Announcement

Collapse
No announcement yet.

One (big) reason Kubuntu (or the "Linux Desktop") remains unpopular

Collapse
This topic is closed.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    One (big) reason Kubuntu (or the "Linux Desktop") remains unpopular

    Hi guys (and gals)!

    I just wanted to share one idea of why the linux desktop, particularly Kubuntu, still sees slow adoption rates (although the trend is positive). First off is the issue of stability vs new features. The release cycle is a bit frequent, but the way things are laid out, it has to be in order for any software to be kept up to date. I argue that this is a faulty model.

    Having a system that is stable and workable but without the luxury of updating choice user software (OpenOffice, Firefox, media player, even to a great extent the KDE desktop itself) is not good. It's quite a pain. Here's the thing: Lucid 10.04 is the only distro to work properly with my system (and a few friends', with some heavy tweaks), since distro X or Y some years ago. Having spent countless hours debugging and tuning the system to play well with our hardware, waded through countless forums and chats, and invested plenty of sweat and tears into personal debugging, we all found a system we could be very proud to call our own, and we were happy to contribute back to the community.

    Now, satisfied that our computers interface perfectly with all our hardware, we've moved on to jobs, school, and so on that afford us little time to invest heavily in repeating the process for new distros. It's a well known (and disenchanting) truth that some distros work, and some don't, through no fault of the developers (usually). But again, here's the thing: there's no guarantee (or even likelihood) that all of the hardware will work properly with the next incarnation of your favorite distro, kernel, X, hal, or whatever new drivers come our way.

    This is a somewhat broken model for many people, particularly those with average assortments of hardware that experience changing levels of support. It's rare in my experience that every version of Ubuntu from their starting version to 2+ versions later works just as well for them. The same tricks/tweaks rarely work in the same way with all of the structural changes in the distro.

    My point is this: Once we get a distro working properly with our hardware, we're out of luck for all future software updates as well. The solution? Keep rolling out new versions of the software, installable at the user's discretion. It's very difficult and time consuming to obtain new copies of OpenOffice, Firefox, KDE software, etc within the framework of the distro we worked so hard to get working properly. All software except that which interfaces directly with the hardware should be constantly updated and provided in (a) user-selectable software source(s). We (I speak for the group I'm talking about) are growing very weary of Linux in general (which is unwarranted) because of the need to invest large amounts of time into keeping abreast of the essential feature updates.

    Some questions you might be wondering:
    Why just update the user software and not the hardware-interfacing stuff?
    - Because the hardware stuff is likely to result in breakage when updated, or otherwise necessitating sweeping changes in architecture with all attendant bugs. User software doesn't because they're just programs that run atop the framework established by the former.

    But the same argument applies to user software: if it's not broken, don't fix it. If your hardware-interfacing software (kernel, X, drivers, hal, hardware support) is stable and you don't want to change it, why should you bother with the software either if it's stable? It's all or nothing, and the distinction is artificial.
    - New versions of Firefox are required to stay abreast of changing web standards. New versions of OpenOffice are required to maintain feature-parity and compatibility with Office products. More fundamentally, new versions of KDE (particularly 4.5.2 in my Lucid case) address essential bugs/stability fixes plaguing previous versions, even those offered through "semi-offical" PPAs. The software world is changing constantly, but my hardware remains the same. Most people actually tend to keep their expensive computers for 2 years (or more) and don't want to risk breaking their systems for every release.

    What about "rolling distros" that constantly update their software?
    - Many of these distros are perfect examples of the minefield principle. This is because they update both user software and hardware-interfacing software on a rolling schedule, ensuring constant annoyance (or breakage) in some cases. There's a good reason the likes of Ubuntu and set-release distros are more popular-- they maintain hardware compatibility and emphasize stability after release. But the likes of Ubuntu et al use this philosophy too much and too little, simultaneously. Instead of continuing the focus on stability, either: 1) A new distro is released and the old is unsupported, or 2) LTS releases focus entirely on stability and lose software relevancy. A proper approach would be 3) Less frequent releases, emphasis on hardware stability within each release, but provide the latest user software. Rolling for software, not for hardware. That is, provide rolling user software updates such as Firefox, OpenOffice, KDE SC, GIMP, etc but focus only on security/bugfix updates for the kernel, drivers, X, modules, hal, CUPS(?), etc. This would keep the distro relevant, up-to-date, and functional for a span comparable to Windows and OSX. It may even be perfectly feasible to focus entirely on LTS releases every other year or every 3 years (with the option of an "experimental" release in intervening years).

