Well, let’s the blame where the blame is due, shall we?

1- Dell. Dell should take most of the blame. I certainly don’t mind them selling computers with Ubuntu pre-installed, but they should’ve put a HUGE warning that the buyer is not buying a Windows system, and asked the user to confirm it more than once. I’m sure Abbie bought her laptop because it was cheaper than the rest. Dell should’ve warned her.

By the way, I tried to look up Ubuntu computers on Dell’s website few months ago and I couldn’t find them in the products list. I had to specifically search for “Ubuntu” to find them. So maybe Dell fixed their part of the problem now.

2- Verizon. Regardless of Windows or Linux, I don’t understand why Verizon insists on using a CD in order to use their internet service. With my current ISP, I just plug my router and (sorry to quote Steve Jobs) it just works. If Abbie bought a netbook (or a Macbook Air or Mini for that matter), wouldn’t she face the same problem?

3- The Linux fanboys. Oh boy. Their bloody attitude is just so barabaric. They insulted her, called her dumb, harassed her, and so on, and for what? Because Dell and Verizon didn’t help her in a timely manner? AARRGGHH. [insert comment about them giving the good Linux people a terrible name].

4- Abbie. Well, it’s easy to blame the victim, but I have no idea what really happened. The news story doesn’t really give that much details about it. How did she go from “can’t connect to the internet” to “dropped out of college”? Does she not have any other way to access the Internet? How did she register in the first place? How did she connect to the Internet before buying the doomed Ubuntu laptop? I just don’t get it.

Surprisingly though, I don’t blame the news channel that published the news (even though it’s so much fun to blame the media). The two reports were just saying what happened. Perhaps the Linux fanboys were so angry because the only coverage they ever got was bad publicity, but that’s nothing unusual. No one ever sings the praises of Windows on TV either, it’s always “virus destroyed this” or “hackers stole that”.

Maybe the Linux fanboys who sent all that hate mail to the news channel should’ve instead pointed out where Linux has been quite positive, like in South America.

“Well, let’s the blame where the blame is due, shall we?”

That should’ve been “Well, let’s put the blame”

Anyway, after reading the news story on wkow.com. Wow, she apparently paid $1100 for the laptop. Now I’m even more puzzled. How did she order Ubuntu instead of Windows? No wonder Dell changed their website and hid Ubuntu.

As much as I hate OEMs that push Ubuntu to unsuspecting consumers (they do that to get better bulk OEM prices for Windows btw), I still can’t understand how “can’t connect to the internet” turned into “dropped out of college”. Same way with the “can’t open office docs” problem.

“Can’t connect to the internet” is a two-day problem. You try to connect, you fail, then the next day you pay 40 bucks to the technician who comes and fixes everything. Similar to the can’t open doc files problem: Technician comes, he tells you you need Windows to have full doc compatibility, you pay the dough required for the Windows license (and learn a leason not to buy weird unproven technology next time), and that’s it.

I hate it when users have a problem with their computers, and instead of calling some professional to fix the problem, they will wander around for months in system settings and websites trying to figure out a solution by themselves, and then blaiming the computer for the loss of productivity incured.

tl;dr: The girl was a female Wally.

PS: The funny thing with doc files is that the specification is now open (it’s even in wikipedia). In fact, it appears that doc and ppt support in OpenOffice stopped improving when microsoft opened the spec, because loons couldn’t show off their mad reverse engineering skillz…

“Well, let’s the blame where the blame is due, shall we?”

Duly obliged.

1. Ubuntu.

That’s it. No one needs Ubuntu on a laptop. No one wants Ubuntu on a laptop. Linux on a user-facing machine is plan and simply a waste. OOo? Apt/YUM? Broken drivers? Ultra-short support periods? Where do I sign up?

Oh, yeah. Linux loons want Linux everywhere. The IT press complains when Linux loons can’t have things their way. You simply can’t win.

“Maybe the Linux fanboys who sent all that hate mail to the news channel should’ve instead pointed out where Linux has been quite positive, like in South America.”

You mean when I am abjectly poor, I get to be stuck with Linux, too? Swell!

“Can’t connect to the internet” is a two-day problem. You try to connect, you fail, then the next day you pay 40 bucks to the technician who comes and fixes everything.”

There are only a few things an average ISP technician is allowed to do, and I don’t think fiddling with Ubuntu is included.

“Similar to the can’t open doc files problem: Technician comes, he tells you you need Windows to have full doc compatibility, you pay the dough required for the Windows license (and learn a leason not to buy weird unproven technology next time), and that’s it.”

Dell’s own Linux loons squad told her to stick to the damned thing, and Schubert knew jack all about what this “Ubuntu” was all about until much later on. I think by now she would be quite familiar with the concept of an “operating system”, but I still wouldn’t vouch for her ability to install one by herself.

@JoeMonco

“You simply can’t win.”

Oh how much easier life would be if the slaves didn’t revolt.

Dell is a sht. Seriously, only stupid people buy Dell online or on phone nowadays.

Guess I have more of a reason to start up Suckware.

