Slashdot Mirror
Linux and the New Computing Order
Chris Siegler writes "An Op-Ed over at Dr. Dobb's, on whether Linux can change to meet the needs of the mainstream user, and the consequences if we don't.
" Interesting piece-talks about the potential for fragmentation, and that all of the big name companies coming to play are probably the greatest potential cause of that. I'm not incredibly worried about fragmentation, but more about how things change can alter the community. What do you folks think?
Personally, all of my hardware is hand picked to run Linux, but what about the new user who buys a PC, then hears about how much better this "Linux" OS is over Windows, goes out and pays $50 for Red Hat, and finds that he can't get his USB mouse to work, his PNP sound card etc? Should Red Hat put a message on the screen that says:
"Screw you buddy! You bought hardware to run Windows? So run your Windows and shut up!"?
Among all criticism directed toward Solaris I never heard anything about its inability to run on HP-PA -- if you want HP-PA, you buy HP-UX, and if you want Solaris, you buy Sparc.That's a totally different argument. Solaris doesn't claim to run on HP-PA, but Linux DOES claim to run on Intel PCs
Of all the comments I've ever posted, this is definately one of them
I think you're absolutely right about what the average computer user wants. I don't know if Linux CAN deliver the experience that they desire, though I see no reason why not - it's really just a matter of packaging and pretty (software) wrappers.
:-) products. For instance, my wife does just about all here work on her computer at home. She is a pure user, without a trace of geek (though maybe a little wanabe). Mostly she uses Filmaker Pro to keep track a her client's business for him, but she also does some web page creation/editting using FrontPage. Before her recent upgrade to FP 2000, she experience continual flakey behaviour, forcing her to repeatedly reload FP. Just before the upgrade the only way she could open a page to edit was by first opening a page she hadn't touched in several months and then opening the one she wanted to edit. Things were looking grim, as she was running out of old, untouched pages. Even now, when she alters a gif, she has to go through a ritual dance of saving the gif, reloading and refreshing pages, closing and re-opening applications to see whether a simple change has had the desired effect.
Where I really can't agree with you is the suggestion that low (computer)knowledge users get the experience that they desire from MS (gee, I've never noticed before that their initials suggest a nasty disease
I assure you she does not enjoy the windows "experience" and totally fails to understand why software companies think they can release non-working programs then charge a fortune for "upgrades" with "features" that she has no use for and that rarely work any better than the program that she wants to use, if only it would work.
Microsoft knows what the average user wants and slickly promises him/her just that, but they *don't* deliver, which is why the chairman of GM can make pointed jokes about MS software reliability. Lack of performance, whether speed or reliable behaviour, is also a significant nuisance to an ordinary user and if they don't wise up soon, this is what will really collapse the MS house of cards.
On the whole, I liked this article. I just have a few things to comment on.
Now, I don't consider myself an elite hacker or anything, (I do write and compile a lot of software though) but I haven't noticed any compatibility problems on either of my machines. But maybe that's because, like most of the world, I don't try to be bleeding edge. Neither will the mainstream. Companies like Red Hat are terribly concerned with compatibility issues, and because their money is on the line, they make sure that compatibility will be a matter of upgrading x number of packages. The cooperative and open nature of the open source community makes the turnaround time on fixing broken stuff much shorter, we all know that. But why? Because if someone with a large stake in Linux needs compatibility with new libraries or compilers, they can easily
- fix it themselves
- use their resources to aid the developers in charge of the project(s) in bringing it up to date
History seems to show that Linux companies usually choose the second option, since that is the fastest and most efficient way to go.This is the new Coke dilemma. Sorry, I just had to say that. Seriously though, this does happen. It's the problem best exemplified by EGCS. EGCS was a split from the main development track of GCC, but eventually it proved to be more promising than its predecessor and it was brought back into the fold. Our community is better for it.
To a point. The Linux community, through projects like GNOME and KDE, has made the important changes that the mainstream will want. But as a partisan, I have to say this: Let's not forget our Free Software roots. It's terribly important that we focus on the ideals that got us this far, like unabashed source availability and respect for the owners of open source projects (go here if any of this is new to you). These are the ideas that have protected and nurtured the movement this far, and they will continue to do so in the future.
-k
... I'm afraid.
As more and more organizations come to depend upon the platform they will be less and less willing to accept churn such as the libc cutover.
Someone will have to concentrate upon bugfixing, consolidation, standards compliance, documentation, automated regression testing, etc, etc. This isn't glamorous but there's a market need and it must happen.
So the fragmentation will be in versioning: people who rely upon the platform to support mission critical services and commercial software will be running two or three year old distributions, while the hackers will be running bleeding-edge stuff on their desktops.