    Some real-world examples:
    - Sony Vaio P. Hardware works perfectly under Karmic with Poulsbo, terrible in Lucid and Maverick. Because it was stuck on older versions of Firefox (which went out-of-compatibility with some web sites), OpenOffice (which doesn't support the many essential docx features like the new one), KDE (with bugs that will now never get solved), I had to go back to Windows 7.
    - Anything with Intel GMA8xx to 9xx graphics: Hardy/Intrepid+self-installed 2.7.1 driver was fastest and most stable. There were many threads devoted to this. Intel graphics support has since ditched this kind of acceleration, leaving these systems in the dust (but supported perfectly, speedily, and stably by Windows 7 Ultimate x32/x64). This was made impossible by later versions of the driver and kernel. Hardy systems are now mostly unusable for modern work, as Firefox, OpenOffice, etc are quite antiquated. So all of these systems got Windows, too.
    - My friend's UMPC (forget the name) had a special webcam and fingerprint reader he could only get working in Intrepid and (somewhat) Jaunty. These are now obsolete in terms of software, and he reverted to Windows 7.
    - Brand new Dell laptop -- got it all working with Lucid finally, but Maverick breaks some things again. This is a core i7 system with new NVIDIA chipset, so it's not antiquated. But I made the mistake of upgrading to the new KDE 4.5.1 via kubuntu.com's instructions. This is apparently a dead-end, because they are canceling the PPA, even though I thought LTS Kubuntu meant the dot upgrades would make it. This left a very bitter taste in my mouth, and I now write this on Windows 7. Dual-boot for now, but without software updates (or the promise of more pain getting hardware to work again in Maverick), I really don't want to spend another few weeks tinkering.

    Windows 7, for the record, works perfectly on all of them, even the older ones (with some tweaking). It will continue to support all software updates well into the future. Even windows XP receives software updates for virtually every piece of major software, and it's almost 10 years old. Support is just now beginning to dwindle with IE9, WMP12+, etc on the horizon, but currently Firefox keeps itself up-to-date, Office 2010 installs easily, etc. Not even the current LTS release of Ubuntu can claim to be as up-to-date... it already lags behind the competition, and Maverick isn't even out yet.

    tl;dr version: keep updating software of user's choice for the lifespan of the distro, at least for LTS. I thought this was common sense, but the developers do love tinkering, so it's hard to get the message across that the mainstream (or even geeks like me) actually want to keep a stable, modern system without screwing around with their hardware every few months. This means keeping the functional hardware stuff in place, but keeping the software up-to-date. Anyone?

    #2
    Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

    All software except that which interfaces directly with the hardware should be constantly updated and provided in (a) user-selectable software source(s).
    Well, I can't agree more. At least we should be allowed to upgrade the "bundled software" (or as it is called "included software" in the *buntus installer slide-show) constantly. Maybe a "included software" ppa should come default. I don't know how intricate the technical details may be, but it sounds very plausible to me.

    Regards.

    Comment


      #3
      Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

      I think everyone is being acknowledged in this regard by Debian. Where they are attemping something a little more in line with what your asking. I ran across this article reference on the mepislovers forum http://raphaelhertzog.com/2010/10/04...-distribution/
      They are calling it "CUT". Basically a snapshot that could be froze on a faster cycle.
      Where would this leave all the derivative works? With more time for... testing, developing maybe even a vacation :P

      Comment


        #4
        Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

        Since switching to a "rolling release" distro about a year ago, I found that I (paradoxically) spend far less time maintaining the systems compared to *ubuntu with their 6 month release schedules, and more time getting actual work done. YMMV.

        Comment


          #5
          Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

          Originally posted by sixonetonoffun
          I think everyone is being acknowledged in this regard by Debian. Where they are attemping something a little more in line with what your asking. I ran across this article reference on the mepislovers forum http://raphaelhertzog.com/2010/10/04...-distribution/
          They are calling it "CUT". Basically a snapshot that could be froze on a faster cycle.
          Where would this leave all the derivative works? With more time for... testing, developing maybe even a vacation :P
          Interesting link, but not quite there -- both hardware and software support are kept at "testing" stages, meaning the drivers, kernel etc keep changing, leading us back to square one. A solid hardware foundation with rolling user software is what I'm getting at.

