archives of Mo's blog at http://blog.linuxgrrl.com/category/fedora/anaconda/ , which I really highly recommend doing, you can see where a lot of the newUI design came from, and a lot of the problems Mo identified in the old design. It's not just redesign for redesign's sake. The post on creating a RAID /home is particularly interesting when it comes to the new partitioning approach - http://blog.linuxgrrl.com/2011/12/14...g-ui-thus-far/ .

Hah, well, I have the benefit here in that I have never actually seen the old Anaconda (Well, I suppose I have, but it was so long ago I forgot all about it). So, looking at the screens of the old UI you showed here, I can see how they would be sub-optimal, but by no means impossible to understand. The installation type selection screen is pretty clear, if not for those checkboxes at the bottom, and I would never choose anything else than custom partitioning, anyway (because you can never be sure what the automatic options would do). The warning is quite severe though, I'll agree with that.

I read through that post, and urgh. The solution feels like demolishing a building and rebuilding it anew, just because the door was too small. Even in those mockups, the design is completely counter-intuitive... If I want to have a partitioning scheme, I would like to see partitions. Not buffets in an RPG. The selections there feel a lot like the old installation type selection, except that it has no "manual" button. It's hard to tell what exactly will be done. Now the part with RAID options is pretty good, that's true, but everything around it contains no useful information... I'd say that the ideal approach should have been fixing the suboptimal choices in the old UI, without throwing the whole concept out. Make it an evolution of the UI, not a revolution. But I suppose nothing can be done about it now. Well, except for making it intuitive again, of course, and I think there will be plenty of ideas on how to do it now, since the first version is out.

Logging into the desktop as root is never correct, as it causes all kinds of processes which are not intended to run as root and do likely do not have proper protection against running as root to run as root. Logging into a console as root is of course fine, though some would argue sudo is a superior model.

The superior model is polkit. And I'd really like to see it implemented into YaST, actually. There was that time when Linus said that it was crazy that you needed root privileges to set up a printer, and he was right, but then it's the whole root privilege model that is flawed to begin with. When opening YaST (or Mageia control center, or any equivalent configuration tool), you usually need to give a root password in order to get to the needed options - and that's again silly, because you are not yet doing anything that requires any privileges. Viewing options should not be restricted like that. It's only applying the options that should require them. And even then, only specific privileges, not privileges to do whatever the tool may please.

So I see the optimal workflow as opening the setup tool, selecting all the needed options, and once that's done, hit a big "Apply" button which then gives a summary of what you changed and what privileges will giving the password grant the program. If you have not changed partitioning, the program should not be allowed to make changes to the partition table, that's only natural. If you see it requiring more than that, then the application must be malfunctioning or compromised. So in the end you need to enter the password only once, and you know exactly what the program will do.

[chithanh]

Remember: logout by default is not shown if there's a single user and only the Gnome shell session is available.

At least one poster in this thread has confirmed that this is not the case.

Really, that's what this silly fuss is about: a default that makes sense for the less techincal user and that whoever needs differently can override in about 30 seconds. The computer version of "first world problems", I'd say

What users point out is things that used to just work are now hidden behind obscure configuration options. The continued dumbing down of the interface until it is useful for exactly nobody. Hiding complexity from the user in a way that makes the system actually more complex and less predictable (remember the disappearing shutdown option issue from the dedoimedo review).

"Fedora 17 had an installer that worked just fine. So you could have continued to use that, and optionally provide your new installer to those who want to test it."

That's not how things work in practice at all. You are by all means free to confirm this for yourself: take the Fedora 17 installer, plop it down in the Fedora 18 package set, and try to build an installer image. Ten to one it won't even compose, but on the offchance that it does, the composed image will _certainly_ be entirely busted. It won't work at all and will probably eat babies.

Thank you for taking the time to reply in detail.

Of course past a point where the old installer was abandoned and unmaintained you cannot just go back to it. The backup plan needed to be in place and made sure to work during the whole development cycle. The lack of backup plan meant that you were now out of options and stuck with the new installer.

"You could have recommended users to ditch the installer altogether and link to a document instead which describes manual partitioning + febootstrap as the preferred install method (Gentoo does it like this, and Arch has recently adopted this way too)"

That's...interesting, but really not in line with how Fedora rolls, I don't think. It would have been just too drastic for a single release.

I don't agree. Admitting that you cannot recommend this installer because it does not yet meet the quality that users expect, and promising to deliver something better in the following release would have certainly been met with understanding rather than rants and angry critiques.

newUI really isn't that terrible, you know; quite a few people have been saying 'it worked fine for me, I don't get the fuss'.

