Sarcasm – You Cannot Do Much Without M$ – EndSarcasm

Well, the whole city of Sunnyvale (pop. 140K) does a lot with GNU/Linux, apparently. They are mostly Google (employees: 54K) and use Goobuntu or some variant of Ubuntu GNU/Linux. “Bushnell was asked why Ubuntu instead of say Fedora or openSUSE? He replied, “We chose Debian because packages and apt [Debian’s basic software package programs] are light-years ahead of RPM (Red Hat and SUSE’s default package management system.]” And, why Ubuntu over the other Debian-based Linux distributions? “Because it’s release cadence is awesome and Canonical [Ubuntu’s parent company] offers good support.””If one of the largest corporations on the planet can use GNU/Linux widely, so can you. QED

I recommend Debian GNU/Linux for all the above reasons except cadence. Debian releases Debian GNU/Linux when it’s ready.

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology. Bookmark the permalink.

36 Responses to Sarcasm – You Cannot Do Much Without M$ – EndSarcasm

  1. oldman says:

    “The deal from Microsoft is not that great.”

    On paper perhaps not, but speaking as someone who has dealt directly with both microsoft and RedHat paid support (we have both at the enterprise level) The microsoft support is superior.

  2. oiaohm says:

    ch
    –That’s a damned good warranty.”

    No, it’s not. Rather, it’s a support contract offer. A warranty would include stuff like “Things won’t break with the next update” or “All bugs will be fixed within xyz hours”.–
    That is not included with out the box products of Microsoft or even standard Volume Licenses.

    About time you read the MS licensing. Default warranty of windows basically says you will get updates for so many years. Not how fast or how good. Note of that All bugs will be fixed within xyz hours does not apply to Windows. There are cases of bugs in Windows not fixed before end of life of the product that were found in the first few days of it release.

    –“Things won’t break with the next update”–
    Redhat does on their commercial distribution maintain legacy support options. So there is a Things will not break in no repairable way next upgrade or update with Redhat.

    ch the true reality is Redhat and SUSE has more legal and contract responsibility to the end customer for same money spent than Microsoft. The deal from Microsoft is not that great.

    This is where you are really screwed. The only reason not to be choosing Redhat over Microsoft is you are doing some function that Redhat Enterprise Linux cannot do.

    Cost and Quality of Support Redhat wins over Microsoft every time.

    ch reality people like you are bring strawman arguments that are simple to disprove and you should know better than use them.

  3. ch says:

    “That’s a damned good warranty.”

    No, it’s not. Rather, it’s a support contract offer. A warranty would include stuff like “Things won’t break with the next update” or “All bugs will be fixed within xyz hours”.

  4. oldman wrote, “No they dont, they sell SUPPORT”.

    Hmmm… what’s the difference between a warranty saying X will send someone to your door to fix things if anything goes wrong and someone selling support to fix things if anything goes wrong? A warranty is just a clause of contract of purchase. A contract to provide service may be much the same.

    “With a Red Hat subscription you’re entitled to production support*:

    Unlimited incidents
    Available 24×7 around the world
    Not limited to issue resolution—we’re here to help you avoid issues from the start
    Unmatched technical expertise of our engineers
    Extended relationship with partners
    Rich content guiding you through planning, deployment, and operating your IT
    Optional support programs are available for Technical Account Management and developer support.”

    That’s a damned good warranty. I would take that any day over “you may return the CD in the next 40 days for a full refund…”.

  5. oiaohm says:

    iLia
    –A program simply can create a list of all RSS feeds before actually downloading new messages, check it and remove duplicates and only then download new messages, you see it is not that difficult.–
    That is what causes Outlook to malfunction. List getting out of sync with folders. Leading to nasty results to back end storage. Not that difficult to make a program that will fail and destroy your data.

    Problem is not as simple as it first seams. You avoid 1 issue you risk bring another issue straight down on top of you. Thunderbird solution is stable but with a few annoying issues.

    iLia
    –And why only copy why not to move it? It seams to be more intuitive.–
    In thunderbird data storage there is no such thing as a move. There is only create, read, write and delete. This is the 4 standard operations of most storage. So a move operation to you makes a copy then deletes the source. So that if the program crashes half way threw your data is not lost.