          If anyone here uses Windows or Mac, you'll know what I mean. The kernel and such stay the same (barring "security and bugfix" updates and such), but the software on top can easily be kept up to date. I've never changed drivers on my Windows machines (many of my friends don't even know what they are), but we all keep our Firefox up to date (automatically), get the new Office or OpenOffice when it tells us there's an update, etc. In mac it's better, because there is a somewhat-universal software update framework (kind of like in many linux distros).

          Comment


            #6
            Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

            Originally posted by skunk
            Since switching to a "rolling release" distro about a year ago, I found that I (paradoxically) spend far less time maintaining the systems compared to *ubuntu with their 6 month release schedules, and more time getting actual work done. YMMV.
            Rolling is good unless the hardware support rolls away from you. Only user software should be rolling. Then of course, that would be paradise.

            I checked the comments for your link, sixonetonoffun. A comment by Sanders hits the nail on the head. I replied with a link here.

            Comment


              #7
              Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

              Originally posted by gabe
              ....
              I checked the comments for your link, sixonetonoffun. A comment by Sanders hits the nail on the head. I replied with a link here.
              Which was:
              Spend some serious time getting the hardware supported, then freeze it with constantly updating user software above it. Windows and Mac have done this
              But that opinion overlooks some serious difficulties, and belittles some dedicated developers whom, you seem to think, are not serious about their coding.

              First, Apple MAKES and standardizes their hardware against their kernel, so that comparison is unfair.

              Second, Microsoft released VISTA and many found that it would not run well, if at all, on their older PCs and many newer or new ones, even when PC OEMs put "VISTA Ready" stickers on their PCs in order to sell them. Then, Microsoft released Win7, plainly stating that new hardware would be required in most cases. New versions of Microsoft's software will only work with Win7. Microsoft can do this because they still have the PC OEMs by the short hairs. IF an OEM decided to offer Linux in any significant way (on the SAME hardware, with the SAME peripherals as Windows, advertised on their URL FRONT PAGE), which they've never done yet, Microsoft's reaction has been to cut off their ad rebates. With current PC OEM profit margins razor thin ad rebates are often the only profit an OEM makes. We learned recently that Intel was paying DELL to NOT include AMD video chips in their hardware. When DELL decided, in the fall of 2007 to use AMD chips Intel stopped payments and DELL's income dropped 70% !!! That will give you an idea of what Microsoft would do to the PC OEM that started giving Ubuntu equal billing and offerings on their PCs.

              Third, Microsoft makes sure the PC OEMS make sure their hardware works with Windows and Microsoft ad rebates make sure it stays that way. At least one BIOS maker, Foxconn, has been caught putting traps in their code to make sure Linux fails. When ever new hardware comes out Microsoft gets ALL the specs and data. Linux developers get to "green-room" the hardware, i.e., trial&error coding and testing till it works, sometimes reasonably well with THAT version of the firmware, but sometimes not. THEN, the next release of the product has unannounced, to everyone except Microsoft, firmware changes that make the Linux software which worked well with the previous hardware version experience trouble with the new firmware version. Since the new firmware isn't necessarily better or cheaper, I suspect that these mid-product life cycle changes are made deliberately to hinder the progress of Linux.

              Forth, Linux is famous for supporting older, often underpowered, hardware. But, vendors are infamous for NOT doing so. Hence, if you have an older PC with ATI's Radeon 9600 video chip and you try to install their latest driver you get a msg saying it does not support that "legacy" chip. So, you download the version of their driver for that chip. When you attempt to install it you are blocked with a msg from it saying it will not work with the current Linux kernel. So, the Linux user with the ATI Radeon 9600 chip in his laptop is faced with running a kernel old enough to use the version of ATI's driver that supports the 9600. Otherwise, the FOSS radeon driver is used but it doesn't support all the features in the chip. In doing so she gives up the power and features of the modern kernel and is unable to attach and use newer hardware, or features in the kernel space that give current Kubuntu its power. Also, the same video chip in different PC OEM PCs work differently, so a patch that allows the video driver for the Intel Mobil Series 4 chip in my Sony to work well, doesn't help the HP PC with the same chip because the hardware design and chip incorporation is different. Both Sony and HP work to make sure their video drivers work with Windows. They don't work with Linux Xorg developers in the same way, i.e, passing on proprietary data and spec info necessary to create a perfectly working driver.