“1. Ubuntu.
That’s it. No one needs Ubuntu on a laptop.”

I’ll give you that, no one needs it. Similarly, no one needs Macs because Windows works well. No one needs Ford because Chevrolet cars work well. In fact, no one needs Dell because HP computers work well.

——
“No one wants Ubuntu on a laptop.”

Well, I do. So unless you are denying my existence, I don’t think your point makes sense.

——
“Linux on a user-facing machine is plan and simply a waste. OOo? Apt/YUM? Broken drivers? Ultra-short support periods?”

Well, nothing in the news story mentioned these problems, at all. The two main problems were:
1- Verizon wants the user to use a CD, which isn’t compatible with Linux.
2- Microsoft Word doesn’t work on Linux. (note that they never said that they tried OOo and it was terrible or didn’t work. They simply said that Word doesn’t work on it, which apparently the college required)

Now, I totally agree that the above ARE INDEED PROBLEMS IN LINUX. However, there is nothing Canonical (or any Linux company) could do about them. Should Canonical buy Verizon? or maybe buy Microsoft?

Linux has problems that are its own doing (GNOME Shell, for instance), but the above two are external to it.

——
“Where do I sign up?”

ubuntu.com

——
“Oh, yeah. Linux loons want Linux everywhere. The IT press complains when Linux loons can’t have things their way. You simply can’t win.”

I’ll have to agree with that. Linux fanboys are a huge problem in Linux.

——
“You mean when I am abjectly poor, I get to be stuck with Linux, too? Swell!”

One of Linux’s selling points is the zero price. What’s your point?

“One of Linux’s selling points is the zero price. What’s your point?”

Yet at the same time, one of Linux’s cons is that it is free ;-)

“Well, I do. So unless you are denying my existence, I don’t think your point makes sense.”

So you are one of those who want Linux “because it’s Linux”? You know, like those want a Ford “because it’s a Ford”.

Why see things as solutions to existing problems when you can make them matters of emotional decisions?

“Well, nothing in the news story mentioned these problems, at all.”

I am pretty sure unless otherwise mentioned, people are bound to have satisfactory experience with ghastly features like those.

You are making a lot of sense here, don’t you?

“Linux has problems that are its own doing (GNOME Shell, for instance), but the above two are external to it.”

Right. Having a fragmented API is indeed a problem extrinsic to Linux. No wonder you only see it in every distribution.

@IMGX64:

_Nobody minds Dell selling computers with Ubuntu pre-installed. Except the people who bought those computers.

Which is surely the point.

@JoeMonco
“So you are one of those who want Linux “because it’s Linux”? You know, like those want a Ford “because it’s a Ford”.”

Not at all. I want Linux on my own laptop because it’s the best for me. I don’t want it on everyone’s laptops, and I certainly don’t want it on laptops of people who never asked for it (like Abbie). If they’re happy with what they have, good for them. And as long as nobody calls me to fix a problem with their computers (which would be much more frequent if they had Linux), I’m happy.

——
“Why see things as solutions to existing problems when you can make them matters of emotional decisions?”

I don’t make them emotional matters. But either way, there is a difference between using something yourself and recommending it to others. I use a lot of things in my life that I would never recommend to anyone (even when they ask “what would you choose if you were buying it for yourself?”).

——
“I am pretty sure unless otherwise mentioned, people are bound to have satisfactory experience with ghastly features like those.”

I don’t know about that. But again, the only people who should buy computers with Linux pre-installed are people who already know about it. I disapprove of giving it to unsuspecting people.

——
“Right. Having a fragmented API is indeed a problem extrinsic to Linux. No wonder you only see it in every distribution.”

Err, What does Verizon and Microsoft not supporting it have to do with fragmented API?

——
@DrLoser
“_Nobody minds Dell selling computers with Ubuntu pre-installed. Except the people who bought those computers.”

Unwittingly. You forgot to add “unwittingly”, in which case I would agree. However, there are people who intentionally want to buy a laptop with Linux pre-installed. (You know, with Linux being unpredictable about drivers, having a computer tested with Linux by the manufacturer is quite useful)

“But again, the only people who should buy computers with Linux pre-installed are people who already know about it.”

Nonsense, if you use Linux then you’ve accepted the constant breakage and problems; installing the OS yourself is nothing compared to the maintenance tax you’ll pay.

Windows users however, expect their PC to just work, and usually it does – the added strain of learning how to install an OS (Windows is an OS?! I thought it was an Office suite! – normal response, by the way) is completely unjustifiable for them, not to mention the time and effort of actually doing the install themselves, manually.

Imagine car buyers setting the engine up after purchase – no, I can’t either.

“I disapprove of giving it to unsuspecting people.”

Do you? Do you really disapprove?

“You know, with Linux being unpredictable about drivers, having a computer tested with Linux by the manufacturer is quite useful.”

Nonsense, it’ll just break after the next update (or the one after that, if the first one failed to do it).

Given that you know that, you should be working towards a stable ABI/API in Linux, if you find that you can’t do a thing about that, start thinking about alternatives.