This intertia will upset the kernel developers and kernel development will become less glamorous. The focus of innovation will move even further toward end-user applications.
This is all good.
I'm sure I'll get flamed to hell on this. But this is what I think.
.02.
I don't think that Linux CAN meet the needs of the mainstream users. Most end users when they get their PC don't want to have to have to learn commands, figure out short cuts. Then want to take it out of the box, plug it in and have it run. This is why windows has become the high focus OS that it is today, you don't have to learn commands. You just point, click, and the application runs, you don't have to compile applications and then compile it. I use UNIX in my daily life, on my home computers and at my place of work. But my use of UNIX didn't come in one day, with windows you can use it from day one no extra commands to learn, nothing to do really. UNIX is clearly the more stable OS, and for a server nothing is better. But you actually have to work to learn UNIX, its really not hard but you have to learn multiple commands and figure out HOW to use things at first if you don't know. Sure there is man pages and HOWTO's but end users with a new PC don't want to read all that, they want it to work right then with no waiting on how to figure out how to use it.
I'm sure you script kiddies out there will be all over me for this, but I hope the users of Slashdot with a brain can figure out where I'm coming from. I will probably get stuff like "WINDOWS SUCKS!! USE LINUX!! YOUR OS SUCKS!!". But come on guys, read what I had to say, and think about it for a minute.
Thats just my
In Linux, we have a user-friendly front end, but we don't require it.
Linux needs two front ends. One is the simplistic interface for someone who just wants to check their email and write memos. The other one is the current X-based "experts only" mode.
So long as the OS under the hood is rock-solid, the end user doesn't care. Having that second interface gives you more control (thus more power) at the cost of needing more mental effort. Drop that complex interface, and you lose the hackers. But here, you can have your cake and eat it, too.
--The basis of all love is respect
You left out WinCE :)
I've written a lot of code for Windows, though, and I think you make a good point with regard to fragmentation. Generally, the APIs are the same and 9 times out of 10, you can just call the Win32 function without worrying about it.
The bite comes on that 10th time. For example, the Windows Remote Authentication Server API (RAS) has significant differences between:
Windows 95
Windows 98 (and Windows 95 OEM Service Release 2)
Windows NT
It's a little tense when you have to explain to a executive why your nifty RAS program does stuff on your NT workstation that it can't do on your customers' Win95 machines. "It's all Windows, isn't it?" No, it's not.
FWIW, I also recall one case where I used some Microsoft example code that obtained a handle to a device context. The example declared the value as an int and everything ran fine on the Win95 machines. It crashed horribly on WinNT, though. A little debugging revealed that the value on WinNT was greater than an int could hold. I declared it long and everything worked ok. What peeved me was that this little quirk was not mentioned anywhere in the documentation. As far as the documentation was concerned, the function calls were the same between Win95 and WinNT.
Other people who program Windows will justifiably call me a wimp for whining about having to get the OS version sometimes. That's fair. I also shouldn't be too hard on MS for errors in documentation that is, most of the time, pretty helpful. Finally, a couple of bad experiences from one developer is not grounds for condemning an entire operating system.
But I think it's also important to point out that the Win32 API is not the same everywhere. If having to check my OS version before a function call is not "fragmentation", then what is?
(Sorry for the length of this post)
Save the whales. Feed the hungry. Free the mallocs.
...will not come to pass until it is conventional wisdom that Linux will inevitably win. For MS to release their own version of Linux would be perceived as their conceding the ultimate victory of Linux over Windows. They will not participate in the destruction of Windows -- until there is nothing left to save.
Geeky modern art T-shirts
>Just as Windows uses arrays of Dialog Boxes to configure everything, Linux would need the same. This is not to say that is a bad thing though!
Maybe, maybe not. I had an ugly experience last night with the dreaded "Windows Dialog Boxes" last night. My wife uses Windows (what can I say? I've tried) and I have upgraded her machine by building a new AMD K6-2 box and them moving everything over from her Pentium. I used drive copy to clone her hard drive so I could set up the new box with a new drive while she still had her machine running.
So everything was going great until I was ready to move the NIC. ARGH. Plug and Play and install dialogs are broken horribly. God I just wanted to find the config file and massage it by hand to get the bugger installed. I had to go through all kinds of gyrations to convince windows to take the NIC.
If Linux had dialog boxes and still had the underlying open configuration mechanisms, the would obviously be fine. It just makes me shudder to think of Linux having ANYTHING like Window's dialog box interfaces!
Bleah!
Pnp was "being addressed" for four years now.