              Fifth. You PAY for Win7 and Microsoft pays for product designers, ergonomic testing, demographic studies, etc., and when you get new PC it comes with everything working (most of the time). Just like you get $2.75/gallon gasoline because the US gov is subsidizing it, Windows is cheap to buy, relatively speaking, because the COST of VULNERABILITY is subsidized by users and our gov. Because Microsoft cannot secure Windows reasonably well, and doesn't even try, you have to buy AV software and decent firewalls. Even with those millions of Windows boxes get compromised and Windows users have hundreds or thousands of dollars stolen out of their accounts because Windows and ALL its 3rd party AV products can NOT block simple keyboard loggers. So, just like it happened in my city this past month when hundreds of bank accounts had money stolen, the banks made good the losses. BUT, the banks DO NOT eat those losses, they pass them on to ... guess who? ... YOU and ME, the rest of the banks customers. Ergo, the general public is subsidizing Microsoft's insecurities.

              It's getting so bad that yesterday a security consultant advised that Windows users stop using Windows for banking or online shopping and start using a "LiveCD": http://voices.washingtonpost.com/sec...e_bank_on.html



              "A nation that is afraid to let its people judge the truth and falsehood in an open market is a nation that is afraid of its people.”
              – John F. Kennedy, February 26, 1962.

              Comment


                #8
                Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

                Not debating the merits (or lack of them) in the 6-month cycle, but you have to note that Ubuntu have been doing this since October 2004, and in that time have expanded the useage of Linux vastly, as well as keeping itself (and linux, gnome, etc) in the spotlight for at least as long. So something must be working, at least a little

                There have been small changes: Firefox now gets regular updates in version for all supported releases, iirc. Kubuntu users get to update KDE to newer versions between releases.

                In our particular case, I think the main reason we don't have, say, KDE 4.5.x in Karmic is as much a manpower issue as much as it is just plain a lot of work. We are, after all, a community-developed distro, as is just about every distro out there.

                Sometimes we forget how the community aspect affects how things work, for better or worse

                Comment


                  #9
                  Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

                  I certainly agree that lack of OEM and vendor support for linux is crippling the driver situation, especially for graphics cards. Poulsbo (which I mentioned) may be an even better example than the 9600, since its support is essentially zero without a very specific configuration of kernel, x, and drivers (in which case it can work quite well for basic usage, light gaming, desktop compositing, etc).

                  Though everything you said is very much the truth, and I heartily concur with the greatest fervor, my point remains the stronger for it. Vendors (including Adobe themselves, who have a linux development department) have expressed time and time again the same qualms I have -- the lack of a stable hardware interface leads to stagnation and splintering of drivers and frameworks, and even if a vendor wanted to produce a driver, it would take all of the manpower you speak of to create one for the major distro(s) every few months, only to have it rendered horrifically obsolete by the kernel, x, hal, and various other API/ABI restructuring and framework changes.

                  This adds great fuel to my initial point -- not only users, but vendors in the industry (exempting those who simply don't try/advertise it because of payoffs) are genuinely discouraged by the same exact painful process that plagues users -- the fact that the interface between computer and hardware is disrupted so drastically every 6 months -- and we're just talking K/Ubuntu here!

                  Ubuntu espouses many wonderful concepts, and has made great leaps in the usability of the operating system and the offered software. That's Ubuntu's success and a main reason for its popularity. One thing that takes away or hurts its popularity, however, is precisely the issue I'm bringing up. As well as it works when we get it working, we're doomed to yet another ultimatum 6 months later: either sacrifice the time and energy to risk reconfiguring a potentially broken system, or stick with obsolete software. For driver makers, it's even worse -- how can we (they) rework this driver to function properly with the next kernel/x (which may be broken in different places and introduce drastically new architecture, drop old methods of working, etc).