    Safe data handling does bring a few headaches. Since lot of times it basically mandates a move operation cannot exist particular with things doing remote access since as part of doing a safe move you have to duplicate and you don’t want to duplicate the remote access.

    What seams intuitive when you get down to the nitty gritty of the code is now not a really safe path for the data.

    Of course there might be away around this but I still don’t think you would be happy all the time. Duplicate the URL but have the new copy disabled until the source is deleted and a crash at the wrong time still might lead to items not downloading. So bug free doing what you want is basically impossible. iLia. Question is really what bug do you want. Maybe if you are able to think that through you will be able to place a valid feature request.

    Some-things are not defined by GUI convention because there is simply no 100 percent safe path. So each implementation is allowed to choose the path and risks matching.

    I have read gpl. You did not.
    “IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW”

    Note required by APPLICABLE LAW. GPL does not contain a get out of it completely. If you provide GPL for medical usage and it kills you can be held to account. BSD and MIT licenses on the other hand not so good. Introducing internally evil code into a GPL project can also be prosecuted due to the contained Applicable Law clause. MIT and BSD you are home free.

    GPL coverage protect from common law does not protect from criminal. So GPL does mandate some responsibility on the part of the authors since they still can be charged under the criminal code. GPL is not a complete waver. Microsoft EULA contains a complete waver.

    iLia
    –No responsibility here–
    Is a complete bogus statement for GPL. You got it right for MIT and BSD. Does this now explain why I have a preference for LGPL, GPL, MPL over MIT and BSD licenses.

    iLia
    –And, oiaohm, Linux source code is open and accessible, but actually many and many bugs go unnoticed for months or years, and maybe known to computer criminals, so open source do not save from the damage caused by the software.–
    True but when the event happens you can track down what coder did the alteration and make sure everything he has done has been reviewed. So a rogue programmer can get found.

    Bugs being known to computer criminals first is true for Windows as well. Since a lot of so called criminal tech comes from intelligence agencies that do have access to the Windows Source code. So the criminal has access to the source code you don’t. So closed source is not really a magic differences here for desktop computers.

    Lot of medical devices with closed source don’t fail due to an attacker. They fail because people don’t know what they will react to. So they get exposed to stuff that like resets the settings the person dies due to device malfunction. A system does not have to be hacked to kill. Lot of implanted medical devices these days that is insane you can configure them at 150 metres without ever touching the person. These include pace makers. More fun some are 2.4 Ghz same a bluetooth and wifi. In fact some are in fact bluetooth because bluetooth radio chips are cheap. So basically anyone with a mobile phone and attack software could kill a person without ever coming into contact even possibly by mistake. Might just kill them by bad luck that the radio noise happens to make the right signal. Basically with medical devices we are to the point of insane. No sane person who really thinks about it wants a device implanted in them that can be reconfigured without physical contact of some form. We need regulation soon(more like yesterday but you cannot rewind time).

    Security by Obscurity kills when it comes to medical devices partially when they are not doing basic physical security. Like it could be as simple as having a chip that detect the present of a super magnet or charging coil to allow programming o changing configuration otherwise. The requirements to make implanted devices safe from attack are not that high. Little physical security to control how and when they can be altered. Not doing saves a few dollars and a little weight.

    oldman
    –No they dont, they sell SUPPORT. RedHat is just as capable of disclaiming warranties as any company trying to keep from being sued.–
    Yep of course the any company include Microsoft.

    Not everything RedHat can disclaim due to GPL and MPL. So really RedHat as more responsibilities to you than Microsoft.

  6. oldman says:

    Nothing in FLOSS prevents an OEM/retailer/service provider from selling a warranty.

    RedHat and others make a good living doing so.”

    No they dont, they sell SUPPORT. RedHat is just as capable of disclaiming warranties as any company trying to keep from being sued.

  7. Nothing in FLOSS prevents an OEM/retailer/service provider from selling a warranty.

    RedHat and others make a good living doing so.

  8. iLia says:

    Fun right iLia what closed source lets companies get away with.

    Have you read GPL?

    NO WARRANTY

    11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
    FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
    OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
    PROVIDE THE PROGRAM “AS IS” WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
    OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
    MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
    TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
    PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
    REPAIR OR CORRECTION.