“Not at all. I want Linux on my own laptop because it’s the best for me.”

In what way? That stuff keeps falling apart left, right and center?

Oh well, one man’s turn-off is another man fetish, perhaps.

“Err, What does Verizon and Microsoft not supporting it have to do with fragmented API?”

Not at all. Applications run on sheer magic. API is just a myth.

It’s no ordinary magic either, it’s ToeFungiMagick™ by Stallman©(R) and it’s what makes Linux work so well.

@ReverseControllerSE
“Nonsense, if you use Linux then you’ve accepted the constant breakage and problems”

I can’t see any disagreement between what I said and what you said. Yes, I accept that Linux has problems, and I’m ready to live with many of them. What’s wrong in wanting to reduce them by buying a computer that was tested with Linux instead of doing all the compatibility research myself?

——
“installing the OS yourself is nothing compared to the maintenance tax you’ll pay.”

The point is not having to install the OS, the point is it being tested with the computer.

——
“Windows users however, expect their PC to just work, and usually it does – the added strain of learning how to install an OS (Windows is an OS?! I thought it was an Office suite! – normal response, by the way) is completely unjustifiable for them, not to mention the time and effort of actually doing the install themselves, manually.

Imagine car buyers setting the engine up after purchase – no, I can’t either.”

Exactly! That’s what I just said above. I don’t want people who don’t want Linux to have it on their computers without their consent.

——
“Do you? Do you really disapprove?”

Yes I do. What makes you think otherwise?

——
“Nonsense, it’ll just break after the next update (or the one after that, if the first one failed to do it).”

And you know that because you have a time machine and tested all hardware configurations with future versions, right?

——
“Given that you know that”

Know what?

——
“you should be working towards a stable ABI/API in Linux”

Who do you mean by “you”? I’m not a kernel developer.

——
“if you find that you can’t do a thing about that, start thinking about alternatives.”

I can’t do a thing about it, yet I’m fine with Linux. Why should I start using Windows? (unless you mean something else by “alternatives”)

——
@JoeMonco
“In what way?”

I could craft a nice numbered list of all the reasons I use Linux, and why it’s better for me than Windows, but what’s the point? You’ll either disregard them as TMs, say that I don’t really need them, write a long reply about how AnythingNotNTIsCrap, completely miss my point, or just begin to argue about tiny details that don’t matter to the main point. You’ve done all that before.

You already know many reasons people use Linux, mine is a subset of them. End of the discussion.

——
“Not at all. Applications run on sheer magic. API is just a myth.”

That’s not the reason. Verizon and Microsoft don’t support Linux because it has negligible market share, not because of stable APIs.

I just wonder how the most widely used proprietary software on Linux (Flash) manages to be distributed as a single binary (libflashplayer.so) that works on pretty much every Linux distribution.

“I could craft a nice numbered list of all the reasons I use Linux, and why it’s better for me than Windows, but what’s the point?”

See? I like that. Whenever serious questions are thrown in regards to why Linux supposedly “works best” for someone, this is almost the one single answer I get every time. Coincidence? I think not. And, to be honest, if your answer can be at all condensed to a set of TMs, this site is probably _not_for you anyway, don’t you think?

“[S]ay that I don’t really need them[.]”

I use Linux in a headless box, and I know exactly the capabilities a typical distribution has to offer, how relevant they are to running a laptop, and under what circumstances things break due to in/extrinsic factors. On top of that, I also happen to know what the Windows equivalents are of the Linux software in most cases. I am not an NT lover, but simply a pragmatist who believes the best tools are the paths of the least resistance to getting the work done, and Linux hardly counts as the path of least resistance on a server – let alone a user facing device like a laptop.

But, hey, if your reason to use Linux is to have something to much around with (or, as some say, “customize”), then I have almost certainly no opinion about it. But, then again, I don’t think most of us here are hobbyists or are interested in messing around with things for the sake of the exercise. Your opinion, thus, in this case, is simply worthless to us.

“That’s not the reason. Verizon and Microsoft don’t support Linux because it has negligible market share, not because of stable APIs.”

Right. Adobe senior engineer Chris Cox once said that there would be no Linux support from Photoshop unless its market share rose above 15%:

http://forums.adobe.com/thread/487814

Of course, we are all aware of the fact that OS X market share has always been below that number, but, still, a Mac version of Photoshop exists.

Frankly, even if Linux collective market share does by any odd chance rises above the unlikely 5%, what that simply means is that you will have 6000+ different broken little OSs occupying 5% of the surface area of a pie chart. In other words, the number simply means jack all when you don’t have a consolidated target for anyone to aim at, and Linux is exactly this kind of money sink that no one wants to throw money into unless there is one or two specific areas where one or two distros stick out significantly above others.

Trust me – market shares are all about the API.

“Right. Adobe senior engineer Chris Cox once said that there would be no Linux support from Photoshop unless its market share rose above 15%:”

1. Taking words from an engineer regarding a company’s product strategy seriously.
2. Thinking that X% marketshare means X% of total photoshop users.

