SuSE engineer needs to get a clue

I am so f*cking pissed off atm with a SuSE engineer: Thomas Renninger.

For two months now I am providing the requested debug logs to fix a bug where the kernel 2.6 just doesn’t load up without passing the “nolapic” parameter. At the very end, after he has all the debugging info he needs, he will reply: “this is an old machine, it’s probably a BIOS bug, I won’t fix”. And this on a machine that kernel 2.4 worked perfectly, Windows/BeOS/BSD work perfectly, and only kernel 2.6 fails. You know, he is right on one thing. It IS probably a BIOS bug, but a modern OS must be intelligent enough to recognize such failures and find workarounds in order to deliver the right experience. Isn’t the Linux guys who say that it supports more older machines that Windows doesn’t?

MY ASS. People should stop believe such f*cking hype. Most GNU/Linuces require more special hardware and more RAM than Windows. In its popular configurations (RH, SuSE, Ubuntu, Mandriva) it’s a bloated piece of shit. And it shows every f*cking day.

And no, I am not unhappy because my pet bug wasn’t fix. It’s because it seems that these guys seem to have no product management and the engineers have no clue about real world marketing and what the userbase expects from their product. While as a power user I understand that messing with ACPI and buggy BIOSes is a pain in the ass for an engineer, my alter ego, the Joe User, expects this machine to work out of the box exactly because every other OS supports it. For the Joe User, this is a failure of Linux, plain and simple.

Post a comment »

Ludovic Hirlimann wrote on July 19th, 2006 at 6:03 AM PST:

Can’t be more than true. This is why linux will stay a geek’s os in the years to come. people do not care and can’t figure out arguments to pass to the kernel just to boot a machine. Joe users want things that work out of the box, like TV sets etc … Why should a computer be more complicated to use than any other device in the house ?

Roeland wrote on July 23rd, 2006 at 6:43 AM PST:

ok. so what’s your point ? You are not able to pass this option to the kernel ?

Are you just as fucked up by the fact that several pieces of hardware don’t work with older windows versions while they work with linux ?

the PIC requires some pretty advanced code. Some of the implementations, like the Abit BP6 for instance, are at the hardware level wrong. Still, the APICs are detected and being set up. Linux *thinks* that your hardware is correct working but in fact isn’;t because of your sloppy pet.

Why do other OS wkr fine ? Do they also use a different interrupt scheme ? Or does Novell keep old code so that newer hardware cannot be used, unless someone reuires an opposite command at boot time ?

The manual(s) that are written, cover how to add such an option. So what’s being done:

1) the bug is found but the work-around is useless as:
2) you can pass on a command at boot time that can be
placed into a config.

So in my opinion, the problem is solved but you refuse to have it that way. People help you and you are ranting your way…

Let’s turn around. For a specific HP All in one machine, under linux, I run yast, no CD’s are required, all out of the box. Cool, right ? It scans, it prints, it does all what you want.

Now, take the windows part: It requires a CD with well over 350 Mbytes of ‘drivers’ and requires additionally some 150 to fix a few things.

In your case, you tell MS that the printer doesn’t work. You send in debug info. MS tells you to install the software of the device, reboot a few times, plug in the device and off you go.

You happily refuse this as you feel that this isn’t needed; linux for instance does all with the device it was designed for _and_ out of the box.

So, now, MS needs to get a clue ?

I think that someone else needs a clue here.

Ludovic also doesn’t seem to know any details about this kind of stuff. The administrator needs to have one single command added into a little box under yast, that’s it. Besides, if you have such problems and call MS, I have a surprise for you. They will tell that your configuration is not supported and then hang up….. (real life example when you move an used XP disk to a new MotherBoard)

This is the admin speaking...
Eugenia wrote on July 23rd, 2006 at 8:14 AM PST:

>The administrator needs to have one single command added into a little box under yast, that’s it.

This is EXACTLY my problem. You don’t understand. When I did that bug report I ALREADY had a solution how to go about it. But the reason I did the bug report was because I REFUSE to pass parameters to kernels. I EXPECT software to JUST WORK. IF ALL OSes were FAILING in that machine, I would say “screw it, it’s indeed bad hardware”. But from the moment that kernel 2.4 works, BeOS works, Windows works and FreeBSD works, and only kernel 2.6 kernel fails and requires workarounds, then I HAVE A PROBLEM with the USABILITY in that case.