    12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
    WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
    REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
    INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
    OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
    TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
    YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
    PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
    POSSIBILITY OF SUCH DAMAGES.

    No responsibility here.

    Maybe BSD License?

    THIS SOFTWARE IS PROVIDED “AS IS” AND WITHOUT NY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF ERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.

    Nope!

    MIT?

    THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

    No!

    And, oiaohm, Linux source code is open and accessible, but actually many and many bugs go unnoticed for months or years, and maybe known to computer criminals, so open source do not save from the damage caused by the software.

    What you want when you make 10 copies of a folder for it to download it 10 times so eating up your bandwidth.

    there are other way to prevent it. A program simply can create a list of all RSS feeds before actually downloading new messages, check it and remove duplicates and only then download new messages, you see it is not that difficult.

    But what ThunderBird does when user wants to drag&drop a subscription? When the user copies one subscription record into another folder, TrashBird creates a new record with the same name but without feed address, and then user has to copy feed address from the initial record and past it into the new record, and only after all it he can remove the initial record.

    And why only copy why not to move it? It seams to be more intuitive.

    No surprise that in Linux/Unix world users are called “loser”.

  9. oldman says:

    “Sorry iLia you don’t know these topics so you are just going to end up with me listing dead bodies caused by closed source and lack of auditing”

    Please do so URL’s please.

  10. oiaohm wrote, “the reason for closed is patents those hardware makers don’t own”.

    That’s only valid if software patents are legally enforceable. That’s not true in most of the world and shortly will not be true in USA. Software is protectable under copyright, not patents. Software created according to some specification and licensed under GPLv2 is not encumbered.

    “Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.”

    See? Software is not any of those things that are patentable. It might be patentable as part of a machine but shipping GNU/Linux on a CD etc. is not covered. So, there are issues of patentability only if GNU/Linux is installed in some device. That can simply be covered by the deal to sell/buy/transfer the device, nothing to do with software development.

    Currently the courts have this wrong but SCOTUS will sooner or later straighten that mess out. I hope sooner or the US economy may never thrive again. Imagine every “smart” device needing 1000 patent-holders to be paid. Imagine a world where software is free. Can USA compete against the world with swarms of piranha/lawyers holding them back? Manufacturers already flee USA. Preventing legal importation of modern technology will make USA a backwater if software continues to be patentable in USA.

  11. oiaohm says:

    iLia lot of medical devices turn out to be hack-able to possibility of killing the person who has them fitted. Because they are not maintained properly at all. There is a lot of calls to open source medical devices for auditing due to the fact many have been demoed being made malfunction.

    When a computer based device in the USA is approved for implanting the software is in fact not checked at all this is true for most countries. All that is checked is that it will not chemically react in the casing. Humans die every year due to software malfunctions in implanted devices.

    Raising medical devices means you don’t understand the problem. There is no requirement for a company making a medical to have there code base audited at all. So it passes human testing its good enough that is it for medical devices. Now if you are the poor sucker who causes the bug event or is near a wifi point broadcasting something that causes a buffer overflow or otherwise bad luck you if it stops your hart brain or something else critical.

    Sorry iLia you don’t know these topics so you are just going to end up with me listing dead bodies caused by closed source and lack of auditing.

    iLia
    –But it makes it much more easier to understand how it works.–
    It also makes it simpler to audit that the driver is doing what it should be doing and not killing the user in the medical case.

    You don’t say after a car crash you cannot look at the wreckage. But when a medical device kills someone even with a court order you may never get access to the source code.

    The answer NO ONE is more true with closed source medical devices. Who will know if the company who made the medical device just deleted all record of the source code. Answer NO ONE they are not required to present the source code to anyone or even archive it.

    When you come with a court order you did not retain the source code so since the device fully fried you cannot prove if it was software or not without getting one cut out of another living person at times. This has really happened a few times so far. In fact by not keeping the source code they have done nothing legally wrong. So yes detect a issue want to hide blame for a while delete the source code and all records knowing the fault is no longer your problem in the world of medical devices.

    Fun right iLia what closed source lets companies get away with. Particularly when your life could be on the line.