“1. Taking words from an engineer regarding a company’s product strategy seriously.”

No. We were saying that even if Linux market share was $a% then 6K distros would cover it and therefore not consolidate Linux as a target.

Frankly, even if Linux collective market share does by any odd chance rises above the unlikely 5%, what that simply means is that you will have 6000+ different broken little OSs occupying 5% of the surface area of a pie chart. In other words, the number simply means jack all when you don’t have a consolidated target for anyone to aim at, and Linux is exactly this kind of money sink that no one wants to throw money into unless there is one or two specific areas where one or two distros stick out significantly above others.

“2. Thinking that X% marketshare means X% of total photoshop users.”

You think we’re dumb enough to think that?

“1. Taking words from an engineer regarding a company’s product strategy seriously.
2. Thinking that X% marketshare means X% of total photoshop users.”

3. Taking Bassboy’s ability to comprehend seriously.

4. Shutting the f*ck up.

Alright then, I’m a hobbyist.

——
“I use Linux in a headless box.”

May I ask why do you use Linux? I’m genuinely curious, after all what you’ve said about it.

“May I ask why do you use Linux? I’m genuinely curious, after all what you’ve said about it.”

You want an answer? I’ll give you an answer.

The headless box is for running VMWare Server and VMWare Server alone. If I now migrate the damned thing to Windows, then I am probably better off migrating every VM to Hyper-V as well, but I am too busy for that kind of work for now. Or, I can migrate to Solaris, but VirtualBox kind of stinks, and I don’t want that.

As a result, now I am currently stuck with a host OS that is more than happy to eat up all the memory for the page cache without leaving anything for the VM instances, and a software suite that is perpetually stuck at version 2.0.2 without any obvious reason. I loathe the set up as we speak, but sorting it out is quite frankly at the top of my priorities right now.

So much for being a cheapskate…

*not at the top

“Yes, I accept that Linux has problems, and I’m ready to live with many of them.”

Very well. But why is your next response:

“The point is not having to install the OS, the point is it being tested with the computer.”

I thought you accepted the invariable failure to operate properly inherent to Linux?

Furthermore, the test is meaningless, unless a stable ABI exist – drivers will simply break at any given update (probably the very next one).

This exact thing, if I understand correctly, happened to Dell Ubuntu machines.

“And you know that because you have a time machine and tested all hardware configurations with future versions, right?”

I know that hardware vendors do not like supporting their current, much less their older, discontinued, products.

Without a stable ABI, drivers will become incompatible and hardware will no longer operate (properly).

It’s inevitable.

Meanwhile in Windows, we continue to use olden computer ware, because, you see, Windows 7 supports Vista’s (and in some cases XP’s) drivers.

“I’m not a kernel developer.”

You’re a user; you have a right to complain.

There is a reason we mock Linux with WorksForMe™ and FixItYourself™ – those cretinisms have no place in any software vendor’s vocabulary.

“I can’t do a thing about [stable ABI/API nonsense], yet I’m fine with Linux.”

You are not. Your desire for a “tested” Linux distro and “tested” hardware, shows that you clearly are not FineWithLinux™.

I think the whole stable ABI/API is blown out of proportion. First, the majority of Linux drivers are open source and live in the kernel source repository. Any ABI changes are irrelevant because distributions recompile everything every release anyway, and any API changes will easily be found because the code won’t even compile.

Now, for proprietary drivers (mostly graphics and wireless), the changes are serious problems. This is one of the biggest reasons Linux users hate proprietary drivers. In fact, the kernel developers don’t even want companies to release their drivers under the GPL, they just want hardware specifications so that they can write their own drivers. When companies do that (again, graphics and wireless adapters are the biggest problems), everyone gets happy; companies don’t have to maintain drivers, and Linux users get the drivers they want (which will rarely break).

Of course, there is the problem of fixing bugs, which will break things that depend on these bugs. IIRC, Windows specifically tries to keep buggy software working by staying compatible with previous bugs (compatibility modes). I once read an article by a Microsoft Engineer about how they spend a long time testing popular software (one example was The Sims), and programming Windows to behave differently when it encounters one of them. Wine developers have always complained about this.

So in short, Linux developers want to do TheRightThing™, Windows developers want to make things JustWorks™. The former works better for open source software, the latter works better for proprietary software.

When some software I use breaks because a bug was fixed elsewhere, I file a bug report (i.e. complain) and hope for the best. It annoys me to keep using the old version, and I also understand that most other people don’t have this patience.

——
“This exact thing, if I understand correctly, happened to Dell Ubuntu machines.”

I’m not doubting you, but can I have a citation?

Actually, I believe Canonical is working to fix this¹. Until few months ago, there were two kinds of Ubuntu-tested hardware: Ubuntu-Ready (which is only tested with a single version) and Ubuntu-Certified (which requires the manufacturer to send the computer to Canonical so that they can test it with every update). Now Ubuntu-Ready is discontinued, and so any hardware listed on the Ubuntu Certified hardware² page is supposed to continue to work across updates.