                  You mention Windows is an insecure, unfair, closed operating system. I agree. This is why I want to have a viable recourse with Linux. And a LiveCD is not a valid choice of operating systems (though one could argue that Ubuntu in current form is essentially a LiveCD regardless-- hit or miss on your hardware, remade every 6 months and persistent to small user changes/files but frozen in time to all major software upgrades).

                  Kubuntu users get to update KDE to newer versions between releases.
                  I've noticed. Now I'm locked at KDE 4.5.1, which was notoriously buggy and was now cut from support, so it'll never be upgraded or easily downgradable. In other words, stuck right back at square one. Upgrading aside, if even downgrading collections of software were a bit more straightforward... ppa-purge anyone?

                  In our particular case, I think the main reason we don't have, say, KDE 4.5.x in Karmic is as much a manpower issue as much as it is just plain a lot of work. We are, after all, a community-developed distro, as is just about every distro out there.
                  Manpower is about priorities. You guys do some fantastic work. Hell, pretty much all of it fantastic. It's simply a matter of priorities and release cycles that cause these issues. If time weren't spent every 6 months on trying to make brand new kernels play with brand new Xorg components and brand new base drivers and brand new hardware interface frameworks, etc, that time could be spent on a solid LTS every 2 or 3 years (or invested into the current one-- 2.6.32 is NOT a bad kernel, for one thing), and the remainder just spent on rolling out software for it. User software, that is. If a user gets new hardware that's absolutely unsupported by the current kernel/drivers, the official kernel PPA or preferably driver PPA is the place to go, NOT a whole new distro which can cause many new problems, contain many unwanted changes or breakages and require manpower and bugfixes of a whole community.

                  As much as I hate to say it (and as much as Google botched their own development cycles), Android is somewhat on the right track. Because of a limited set of standard kernels and hardware interfaces, it's trivial to rip drivers out of one device and put on another. Phones that never ran Android can just hijack other Android systems and drivers willy-nilly, which works because each Android system is far less monolithic than the current Ubuntu releases. Imagine handing the Poulsbo driver to your buddy with the identical netbook and a subtly different system and expecting it to actually "work." Not a chance!

                  With a solid groundwork of basics, the software can be allowed to roll on top, major releases between can be far less frequent but the software remain up to date for much longer. There would be far less server space required, too -- Canonical is maintaining (seperately!) 5 different versions of Ubuntu: Hardy, Jaunty, Karmic, Lucid, and Maverick...the entire repos. And then compile for each one of them every time a security patch comes up. Imagine instead a single release every 2 years with 3 years of support (current LTS desktop model) or every 3 years with 5 years of support. Clearly, rolling out significant software updates would be easier, and the quality of each release's base would likely be quite high. The software sources would obviously then include: Security/bug fixes only (for servers), Stable upstream (for users to have latest stable/non-beta software), and Bleeding edge (CVN-type builds, snapshots).

                  This certainly does not belittle the work of dedicated developers. I can't fathom myself coding for a proprietary technology (such as Flash video acceleration, Poulsbo, any Android closed driver dev) when I know the carpet would be pulled out from under me every 6 months. If a new Vista came out every 6 months, imagine the hell that would cause. But the API and kernel difference between Vista and 7 (like between 2000 and XP, or 95 and 98/ME) is relatively small compared to the differences between these 6-month releases sometimes! A single kernel upgrade can simultaneously render most drivers dysfunctional, remove a method of video acceleration, drop support for the old squashfs and make your compressed system useless, introduce 2 scheduling bugs that grinds your sql processing to a halt, and cause you to entirely replace your version of X and mesa, which are independently developed. With kiddie hacks, plenty of software and drivers "designed" for windows XP works flawlessly on 2000. Obviously without recompilation.

                  The benefits of linux over Windows espousing the same strategy is obvious. Linux boasts improved security and community support. With a slightly less hectic/changing landscape (more "standardized," at least within the most popular distro), developers would feel more welcome to code such things as drivers without fearing facing obsolescence in a few months.

                  Comment


                    #10
                    Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

                    Now I'm locked at KDE 4.5.1, which was notoriously buggy and was now cut from support,
                    On which planet? Certainly not this one. KDE 4.5.1 is famously stable and I just got an update this morning.
                    "A nation that is afraid to let its people judge the truth and falsehood in an open market is a nation that is afraid of its people.”
                    – John F. Kennedy, February 26, 1962.