    Video card drivers is a very complex problem. The Linux world let the graphical stack rot as they focused on super computers. Opengl 3.x was classed as impossible for Open Source to achieve and it has achieved it.

    iLia
    –You know there are some stable GUI paradigms over there and if you don’t follow them it is your problems.–
    Copying or moving something that does a remote operation is not define by GUI paradigms so get over yourself. What you want when you make 10 copies of a folder for it to download it 10 times so eating up your bandwidth.

    Something that are not define by GUI conventions you do need to read the Manual. Its not only thunderbird that deletes remote operations when you do that.

    At least Thunderbird does not copy outlook and the magic reappearing and disappearing folder since there was a remote operation and you moved it that normally chains onto a full outlook storage database failure.

    Yep you are complaining about something that causes a fatal failure in Outlook. So get over yourself iLia. You only temp had link failure you did not lose your records.

  12. iLia wrote, “medical equipment and similar things. Who is going to maintain their drivers?”

    The kernel boys and girls have offered to do just that. They don’t even need a device if they have the detailed specs and can ensure that the driver works indefinitely in future releases of Linux.

    “Hardware vendors who submit specifications will now receive “a complete and working Linux driver that is added to the main Linux kernel source tree,” according to Kroah-Hartman.”
    see Linux Community Offers Free Drivers to Hardware Firms

  13. JR says:

    For the posters with a sense of humour
    Here is some more:

    http://xkcd.com/612/

  14. iLia says:

    iLia so you did not read the manual on thunderbird.

    Of couse I did not, why I should read manual of a GUI program?

    You know there are some stable GUI paradigms over there and if you don’t follow them it is your problems. Why on Earth a simple drag&drop should modify the content of the moved object removing the most important information?

    Open source driver does not make the hardware free of charge iLia

    But it makes it much more easier to understand how it works.

    Of course closed source driver perform better.

    my dear oiaohm, you are speaking here about very popular videocards, and still the open source community is not able to yield a good driver, and what about small companies with only thousands or tens of thousands of sold devices? You know medical equipment and similar things. Who is going to maintain their drivers? You? No, you prefer to BS on the Internets. Me? Absolutely the same. Mr.Pogson? Guess yourself!

    NO ONE

  15. oiaohm says:

    iLia PS if Linux kernel went GPLv3 taking away hardware makers means to use patents to protect there products over night Linux would not have any more kernel submits from hardware developers.

    So hardware makers other than a few odd balls don’t give a rats about the source code of there drivers.

    Some accept-ions are video cards where sections of the video card drivers is covered by mpeg la patents. So the reason for closed is patents those hardware makers don’t own nothing else.

  16. oiaohm says:

    iLia so you did not read the manual on thunderbird.

    That move folder cease change is a feature. Of course you missed that you could lock a folder from being moved or that you could cut and paste special to keep it. Issues with no reading the manual features that are good for some people bite you.

    –No it is almost impossible to make money on open source (SUN), and I don’t think that many companies are ready to open their source code, no one wants to help it’s competitors to develop similar products. What good got SUN from IBMs or Microsofts Java implementations?–

    Really you are trying to prove you are a idiot right. IBM JIT alterations ended up mainline Java. You should have stopped with just Microsoft Java implementations. Of course you missed Redhat Java implementation that has also helped SUN or what about the Apple Java that has helped SUN. Microsoft was the odd ball out not attending java development meeting to share what they had changed.

    Really for a list of companies willing to share there source you don’t have to look past the Linux kernel commit logs. iLia you are talking many thousands of companies who make hardware.

    What is the point of making the hardware if it does not run on an OS.

    iLia
    “no one wants to help it’s competitors to develop similar products.”
    That is what patents are for. GPLv2 of the Linux kernel allows limited patent grant. GPLv3 would not.

    Open source driver does not make the hardware free of charge iLia. How many companies bill you for the drivers anyhow. This is the reality. Drivers are a loss on the books that has to be recovered by hardware sales. So it really makes no sense prevent others from providing you with patches to them. Since those fixes you have not had to pay for being developed.

    Really do you think closed source drivers stop your competitors from reversing how your hardware works if they want to. Closed source is only like an envelope around a letter. It does not stop those with motive working out how your hardware works.

    Heck if your competitor is truly evil they will just produce full clones of your product your branding and all and send counterfeit into the market. So don’t have to bother breaking your driver doing that.