Parallel port scanners - I believe some are supported; if yours isn't, write a driver.Oooh I was waiting for this! I work full time and have a family-- I don't have the time to write and maintain a driver. My wife gets upset when I spend too much time using the computer as it is.
Anyway is that the answer we want to give the newbie who barely knows how to install Linux, and has probably never coded? Write a driver?
Of all the comments I've ever posted, this is definately one of them
There seems to be an assumption in the computer business that there has to be one winner. It just isn't so. Despite that some people (typically the PHB types) think it is desireable to not have any choice (because they fear making decisions), it is not. Choice is good. Markets can't be viable in the long run without competition. Any time that one vendor controls more than about 40% of a market, the whole market suffers.
For me, Linux has already won. It works, it works well. I can work happily without going to something else. I don't begrudge *BSD, MacOS or BeOS or for that matter anything other than Microsoft from some market share. I'd be happy if the *BSD's saw increased popularity, MacOS was resurgeant and BeOS carved out a viable niche for itself. I don't even know if it is desireable for Microsoft to be completely wiped out (although occasionally that desire pops into my head). I'd be happy if they were beat down under 50% of the market so there were mainstream viable choices.
A weak Windows is our friend, do you think Linux would have gotten as far as it has if Windows worked as well as advertised?
Of all the comments I've ever posted, this is definately one of them
I believe that the success of Linux means the doom of Microsoft, or at least of Microsoft as we know it. The success of Linux would force Microsoft to play fair, but that is not their core competency. Either they will make it a core competency right quick, or they will die.
Either way, the customer wins.
Apple had it wrong. They used to think that, for Apple to win, Microsoft had to lose; they were correct. They were correct because, for Apple (or Linux) to win, the customer pretty much has to win. And for Microsoft to win, the customer has to lose.
--The basis of all love is respect
Like I said my mom installed Win95. And the printer. The network connection someone helped her with (its her work computer). No she couldn't do everything from day 1. She still can't. But she knows to hit f1 for help. My mo can't remember the "del" command from dos. She does remember to hit the del key in file manager to delete file.
As for the computer crashing - it happens extremely rarely. Once a year maybe. When it happens its usually something I can say "Well yeah hitting ctrl-alt-del does reboot the computer" to. Occasionaly there is a serious problem. But she knows to save often. So she's never lost more than 15 min of work.
As for reinstalling everything - ummmmm. I think that happened when she got the new computer. About 3 years ago. Since then.... No reinstalls. And she would've had to install Linux and everything with the new computer too.
-cpd
This will certainly happen. Creative has released Soundblaster Live drivers, for instance
But wait till you see the community scream when these vendors don't release source code, in truth many won't be able to, even if they wanted to because of patent problems and what not.
Of all the comments I've ever posted, this is definately one of them
I think that you may not quite understand, or that you're deliberately distorting things to force an issue.
Those people with unsupported hardware either need to write their own drivers or ask others to help out. In this community people frequently try to fix their own personal problems, expecting that their work can be passed on to a silent majority of people with the same problem but no means of correcting it.
I'm sorry that you don't have time to work on the drivers you need, but it looks like you've got better things to do, and that's okay.
What's not okay is for you to bitch at other people, telling them that they need to scratch your itch before their own.
If the community spent all of its time satisfying the needs of newbies, nothing significant would ever get done.
The answer we ought to give to newbies is that they can rely on the kindness of strangers. But that that path is a bumpy one, since strangers don't have to be kind. It's better then for a newbie to actually learn something, shed their newbie status and solve their own problems. Who really likes to be helpless? Is it not better to become more self-reliant?
There are a lot of newbies nowadays, and I think that we'd be better off if we made it easier to start down the road towards becoming an experienced hand, rather than to cater to their needs. After all, most newbies are helpless; they'll always have the same needs, and we'll all be bogged down in them. To constantly let newbies remain newbies is analagous to not teaching your kids to read, just because they don't want to.
-- This and all my posts are in the public domain. I am a lawyer. I am not your lawyer, and this is not legal advice.
before their own.
To clarify my post, All of my own hardware works under Linux, but that's because I built my own PC and only chose components that are known to work with Linux. I am not whining for drivers for my own needs
Most people would rather buy a PC off the shelf of a Superstore, and are often stuck with whatever hardware their PC came with. If they want to run Linux they may be out of luck.
I know that the Open Source community scratches their own itches first, this is understandable, it's human nature. However, there are people out there, notably ESR, who make Open Source out to be something that will solve any problem.
If the community spent all of its time satisfying the needs of newbies, nothing significant would ever get done.