[1] http://blog.canonical.com/2011/07/18/ubuntu-certification-is-changing/
[2] http://www.ubuntu.com/certification

“You are not. Your desire for a “tested” Linux distro and “tested” hardware, shows that you clearly are not FineWithLinux™.”

Actually, I’m fine. That would only make life somewhat easier. Compare:
1- I want to buy a new laptop, I look for a suitable laptop, research every hardware part to check if Linux drivers exist for it, buy it, install Ubuntu.
2- I want to buy a new laptop, I look for one with Ubuntu pre-installed, buy it, install Ubuntu (if the pre-installed version is old).

“First, the majority of Linux drivers are open source and live in the kernel source repository. Any ABI changes are irrelevant because distributions recompile everything every release anyway, and any API changes will easily be found because the code won’t even compile.”

So what if something is not integrated with the kernel source? Integrating with the kernel source requires to both give away the source and leave the damned thing at the mercy of a non-affiliated maintainers who knows next to jack squat about the hardware. Who in the right mind wants to support their hardware in such a way?

And having the source compiling means jack all for all that matters. Nothing will work if things are messed up at the binary level. Think of the frequency API behaviors are changed both in the user space and the kernel space. Can you honestly test a driver and expect to work six months after the fact? I don’t think so.

“In fact, the kernel developers don’t even want companies to release their drivers under the GPL, they just want hardware specifications so that they can write their own drivers.”

Nice, so we can have a bunch of amateurs who knows stuff all about the silicon to produce the drivers for us. What’s next?

'Of course, there is the problem of fixing bugs, which will break things that depend on these bugs. IIRC, Windows specifically tries to keep buggy software working by staying compatible with previous bugs (compatibility modes). I once read an article by a Microsoft Engineer about how they spend a long time testing popular software (one example was The Sims), and programming Windows to behave differently when it encounters one of them. Wine developers have always complained about this.”

No, Maxis should have given the source to MS for some shoddy “peer review” and then let them compile the source into binary. And when the next Windows version is released, do the same crap all over again. Because, you know, MS should never just focus on making an OS, but should instead maintain apps for every ISV in existence, and even the whole concept is nothing more than a money sink, MS should keep doing it even if the ISVs no longer show interest in their products.

Or, better still, just give the lot of users the middle finger for ever daring to use an old piece of software on a new OS. That’s the way Linux rolls, after all.

“So in short, Linux developers want to do TheRightThing™, Windows developers want to make things JustWorks™. The former works better for open source software, the latter works better for proprietary software.”

Yes, why gives people something that actually works when you can ask them to muck around with the source. After all, that’s what producing usable software is all about – pointlessly wasting everyone’s time – isn’t it?

You see, kids – here, we loathe this kind of attitude, and we don’t care if you want to the source to be accessible to the Martians. People use software because they want to want a way to get their work done, and however “open” your source is is in every way peripheral to their interests. If your development model somehow becomes a hindrance to achieving their purpose, that doesn’t mean that your development model is just different – it means that your development model is broken.

But, hey, what’s wrong with putting the cart before the horse anyway? It’s just the Linux way™, and it’s different.

Joe Monco, I’d just like to interject for a moment, what you’re referring to as Loons, are in fact, GNU/Loons, or as I’ve recently taken to calling them, GNU plus Loons.

@IMGX64:

That Sim City thing? I believe you’re referring to http://www.joelonsoftware.com/articles/APIWar.html. (Added because you’re nice enough to supply your own references, so it seems rude not to reciprocate.)

It’s a, hem, interesting way of doing business. Fine for Microsoft, but I don’t think it scales down even as far as Apple (not that Jobs would accept such unclean behaviour), and it’d be silly to expect it of Linux.

I do think that Linux could try a little harder to make things that work for more than six months, however…

“I think the whole stable ABI/API is blown out of proportion.”

Nonsense, even MS’ employees admit they are not taking it seriously enough (now what does that tell us about Loonix cronies).

“First, the majority of Linux drivers are open source and live in the kernel source repository.”

Why do you need to have hardware “tested for Linux” then? Surely, the chances of buying any hardware that does not have an open source driver are very slim?

“Now, for proprietary drivers, the changes are serious problems.”

Oh, I see. But these are in the tiny minority you say; so, why worry?

“In fact, the kernel developers don’t even want companies to release their drivers under the GPL, they just want hardware specifications so that they can write their own drivers.”

Because AMD/ATi would very much like nVidia to know all there is to know about their hardware (and vice versa), right?

In fact wouldn’t all hardware manufacturers just love if their designs were made public?

(The answer, by the way, is NO.)

“When companies do that, everyone gets happy; companies don’t have to maintain drivers, and Linux users get the drivers they want (which will rarely break).”

By “everyone” you, of course, allude to the manufacturers’ competitors who no longer need to bother themselves with acts of industrial espionage, since all there is to know has been laid bare?

A stable ABI, of course, guaranties that nothing can break, but then how will we find out what our competition has come up with; and after all, stealing other people’s work is what GNU/cult is all about, in the name of freedom of course, oh I’m sorry GNU/freedom.