    See lose arguement have to run to Sun as an example. There are many other companies like Samsung, IBM and Google who happily make a profit using FOSS along the way.

    iLia you don’t have a single valid arguement here. Problem is you did not know it because you did not understand the problem. Don’t pay attention to who is submitting code to Linux to work out lots and lots of hardware makers do and that list gets larger each year. So hardware makers don’t argue at all with your ideas.

    Even the Nvidia and ATI drivers that are called closed on Linux are really part open and part closed. There is basically no such thing as a fully closed source driver on Linux.

    Result has been longer support time-frame for hardware than windows by Linux. Problem is this support time-frame is starting for a lot of hardware when its no longer buy-able with current distributions by the time you get the drivers.

    iLia
    –If a company loses 10% of its revenues due to open source its products.–
    Reality if the company is at this much risk of being killed by open source another company that releases open source as a competitor most likely will see that company go under at some point anyhow.

  17. iLia says:

    oiaohm, please check this link

  18. iLia says:

    Finally I found why TrashBird didn’t updated feeds. When you drug and drop one subscription into another folder new copy of the subscription doesn’t contain the feed’s address. And it is so moronish to manage subscriptions in TrashBird. Nothing works as expected!

    So Open Source drivers solved one problem.

    No it is almost impossible to make money on open source (SUN), and I don’t think that many companies are ready to open their source code, no one wants to help it’s competitors to develop similar products. What good got SUN from IBMs or Microsofts Java implementations?

    If a company loses 10% of its revenues due to open source its products, this company is no more in the business, simply because the profits margins are seldom higher then 10%, and if you lose 10% of your revenues you lose all your profit.

  19. oiaohm says:

    iLia by the way I have a real example of a company not gone under. I have a geforce 6 video card. Geforce 6 is support by open source and closed source drivers.

    Of course closed source driver perform better. There will be no Geforce 6 driver for Windows 8. Not that the company has gone under its that Nvidia will not be releasing one. With Linux I will be able to use on latest release Linux’s after Windows 8 my Geforce 6 with the Nvidia binary/source hybrid driver.

    Reality of long term support FOSS and part FOSS wins.

    Its the short term support where the Linux system has an issue. How to get the newest of new drivers to end users quickly. To fix that fault most likely does not require an binary ABI.

    iLia to know what is a fix to the problem is knowing what the problem is.

    Short term so users can get the drivers they need quickly is the problem. Not having to wait 18 months or other insane amounts of time for the latest drivers so their hardware works.

  20. oiaohm says:

    iLia
    –It is not a solution, you know company sometimes cease to exist and there is no one to maintain this drivers, or a company can decide that supporting old hardware is to complicated, even if it so easy.–

    The best solution to case of company ceases to exist is that the driver is open source in the first place. Or is at least party open source so it can be altered by other parties.

    Binary ABI for drivers does not assist with keeping the maintained in case of company failure. Result from company failure with binary ABI is people running drivers with security flaws that cannot be fixed by anyone.

    Linux supports hardware that Windows no longer does because of its dominance of open source drivers. iLia

    So Open Source drivers solved one problem.

    Its solving the other problem of how to get the newest drivers to users quickly.

    Talking about long term support FOSS beats windows binary ABI hands down. Its the quick support of new hardware the binary ABI advantage to Windows. The backport solution addresses that problem.

    Sorry the points you just raised is why the binary ABI idea does not work. Because there is no way to update to newer OS versions as they come out for hardware where companies who have gone splat if all that exists is a binary driver that does not work for some reason.

    A open source driver or a part open source driver you stand a sporting chance of making it work if the source company has gone splat even a sporting chance of addressing any security issues.

    Funny how people think that company failure is a positive.

    Ok I have a example for you have a bit of hardware that has a XP driver but no Windows 7. For some reason the XP driver does not work under Windows 7. You have no source code or documentation or anything else the company has gone splat. How useful is the Windows Binary ABI now iLia. Completely useless right. This happens thousand of times over. Even vista to 7 there are some drivers that don’t work by companies that are no more.

    So the binary ABI is of limited usefulness and the limited usefulness is because companies fail to be around for the next OS cycle iLia.

  21. iLia says:

    So Linux being Open Source has to develop a different solution to the problem.

    It is not a solution, you know company sometimes cease to exist and there is no one to maintain this drivers, or a company can decide that supporting old hardware is to complicated, even if it so easy.