Agreed, but then we need to stop pretending that Free Software/Open Source has all the answers. The reality is, if the Linux/OS community doesn't care enough provide something, companies may see the need and provide proprietary solutions, which will make the end user happy, but infuriate the Linux/OS community.
Of all the comments I've ever posted, this is definately one of them
Ooooo, perfect opportunity for my story! My roommate decided to install Linux. This made Windows uphappy. So he moved everything over to that partition and tried to reinstall Windows. He therefore had the opportunity to install both within 24 hours of each other. The Windows install is easy because it makes too many assumptions. For instance, it assumes you want one partition (duh, we all knew that). Then it assumes you have a standard VGA video card, no sound card, and don't want to dual-boot. Practically, it makes sense to say that Linux should not assume all these things, since it's usually not installed on its own, but idealistically it should be equal with Windows. My friend was astounded that it didn't even bother to ask about his video card, so he went to the control panel, clicked on display, and then it had a kernel error and crashed! With nothing installed at all. It crashed. What kind of an installation is that?
Yes it will take awhile to write the drivers. But USB looks like it will be a big part of the future of PCs and Macs, so if Linux is serious about these platforms, it cannot be ignored.
USB has been around a couple of years, yet Linux support is just starting to materialize. NetBSD has had USB support for a year or more.
Of all the comments I've ever posted, this is definately one of them
Why is it the press must focus on overthrowing Microsoft in the mainstream market? Why does everybody insist that Microsoft is the evil empire and that Linux is their tool to defeat it?
I use Linux because if I don't like it, I can change it. That ability will never go away. The whole concept of copyleft has changed all of the rules.
Microsoft is NOT the enemy and it never has been. Linux will only die if the developers lose their love for it and leave. It might lose market share or gain it. It really doesn't matter to anybody except the marketroids. Linux will always be there and most of use don't give a damn what percentage of computers run it.
--- A Jesus Fish eating a Darwin Fish only proves Darwin's point.
Agreed, but then we need to stop pretending that Free Software/Open Source has all the answers. The reality is, if the Linux/OS community doesn't care enough provide something, companies may see the need and provide proprietary solutions, which will make the end user happy, but infuriate the Linux/OS community.
Well, like I had said, I think that the best thing to do is to encourage people to want to learn about things, including the details on their hardware, and how to support it if it isn't already. I don't think that OS is necessarily going to provide all of the solutions, although certainly the more people that really use it, the better it'll work.
As for commercial software providing solutions, I don't really think that this is as likely as you do. Most companies are impatient. Rather than go for steady, modest profits, they frequently go after big markets and big profits. Think of the difference between a small software house and Microsoft. Any niche big enough to support a generic, impatient company is likely to have a group of OS developers already. The tricky bit, we agree, is getting in between the cracks, and not very many companies are going to want to do that.
Too many people want to Make Money Fast, and since a lot of them are in management, they can frequently give orders to programmers to ignore the vital cracks and go for the high profile stuff.
-- This and all my posts are in the public domain. I am a lawyer. I am not your lawyer, and this is not legal advice.
The article raises some good points. Some Linux developers get all up in arms with the mention of things like "binary-only kernel module". Yet the Linux community has so far not delivered on many of the things it needs.
Some examples:
USB? If you're lucky your mouse will work.
PNP? Linux makes this more difficult than non-PNP
Parallel Port Scanners? Forget about it..
3D? 3D works great... IF you have the right card
But if you need an MP3 player, you're in luck! Freshmeat lists 77 entries under Mp3.
That illustrates what tends to be the problem with the Open Source community, they'll produce the stuff that they personally want, but other things tend to be ignored.
Don't get me wrong, I love Linux, and use it for almost everything. I do get frustrated at times because I need to reboot into that other OS to accomplish something.
Of all the comments I've ever posted, this is definately one of them
One that supports links, isn't plagued by the archaic and irritating 'drive letter' scheme and that doesn't do funky things with mixed case filenames.
NTFS is saddled with major annoyances in order to make it be able to coexist with regular Windows (and even MS-DOS) applications.
You have made a good point but you have overlooked a couple of things. Windows users will gradually grow more sophisticated over time. The strategy of pretending that administrative issues don't exist so that you need only be a "user" of your machine will grow weaker over time as Windows users gain experience. Windows only appears to be easy to use and administration free because of several factors including this naivete and, of course, pre-installation. As users become more knowlegable and as other systems are offered preinstalled that boot straight to GUI's Windows will lose this advantage.
Geeky modern art T-shirts
This begs the question: what is it to win, and should linux strive to win at it?
The implicit assumption is that linux will become a mainstream, Joe Lunchbox operating system, to the detriment of Microsoft (Apple, Be, etc).