                    Comment


                      #11
                      Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

                      There was an issue with its use of dbus or something, so previews lagged in dolphin and such. It was quite annoying.

                      On what planet? Come on... http://dot.kde.org/2010/08/31/kde-releases-451

                      Topic changes much?

                      Comment


                        #12
                        Re: One (big) reason Kubuntu (or the "Linux Desktop&quot remains unpopular

                        My points remains:
                        1) KDE4 4.5.1is NOT "notoriously buggy", it is running on this laptop much better than VISTA did.
                        2) You are NOT locked into KDE4 4.5.1 because no one is preventing you from removing it and down grading to a previous release or to Gnome, or Xubuntu or some other desktop or distro, and
                        3) support for 4.5.1 CONTINUES, as witnessed by upgrades received even today.

                        A point that seems to evade you is that this forum, and others like the one you cited, are not primarily for "Attaboy" messages or testimonials, although they do appear from time to time. People post questions here MOST of the time because they are having problems. That a person posts a problem about KDE4 or one of the packages in Kubuntu does not necessarily mean that ALL users of KDE4 or Kubuntu are experiencing that problem.

                        Problems can be specific to hardware, and not everybody uses the same hardware. In fact, even hardware of the same make and model can have different versions of firmware in them. KDE4 may work with one version of the firmware but not the other, so two people, both using a Gateway m675prr laptop can report different results running the same program -- an observation I made because my laptop worked with a certain video driver and my friend's same laptop, made a year earlier with previous firmware versions, would not.

                        Besides differences in hardware there are differences in experience and skill sets of the users. NO matter how easy to use something is, someone, somewhere will find a way to make it difficult, mess up, and then blame everything and everyone else for their problem.

                        Computers are, by any stretch of the imagination, a modern miracle. From the hundreds of millions of JK Flip-Flops in the interconnected circuitry, the microcode in the CPU and firmware, the human interfaces like KDE4, and the tools and languages used to program it all ... it works, and even when the lowest bidder wins. I've taught electronics in college and I'm still amazed that it all even works when we turn them on. But, there are two words which describes why it does all work -- constructive cooperation. Nothing significant was ever built with destructive criticism.
                        "A nation that is afraid to let its people judge the truth and falsehood in an open market is a nation that is afraid of its people.”
                        – John F. Kennedy, February 26, 1962.

                        Comment


                          #13
                          I just went from a system that was running just fine, to one that is almost utterly useless. This is solely because of the issues gabe mentions. I am no Linux novice - it was my sole OS for several years. It got better and better for several years, but it's been going backward ever since. This latest "upgrade" to 12.10 leaves me with no touchpad and no WiFi, and KDE crashes constantly. Even plugging in an external mouse, it works for a fraction of a second but then doesn't respond for several seconds in a continuous cycle. The Windows key still doesn't do anything, even though most of the how-tos I've read recently say to use it. This is all after a clean install, forced by an attempted dist-upgrade because I forgot how painful the last "upgrade" was. Back in version 7.x, my Eeepc fn keys (WiFi toggle, mouse toggle, volume, etc.) worked, but they haven't worked since. I've spent the last 36 hours trying to resolve the most heinous problems, but it just keeps getting worse. At least before the "clean install" it was running okay and my hardware still worked. (Even though GRUB it said it was version 11.10 but once booted it claimed it was 12.04, kernels through 3.5 were installed but only 3.0 was accessible and I couldn't install or remove any packages because of insurmountable dependencies and other errors.)

                          So, it's back to Windows 7 until such time as I get a boatload of time to research and resolve everything, which may end with me going back to 10.10 or so.