—————————————-

“I’m not doubting [that Dell Ubuntu installs broke], but can I have a citation?”

I’ve read about it on LHB, don’t really care about it myself, but if you like, you can go ask them for links, be wary of the sockpuppets though.

“Actually, I believe Canonical is working to fix this.”

How can they, without forking and maintaining their own kernel and all the other core components (well, and everything else actually).

“2- I want to buy a new laptop, I look for one with Ubuntu pre-installed, buy it, install Ubuntu (if the pre-installed version is old).”

The old version will probably break with updates.

If you install a new version, you have no guarantee that it works – Canonical has no control over the components they put together, they can’t fix anything.

“So in short, Linux developers want to do TheRightThing™“

No, they are trying to force everyone to release internal information (useful only as a replacement for industrial espionage – trade secrets, by the way, are the “magic” that keeps you ahead of the competition).

For this ideological cause (it’s really just communism, through different means) they are willing to sacrifice users’ ability to work – this is not a problem within the cult – it is however, an insanely huge issue for anyone who actually wants to get work done.

“Windows developers want to make things JustWorks™”

Which is the only way any platform designers are supposed to work, the platform supports the products that use it, not the other way around (it’s crazy to believe software products would change and shift to follow the OS).

“ The former works better for open source software,...”

Except, that it doesn’t work at all.

“... the latter works better for proprietary software.”

For any and all platforms actually, closed or open source, public domain or proprietary (even GNU/proprietary copyleft platforms).

“First, the majority of Linux drivers are open source and live in the kernel source repository.”

It’s a self-selection process, isn’t it? If you’re non-FOSS, you ain’t gonna be in the repository. This might be because non-FOSS drivers are unhealthy in the sight of the Lord … or it might even be that, what with that ABI/API silliness, there’s no point anyway. They will break in a matter of days. You’re better than that, IMGX.

And incidentally, if you can point us towards a single, solitary, Linux kernel developer who is able to take hardware specifications from any manufacturer you care to name, and produce a device driver that “just doesn’t break,” then we’ll concede your point.

We were wrong all along. What you need is continuous flux and the magic of configure make make install (OK, drivers are slightly different), and then you can absolutely ignore the concept of ABI/API consistency.

Which was, btw, Joe M’s point all along. We had a hilarious conversation, back in the day, with a Loon over on LHB who wanted to get around ABI/API inconsistency by, er, forking off another version of the OS.

Sometimes I lie on a grassy knoll and wonder precisely what happened to that.

Forking would actually work, but then, of course, you’d have to support the whole monstrosity from kernel upwards – no one has the means to do that.

——————————-

I miss the good days we used to have over at LHB – taking Durden nonsense apart… it was always a good read – I had a blast.

Notice how OMG!Ubuntu classified the story as 'hot air’ until they could put a sensible blame on someone so that 'Ubuntu isn’t incompetent’.

““Can’t connect to the internet” is a two-day problem. You try to connect, you fail, then the next day you pay 40 bucks to the technician who comes and fixes everything. Similar to the can’t open doc files problem: Technician comes, he tells you you need Windows to have full doc compatibility, you pay the dough required for the Windows license (and learn a leason not to buy weird unproven technology next time), and that’s it.”

Because she wanted the cheapest possible and brought a 'bread and butter’ machine. Paying I don’t know 50-100$ does not help here.

“because loons couldn’t show off their mad reverse engineering skillz…”

Bitches don’t know bout my pirated HexRays IDA.

Also, the loons don’t develop OOo, it’s still all Sun with a little Novell. And Sun massively scaled down their input since about 2005 reducing the team to I think 1/5 of its original size. That’s why.

“One of Linux’s selling points is the zero price.”

Windows OEM licenses are so cheap, you can’t really argue with that point.

@ReverseControllerSE

Frankly, I’m having trouble following your arguments. Either I’m too dim to understand them, or they don’t make any sense. When JoeMonaco or DrLoser say something, I can generally understand their premises¹ and how they reached their conclusions. But you, you completely baffle me.

But anyway,

—-
“Because AMD/ATi would very much like nVidia to know all there is to know about their hardware (and vice versa), right?”

No. The only thing required is the documentation that says how to make the hardware do its job, not how it works internally. It’s analogous to APIs in software. Just Because Microsoft releases the API of Windows/.NET/etc, doesn’t mean anyone can just make a clone of it easily, or know what it does under the hood.

In retrospect, I think I should’ve used the word “documentation” instead of “specifications”.

——
@DrLoser
“It’s a self-selection process, isn’t it? If you’re non-FOSS, you ain’t gonna be in the repository. This might be because non-FOSS drivers are unhealthy in the sight of the Lord … or it might even be that, what with that ABI/API silliness, there’s no point anyway. They will break in a matter of days. You’re better than that, IMGX.”

You’re probably right about the self-selection, but the real question is, what harm to the hardware manufacturer will happen if a FOSS driver existed? (possibly not even written by the manufacturer itself)

——
“And incidentally, if you can point us towards a single, solitary, Linux kernel developer who is able to take hardware specifications from any manufacturer you care to name, and produce a device driver that “just doesn’t break,” then we’ll concede your point.”