I really don't see the advantage in that. It'll certainly up the demand for commercial applications, and corporations will move to fill that demand. Is that what we really want?
So many linux users (both way back when and now) couldn't give a rat's ass for free software; they want applications. I'm afraid that if we court the commercial market too strongly, we'll lose, not gain, developers - that great mass of developers who create free software.
What linux should strive to gain is more developers and especially more free software developers. Commercial ventures seeking to distribute non-free software for linux should be given a run for their money; they should be pressed to advance the state of the art as fast as they can. The point is that we, as a community, don't need more users; we need more developers. Linux users should aspire to become developers, and existing developers should aspire to use their skills on Linux. If we slowly give that up and let corporations handle it for us, then we're also giving up control of Linux, GPL notwithstanding.
--
--
There is no premature anti-fascism. -Ernest Hemingway
But who set it all up? I have seen my mother install win95. And M$Office. Yeah she just hit the typical install thing. But it was simple.
I installed RH5.2 about 6 mos ago. I had no problem. I installed KDE and Gnome. Fairly easy. Thing is I know basic unix stuff. My mother could not have done it.
With linux you still need to know commands. With Windows you don't. Although you should. I'm sure my mother would be uncomfortable in linux, and therefore would not use it. Yes KDE and Gnome are easy to use. But you still have to know stuff. Most people already know Windows.
Why would they learn something else?
Cuz it's more stable? - nah. Mom's computer rarely goes down and then its an exscuse for coffee.
Freedom(speech) of Code? - nah. she could care less.
Free($$$) of code? - That she might bite at. But she'll probably pay for something familiar. How much will you pay for home cooking like mom's?
Cuz she can run a server? - Mom's response - Whats that?
I still argue that linux's main place is in the server arena, in the hands of computer people, and used by people willing to learn. The masses tend not to want to learn.
-cpd
I agree with your points, but not your conclusion. (just realized I was replying to the reply, this actually goes to the original post).
Never say never.
The points you bring up are certainly going to make it much harder for Linux, or any *nix-like OS, to become a 0-skill, entry-level, first day OS. Look at DOS and Windows. Same old DOS still sits under W95 and is still just as ugly as ever. The new user is shielded from that.
To do the same to Linux is probably not impossible, though not very desirable. My opinion is that Lunix is moving in the right direction to make it more accessible for the novice enthusiast, but will/should probably stop short of the "mainstream" user who has no computer skills and does not want/need them to do their job.
I'm a big fan of choice. I'm not eager to see Mickeysoft wiped out completely, just put in their place. I can see a world where Linux is a powerful, well entrenched, well supported platform for whoever wants to use it for whatever use provided they have the knowledge to put it to work. And then alternate, less powerful, less entrenched equally well supported platforms are available for those whose skill sets are outside of computing who want to have an easy to use tool to make their work easier.
My $.0199967352
Fragmentation IS a concern, and it may already be happening. In the article on here yesterday or Monday about SGI's new Linux servers, I pointed out that they specifically mentioned two specific changes to the kernel -- a replacement TCP/IP stack and more efficient NFS code.
I asked if anyone had seen a mention of these on the kernel mailing list. There was no replies suggesting that there has been any contribution by SGI to the stock kernel of those technologies.
So there's a bit of fragmentation there. What happens if software starts needing a particular NFS implementation, or a particular TCP/IP implementation. Already you see packages coming out for RedHat that are a bitch to install on non RedHat systems. Hell, IBM's viavoice stuff they released a while back bombed out on non-RedHat 6.0 systems.
As time progresses this is only going to get worse. The work being done to define a standard base of libraries doesn't enforce application vendors to build against only those libraries -- particularly since (AFAIK) the compilers usually link against the most recent ones on the system, and most if not all distributions will have more than just those base libraries on them.
Under Windows you occasionally see that problem too, with differing versions of DLL's getting mixed up, overwritten, or lost and the general instability that results. What happens when more applications start coming out requiring versions of the system libraries that aren't necessarily compatible with each other? Are we going to end up with 100 meg of differing libc/glibc libraries on a Linux system? Even with a system like RPM, that's a real pain to keep track of, particularly when you're building new software.
Has this been addressed by anyone? It'd seem its an issue for things like Gnome and KDE. If its March 2000, and I've recently upgraded by Gnome and now have GTK 2.0 on my system, will the binary of Communicator 5.0 I'm running still work? Switching to GTK 1.2 broke most of the older apps I had.
Using the typical is-linux-read-for-the-masses example, would my Grandmother be able to straighten out that mess? Hell no!