You just don’t get it. It is not about if it’s easy to add a small parameter to a kernel or not. It is about OFFERING a USER EXPERIENCE that it’s EXPECTED in year 2006.

Roeland wrote on July 24th, 2006 at 6:54 AM PST:

You ignore the fact that the reason the 2.6 kernel doesn’t bot on that specific piece of hardware, may have something to do with a thing that’s called progress. Like detecting a special APIC that can do more,; for 99 out of 100 systems, this works, just like the systems here.

The message is clear — your hardware fails because it’s being used in a 2.6 kernel. while *BSD, BeOS, windows and the 2.4 kernel doesn’t.

You in one word say that, since you don’t use your trunk, your trunk is working. Now, I look at your car and I cannot open it. I say — your trunk is broken — you say — heck, I have never had an use for it so it’s my fault that I cannot open the trunk.

My Fujitsu-Siemens Amilo D1840W boots fine without _any_ parameters. Probably FSC (uniwill bo be precise) has created hardware that doesn’t need a work-around.

If you can’t grasp the concept of understanding why something may fail, you’d better off not writing about things like SLED 10.

Ok, now, let’s go another route. Windows 2003 sucks ! I have been asking MS to fix a problem in a RAID driver. I have to manually add a driver during initial installation. Ok, I know the work around but I don’t want to use this work around, so windows 2003 sucks. Same story….. same sh*t.

To make things worse, if W2003EE is being installed it doesn’t recognize a specific RAID chipset; so you use the floppy drive to add a ‘driver’ to the OS. OK, System detects the RAID array now and I can continue. And what happens is that during installation, W2003EE states that it has a newer driver on it’s media…… how does it know ? It couldn’t install it the first place ? Right ?

There always will be some specific piece of hardware that fails and if a work-around can be used, you should use it.

The baove mentioned bug could be fixed by MS, but first we had to have a support contract of € 10k+. Ok, give me the work-around. I document it and go on with my life.

Maybe you should do too for your failing hardware. At least you received an answer and the engineer realized that a fix for you could botch other systems; like having it default passed on to the kernel, dumping specific things the kernel now cannot do because mrs Eugenia thinks that it should be incorporated in every distribution and now that they don’t, the engineer is an idiot…..

Weird, really weird.

Roeland wrote on July 24th, 2006 at 7:33 AM PST:

>MY ASS. People should stop believe such fucking hype. Most >GNU/Linuces require more special hardware and more RAM than >Windows. In its popular configurations (RH, SuSE, Ubuntu, >Mandriva) it’s a bloated piece of shit. And it shows every >fucking day.

regarding hardware — you always should pick the right hardware for your OS. It’s not hard to understand that a C64 probably won’t run XP. Still, you seem to think that any piece of hardware is specifically designed for linux, or windows. People can tell yu quite a few stories about non working hardware on both (no, all kind of) operating systems.

Regarding the bloat you are referring to, I don’t understand. You can, depending on the knowledge of computers you have install almost anything you like and need, or omit anything you like or don’t need. There are defaults, like other operating systems.

In order to have OS a) to perform the way you want to, you manually add “bloat”, like say, MSO. Something that comes in the form of OOo with other operating systems installed, yes maybe even by default.

Now. there is one point — in order to remove or not install parts, you need to know what you do. This applies to all kind of operating systems. Maybe I cannot strip down W2003EE, does that make windows bloated or does that tell a bit about the (lack of) understanding of windows ?

My ass, indeed.

This is the admin speaking...
Eugenia wrote on July 24th, 2006 at 7:35 AM PST:

>If you can’t grasp the concept of understanding why something may fail,

I don’t WANT to know the details. As a user, all I want is for this machine to work out of the box.

You mention Win2003, but you mention a problem with RAID controllers, and that’s NOT what 99.9% of the users have. YOU are NOT a normal case.

My PC is. And there are many PCs that now require nolapic to work btw, it’s not only my case. That’s why SuSE has added that in their boot manager options. Instead of asking users to figure out that they need to use nolapic to go ahead, they should go ahead and fix it.

Again, my problem is way more common than your RAID one.

Comments are closed as this blog post is now archived.

Lines, paragraphs break automatically. HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

The URI to TrackBack this blog entry is this. And here is the RSS 2.0 for comments on this post.