"You could have labeled Fedora 18 as "Forever Beta" and not make it an official release, just something for the interested."

That again would have been a valid approach, but I'm not sure it buys us a lot. At this point we have the same product and it's just about massaging the messaging. That's something we already tried to do with the release announcement and other release documentation, and see how much slack that's bought us.

I do think that some psychology is at play here.
newUI is bad enough for some high-profile users to switch to another distro. How is that not terrible? Remember these are not users who are indifferent towards the distro they use, they have chosen Fedora and put trust in your releases. Now they feel that their trust has been betrayed.

And that people don't read the release documentation should not come as a surprise. Look, http://fedoraproject.org/ says "Fedora 18" in large letters so they think that is what you want them to download and install. The release announcement is not even linked from that page afaict.

[V!NCENT]

The privilage model is not flawed, but not perfect either. That is why Fedora has SELinux (Security Enhanced Linux), that implements MAC (Mandatory Acces Control).

You should not even have to run anything as root, except some commands, in which case "su" in a terminal is better than "sudo", because with sudo you either constantly have to type the damn sudo command, or automatically get root rights for a period of time, so the user looses track of what commands run as root, and what runs as user. Better just have a root terminal where one can be sure what commands will run as root and what commands run as user.

[V!NCENT]

And what the hell is a "high-profile user"? Someone who uses his computer for a lot more than just simple avarege tasks? I would expect that 'power users' would know how to get around a simple fscking installer... :S

[birdie]

well birdie, maybe it is time for you to stop posting bullshit?
LINUX has stable APIs AND ABIs for more 20 years.

Internal constructs are neither APIs nor ABIs.

Please stop posting until you started to think. People like you with their 'give me. NOW' attitude and no clue at all make me sick.

Stable API/ABI? Are you out of your mind? I guess either you are batsh*t crazy or you don't understand what ABI/APIs are.

Look here http://upstream-tracker.org/ - and keep telling me about imaginary stable API/ABIs. I've never seen such an idiot before.

Maybe this http://www.mjmwired.net/kernel/Docum...i_nonsense.txt will set you straight.

[GreatEmerald]

You should not even have to run anything as root, except some commands, in which case "su" in a terminal is better than "sudo", because with sudo you either constantly have to type the damn sudo command

Oh man! Writing five symbols is just so difficult! Especially since there are few tasks where you need extra privileges in rapid succession, anyway.

[uid313]

In conclusion, you don't give a five year old the keys to the car and tell him to drive. By the same token, you control who has root access in GUI mode.

Even if you want to do an administrative task from a GUI you still do not login as root!

You run 'gksu nautilus' to run an instance of the file manager with superuser privileges.

[eliac]

At least one poster in this thread has confirmed that this is not the case.

What users point out is things that used to just work are now hidden behind obscure configuration options. The continued dumbing down of the interface until it is useful for exactly nobody. Hiding complexity from the user in a way that makes the system actually more complex and less predictable (remember the disappearing shutdown option issue from the dedoimedo review).

...

Well, there's a bug where the item doesn't appear until the next time gnome shell is restarted. Still, the designed behaviour is exactly the one I mentioned: no log out if you're the only user _and_ there are no other session options in GDM.
And frankly, most users don't care about "obscure configuration options", because most users won't have to change a thing. The minority that has reason to - and I'm among them - has no trouble doing so.

Let's be clear: most normal users out there don't set up a workstation as a file server, nor do they install multiple desktop environments on their computer. A good percentage of us do that _because we are a niche of tech-savy users with different needs and skills_. But if you design for the mainstream users, then you have to choose your defaults and behaviour in a way that makes sense for them, not for your current actual niche.
How many real users do you know that had a reason to log out in a single user single - single session case but were unable to set the needed gsetting? Because I know nobody that wasn't already looking for a reason to NERDRAGE about Red Hat conspiracies and "dumbing down of interfaces" and that had real trouble with this issue.

As for the "review": bugs happen. I never met the specific bug he mentions and I don't know if he even reported that. I know that he harped about the boxes icon (because he apparently can't _understand_ what it represents, so it must be broken) and then was unable to actually use the boxes to manage his VMs, going on to literally draw troll faces over the window of an application when he couldn't understand how it worked.
He's free to dislike anything and to critique anything, but I think I'll not take him as an unbiased source when it comes to the software quality of Gnome Shell.