                          No, Linux is never really going to get significant market share until the nightmare of instabilities and lack of hardware support are overcome. That means not pressing forward with multitudes of changes just for change's sake and it means concentrating on usability and stability rather than flashy irrelevant stuff like window managers and effects. It also means developers coming to the realization that those people who keep getting bent out of shape because some serious issue is keeping them from getting work done are exactly the ones who need to be satisfied before Linux can claim to be a reasonable alternative to Windows. Inaccurate anecdotal "evidence" aside, Windows is in fact quite stable and secure these days, especially considering the user base, and Linux is far less stable and secure than the fanboys will admit. Windows does in fact work, and work with pretty much any hardware, and let you get your work done. Linux is still very hit-or-miss as to whether your existing hardware will work at all, especially critical peripherals like printers and scanners. There's a very good reason that MacOS hasn't and never will get anywhere near the market share of Windows: in spite of Apple's glamour and fad appeal, real people can't get real work done as easily, efficiently, cheaply, or as well as with Windows. The writing is on the wall for iOS as well, for all the same reasons. The number one reason, though, is that with Linux as well as Apple, the ultimate attitude is "you can be happy with what I give you or shut up and go elsewhere." So of course people do shut up and go elsewhere. In droves. Nobody really *loves* Windows, but nearly everybody uses it.

                          It's clear from everything I've ever seen even remotely associated with Linux that "destructive criticism"=(anything you don't agree with) and "constructive criticism"=(anything that doesn't actually involve criticism).
                          Asus EeePC 1000HE WinXP+Kubuntu 9.10-&gt;looking very hard at 9.04<br />Self-built AMD Sempron 2.4GHz Kubuntu 9.10<br />Self-built Via Epia Nehemiah M10000 WinXP+Kubuntu 9.10-&gt;reverted to 9.04<br />Using Kubuntu since 7.04, Ubuntu since 6.09

                          Comment


                            #14
                            Hm. Resurrecting a thread that's over two years old?

                            Originally posted by SilverWolf View Post
                            rather than flashy irrelevant stuff like window managers
                            Window managers are hardly irrelevant. Without a window manager, you'd have no ability to interact with your programs. Broadly speaking, the window manager is responsible for painting the desktop.

                            Originally posted by SilverWolf View Post
                            Inaccurate anecdotal "evidence" aside, Windows is in fact quite stable and secure these days, especially considering the user base, and Linux is far less stable and secure than the fanboys will admit. Windows does in fact work, and work with pretty much any hardware, and let you get your work done.
                            I'll leave the Apple stuff aside. However, I can speak with a certain amount of authority with respect to Windows, having worked for Microsoft during 1998-2009.

                            There is an astounding amount of app-compat and hardware-quirk workaround code in Windows. J. Random User never sees this, of course, but I assure you it's there. Large portions of the operating system exist purely to paper over flaws in hardware or prevent ancient, crufty code from crashing.

                            Furthermore, hardware vendors willingly open their kimonos to Redmond. To ensure that its newest whiz-bang graphics adapter becomes an object of great desire, a hardware manufacturer throws bodies at driver development, sometimes using BIF money provided by Microsoft ("business investment funds"). Rarely are the drivers of extremely high quality, but hey, who's complaining?

                            Against these conditions, Linux is fighting a difficult battle. To keep the system lean, most distributions forgo lengthy app-compat maintenance. Frankly, the amount of app-compat handler code in Windows is almost bringing it to its knees. It may seem stable now, but at some point, a lot of it will get jettisoned unceremoniously. Linux desktops adopt a different approach: it is not the responsibility of the operating system to transparently fix flaws in bad applications. If a newer module or library fixes bugs in an older version, and an application that relied on those bugs now no longer functions, why should the operating system include an under-the-covers workaround just for that application? Answer: it shouldn't. The app needs to be fixed.

                            Hardware is even more problematic. Linux developers must rely on often incomplete and sometimes outdated documentation to implement drivers. Different hardware companies have different relationships with and attitudes toward open source; some actively disdain it. Few are able to throw lots of bodies on such projects, and none receive the kind of funding that would match a BIF grant.

                            Some readers may view my comments as nothing more than extended excuse. That would be incorrect. I'm simply documenting two examples of cultural and attitudinal differences that describe why the gulf between Linux and Windows exists. Windows is the result of tens of thousands of people coding via committee, writing for the largest common denominator possible. Linux is the labor of love of a modestly-sized, dedicated group of very smart individuals, working mostly for free and mostly for themselves, who wish to use their hardware without constraints imposed by shareholders, rights-holders, and the demands of "the market." That an extended community such as ourselves can derive benefit from their work is a testament to their talents.

                            Comment


                              #15
                              Well said Steve, well said. I have argued this with Windows-using friends for years, trying to debunk the relentless "Windows Myth" as I call it. And that myth is: it just works.