Kernel developers aren’t that magical. A developer (or a group of developers) who has experience in the type of hardware (say, Ethernet cards) can write a driver for another Ethernet card, given that he has good documentation of the card.

I don’t know of someone in person, but here’s an example, taken from OpenBSD developers’s criticism of Linux developers²:

“A particularly nasty player in this regard has been the Linux vendors and some Linux developers, who have played along with an American corporate model of requiring NDAs for chip documentation. This has effectively put Linux into the club with Microsoft, but has left all the other operating system communities — and their developers — with much less available clout for requesting documentation. In a more fair world, the Linux vendors would work with us, and the device driver support in all free operating systems would be fantastic by now.”

So yes, it has happened before. The manufacturer lets the developers read the documentation, and they write the drivers. Of course, the OpenBSD object on NDAs, but I digress.

By the way, the whole “Puffy Baba and the 40 Vendors” is interesting, if you’re in the mood to read it.

[1] http://en.wikipedia.org/wiki/Premise
[2] http://openbsd.org/lyrics.html#41

What harm to the manufacturer? Try what benefit to the manufacturer. It’s a vanishingly small market, and the only likely result is that you will end up with a PR nightmare.

Because, as usual, any problem would not be the fault of Linux; it would be the fault of the hardware manufacturer.

So, at MegaCorp I offer out of the goodness of my heart to build and support a driver for Linux (I’d have to be mad, but that’s a different issue). And the kernel guys turn round and say, “We know better than you! We’ll write it! Just give us the specs/API!”

Inevitably the whole thing will go tits up. If it’s an important bit of hardware, such as, say, a traffic light controller, then you, as the hardware manufacturer, are now perceived to have injured or even killed people — through a process over which you were refused any control whatsoever.

Tell me. Which part of “the real world” do you fail to understand?

One more thought: if I recall correctly (and apologies if I am confusing you with somebody else), you are of the opinion that unstable APIs and ABIs simply do not matter.

Well, that’s peachy for the kernel guys.

Here I am as somebody who actually depends on saleable hardware to make my money, and I suddenly realise that I have to change my API. Now, if driver development was in-house, I could do that.

With the Linux kernel? Not so much. Double standards all over the place. Which is ridiculous, because in the real world, people who make stuff dictate the terms to fluff-heads like the Linux kernel developers; not the other way around.

“Because, as usual, any problem would not be the fault of Linux; it would be the fault of the hardware manufacturer.”

How is that? If hardware X works well with Windows but not with Linux, Linux will be blamed. In fact, this is what happens all the time.

——
“If it’s an important bit of hardware, such as, say, a traffic light controller”

Safety critical systems are in a completely different league. I doubt there are any drivers for super-special hardware like traffic light controllers in the kernel. And I highly doubt any sane kernel developer will accept taking responsibility for it.

Yet, Linux is reported to be widely used in embedded systems, including safety critical systems. shrug What were they thinking?

——
“Tell me. Which part of “the real world” do you fail to understand?”

Most of it, actually. But I do understand the “commodity hardware that can be bought by any person and runs well on Windows” part, and having binary-only drivers for that hardware doesn’t make sense to me.

——
“One more thought: if I recall correctly (and apologies if I am confusing you with somebody else), you are of the opinion that unstable APIs and ABIs simply do not matter.”

It all depends on context. Stable ABI/API of course matter and very useful. However, what price are you ready to pay for the stability? Even kernel developers don’t change things for no reason. They only change it when they have a compelling reason.

“Here I am as somebody who actually depends on saleable hardware to make my money, and I suddenly realise that I have to change my API. Now, if driver development was in-house, I could do that.”

Wait, what? How are you going to change the hardware you already sold? Once you sell a model, it will continue to work as you sold it. If you change how the hardware works such that the old driver doesn’t work anymore, then it’s a different model, and you should tell people who are going to buy it that it’s a different model. You can’t sell two different models under the same name, unless you deliberately want people to be confused.

@IMGX64:

I kind of rushed in to the API thing, didn’t I? You’re absolutely correct.

Let me try to recall my thinking (on top of a lot of beer).

A stable API works both ways. The hardware manufacturer (hopefully) provides one, and the systems guys (hopefully) provide one back, eg the standard Linux driver model (which is admirably simple, if rather byte-orientated).

So, it’s a two way process. The Linux kernel doesn’t accept the concept of a stable API, because they should be able to rewrite the driver.

In extreme (and obviously it would be extreme) circumstances, the hardware guys might make a change in their API.

This is not necessarily a problem, if you do driver development in-house. You talk to somebody sane in the OS department and you come to a reasonable compromise. In Windows, it would probably involve a short-term compromise with a user-space driver API, and a long-term fix.

In Linux … well, I doubt the conversation would ever happen. Who, precisely, would you talk to?

In other words, your API is fixed. Their API is randomly mutable, as far as you are concerned.

This is not a good way to attract partners.

(But you were indeed correct, and I apologise for my earlier babbling.)

“The only thing required is the documentation that says how to make the hardware do its job, not how it works internally.”

And the difference is?

“It’s analogous to APIs in software.”

No.

You are mixing public and private interfaces – public are the ones that will be used by everyone (GUI’s, API’s, ABI’s), private interfaces are the ones that are used internally (let’s say how a graphics chip communicates with on board memory, or how a vendor’s driver talks to vendor’s HW).

The public interface of an OS is the “kernel to drivers” interface – it must be public if the companies making drivers are to make them and it must be standard and stable.

The “hardware to drivers” interface is private, since only those who made the hardware really know how to make drivers for it, nobody else (well, only few others) can make them (if quality is a concern), nobody else cares about them – and this interface is neither standard nor stable – it changes as HW manufacturer finds errors within the HW and corrects them at the driver level (they either bypass or change the way HW works) or enables/disables functionality (yes, they do that).

“Just Because Microsoft releases the API of Windows/.NET/etc, doesn’t mean anyone can just make a clone of it easily, or know what it does under the hood.”

Well, actually they sort of can – that’s why ReactOS and Wine can, in fact, exist. They suck, because it’s not at all easy to copy Windows APIs, but they do exist. Microsoft doesn’t react because everyone involved is poor and unsuccessful.

They has no choice though, the interfaces must be public, no APIs means no programs, no programs means no reason to use the Windows platform.

——————————

“but the real question is, what harm to the hardware manufacturer will happen if a FOSS driver existed?”

Either the driver only calls (supports) a very limited spectre of hardware capabilities, the hardware itself is fairly generic (example: USB controller) or the manufacturer no longer cares about it (i.e. as old as the earth itself).

Anything else would mean a potential disclosure of information on hardware design.

——————————

“So yes, it has happened before. The manufacturer lets the developers read the documentation, and they write the drivers.”

Yeah, but those drivers are not FLOSS are they?

Key information is not available to the public – not open and not free.

The statement you quoted:
“In a more fair world, the Linux vendors would work with us, and the device driver support in all free operating systems would be fantastic by now.”

Is patently absurd – it takes years of refining graphic cards drivers before they work properly (in the first couple of months after release they are near unworkable – I’m speaking in Windows quality terms here, nothing in the world of Durden is workable by them), no FLOSS hacker has the motivation to keep going.

Either the hardware manufacturers have the necessary financial motivation to keep up or you’re not going to have good drivers; after the time, when motivation was present, passes, only a stable ABI can enable users to continue, well, using HW they bought.

Let’s say the OS is no longer used by enough users to warrant active support (Linux never is) by the HW maker, but is still supported by its devs: if it has a stable ABI, drivers (hardware) will continue to work, if not, well then you’re f*cked, same for software.

——————————

“If hardware X works well with Windows but not with Linux, Linux will be blamed. In fact, this is what happens all the time.”

Except that it doesn’t – HW manufacturers and OEM’s are blamed all the time, Linux is considered to be the victim of their conspiracies.

——————————

“Yet, Linux is reported to be widely used in embedded systems, including safety critical systems.”

Do such reports actually substantiate their claims?

Would be interesting to know how many changes were necessary to make the Durden kernel work appropriately (it would also be interesting if LY could still recognise their “miracle”), that is, if the version used is even released to the public.

——————————

“...having binary-only drivers for that hardware doesn’t make sense to me.”

Can you reprogram said drivers (in a reliable and useful manner)?

No?

What use would the source (or specs) be to you?

Do you really believe the Zealots of Durden can reprogram them? Reliably? Come now, hasn’t the history of Durden shown us all how it goes – these guy are completely untrustworthy.

——————————

“Even kernel developers don’t change things for no reason. They only change it when they have a compelling reason.”

Unstable API’s and ABI’s are there to make it painful for any non FLOSS vendor to release software for the Durden OS – this is well known, it’s a scheme to force GNU down everybody’s throat.

There was a link to a video where Mark Russinovich talks (mentions, really) about how the First Zealot of Durden (Tori of Valds) actually changed the kernel only to sabotage Mark’s program, you’ll find it at the LHB.

“If you change how the hardware works such that the old driver doesn’t work anymore, then it’s a different model, and you should tell people who are going to buy it that it’s a different model.”

But how are you going to sell hardware which has no drivers, oh right, you’re going to beg the Zealots of Durden to create one for you, and then you’ll take what they give you, right?

Quality, reliability and usability be damned.

Remember: if you create your own, they are simply going to break interfaces at any given time the idiots decide to change something (and they will do that soon enough, or, possibly, merely to screw with you), then it will be your fault for not simply letting the “community” take care of it (as if it could).

————————————

Doctor, by the way, was talking about the perspective of MS – they cannot change ABI’s since they do not control the driver development, while the LY can do whatever the f*ck they want, whilst providing exactly nothing (hence his complaint regarding the double standard).

And I see he posted before me and he meant the HW guys, well, the point of double standard is what matters.