                              There is an astounding amount of app-compat and hardware-quirk workaround code in Windows. J. Random User never sees this, of course, but I assure you it's there. Large portions of the operating system exist purely to paper over flaws in hardware or prevent ancient, crufty code from crashing.

                              Furthermore, hardware vendors willingly open their kimonos to Redmond. To ensure that its newest whiz-bang graphics adapter becomes an object of great desire, a hardware manufacturer throws bodies at driver development, sometimes using BIF money provided by Microsoft ("business investment funds"). Rarely are the drivers of extremely high quality, but hey, who's complaining?
                              I bought a new HP laptop a few months back, the first new computer I've owned in many, many years. It came with Windows 7 pre-installed, and not having used Windows since '96 I decided to give it a try, see if it got better and was what so many people claim. When I turned it on and went through all of the initial user set-up stuff, the first thing it did was update itself, which took about an hour. I kept wondering what was taking so long, but tried to be patient. I soon found out that with only 2Gb of ram it had to constantly swap to disk just to do the updates. Nothing else was open, just the desktop and the updater, and I didn't have enough ram for those alone. Talk about cruft! Once that was done, and I started exploring what was on it, I was told there were more updates that needed to be installed! Hell, it just finished over an hour of updating, now what!? I clicked the update icon and told it to ignore, and it went away. Then later when I went to shut it off it tells me not to hit the power button because it was installing updates! I tried to live with that for about a week. After all, how many updates can it need in a week? Evidently, it needs them every day, mostly "security" updates, because it would do that every single day, and I was warned I couldn't shut it down until it was done. Enough was enough, I put Kubuntu on it, Windows is gone. But it did make me curious as to why Windows is such a memory-consuming behemoth. The first part of what you wrote that I quoted above is the reason in a nutshell.

                              The second part that I quoted I did because it points to the biggest complaint I hear from new Linux users, hardware problems. Again, you hit it on the nose. Microsoft does not develop hardware drivers, hardware manufacturers do. Yet I constantly hear "If Linux developers would concentrate more on drivers working for my hardware...". Linux developers? How about complaining to hardware manufacturers. Considering Linux gets almost no help (code) from manufacturers, and considering the sheer number of hardware devices there are for computers today, the existing driver support in Linux is nothing short of amazing. Adding to it the fact that it's being done in spare-time, for no pay, and is constantly being criticized, it's miraculous.

                              Manufacturers have full-time, paid employees. And Microsoft has a war-chest full of money to ensure compatibility. Microsoft has that luxury because you have to pay for every copy of Windows. You can only have that copy on one machine, if you want to use it on more computers you have to pay for each one. Actually, to be completely accurate, you're in fact only paying rent, you don't own your copy. Doubt it? Read your EULA. Even upgrades have a cost. As a side bonus to Microsoft you're also paying to ensure Linux can never compete. Apple knew this from the onset, which is why they made and make their own hardware. And why it's so expensive. Microsoft doesn't make hardware, so they don't have to absorb the manufacturing cost. Apple does. And Linux... oh yeah, that's right, Linux is free. They get some help from donations, some corporate support (that's written off at tax time), that's it. Ubuntu has Canonical, and they've done an amazing job putting Linux more in the limelight, but Canonical doesn't have near the war-chest that Microsoft has. If they did, by now Linux would be on half of all computers and hardware problems would be a thing of the past.

                              Want better hardware support in Linux? Here's an idea. Take what you would've spent on a Microsoft license and instead give it to your Linux distro of choice. Or to the kernel developers, or the Linux Foundation. If the ecosystem around Linux saw an influx of half as much cash as Microsoft does, you'd start seeing manufacturers jumping on board in droves. The problem with Linux isn't there's too many distos, or there's a lack of interest in fixing the problems it does have, or any of the other numerous complaints pointed at as proof why it can't compete against Windows. The problem is it's financially incapable of wooing hardware manufacturers out of Microsoft's bed. If you'd like to see better hardware support in Linux, then support it. Nothing gets the attention of manufacturers more than money. You can even write it off at the end of the year on your taxes as a donation. Try doing that with your Windows rental.
                              Computers don't make mistakes. They only execute them.

                              Comment

                              Working...
                              X