    The real solution is to have well designed interfaces, both API and ABI and keep the backward compatibility as long as possible, without tweakint them every now and then. When we speak about interfaces you want more stability, than features, simply because when the old version of software of driver you use becomes obsolete and there is no one how want to maintain it you have some problem.

    And when we are speaking about drivers we are speaking not only about drivers issued by huge corporations too big to fail, but also by small companies that disappear every now and then, there are a lot of different devices produced by such companies.

    By the way, about FLOSS quality, ThunderBird for some reason refuses to update almost all my RSS subscriptions. And doesn’t say why. One of the exceptions is Pogson’s feed, I wonder why? Was I hacked by a leftist hackers who don’t want me to read far right sites’ feeds?

  22. oiaohm says:

    iLia binary ABI compatibility has it down sides.

    The issue is a stable binary ABI has already been tried iLia and failed. There was a universal for all Unix, BSD and Linux based OS’s that Linux Solaris and freebsd fully supported. What would have made closed source drivers live very simple. Closed source driver makers were not happy with the wrappers performance so release no drivers it so it gone by by.

    Implementing a stable binary ABI will come with a performance hit this is not avoidable. Even the Linux syscall system is not free of a performance hit.

    –It means that companies should recompile there drivers every time ABI changes, and they should track these changes, and it takes some resources to do, so that is the reason why Linux drivers are usually poorer than Windows, why on earth a company will allocate more resources for creating and maintaining drivers for 1% of users, than for 85%?–

    Or release the source code to the kernel. How much of that process is automated. Every Linux kernel API change for the last 5 years has a matching Coccinelle SmPL file to update your drivers source code. Apply automated patch you are done.

    Maintaining drivers for 1% who are you kidding. Servers and Desktops use almost the same parts these days.

    iLia basically the Coccinelle SmPL forwards porting has worked quite successfully for many years. Reducing maintenance over head to almost non existent for old drivers to newer kernels. Now new drivers backport still has a cost due to lack of automation. Remove that due to another lot of automation and you are talking no extra overhead either way.

    iLia by rebuilding every time as compliers improve bugs that were let slip threw in past get detected before attackers find them and develop an attack in most cases.

    So you want the least number of bugs in kernel space yes or no iLia.

    If you want the least number of bugs you want the drivers rebuild more often you don’t want a binary ABI at all for drivers.

    iLia your arguement here does not hold water. Its going to become more clear after 3.7.

    iLia I know its not the answer you want. You are wanting Linux guys to say yes we have to do a binary ABI then you can go hey we were right all along. Issue is that does not work. You have to deal with the real work closed source driver makers who if they can see the source code and see a path that is faster will use it even if it has big warning signs around it saying we will change this next release.

    Binary ABI idea only works if your OS is closed source to most parties particular the parties writing the drivers.

    So Linux being Open Source has to develop a different solution to the problem. So if a developer uses something he should not it does not cause the worlds worse screw up.

    Some of the window odd ball driver problems come from drivers using interfaces that are not documented and are not meant to be used. So they get updated and all hell breaks lose. So its not only driver developers for Linux that have we well not obey abi instructions. Microsoft had to introduce the signed and approved driver process to get on top of this problem. Even then its not been perfect.

  23. iLia says:

    Linux kernel developers have been attempting for long-time to find away of doing backwards compatible drivers without ABI.

    It means that companies should recompile there drivers every time ABI changes, and they should track these changes, and it takes some resources to do, so that is the reason why Linux drivers are usually poorer than Windows, why on earth a company will allocate more resources for creating and maintaining drivers for 1% of users, than for 85%?

    Linux kernel developers have been attempting for long-time to find away of doing backwards compatible drivers without ABI.

    Instead of attempting to find a way of doing backwards compatible drivers with ABI. They prefer to create new problems rather than solving the existing ones.

  24. TM wrote, “I guess desktop linux users just don’t visit wikipedia, right?”

    The vast majority of hits on Wikipedia are from English-speakers. That’s a small minority of users of PCs, so the data does not represent the universe of PC-users very well.
    “Wikipedia en 46,813 M 2,939 M 39,380 M 4,495 M
    Wikipedia es 9,728 M 354 M 7,468 M 1,906 M
    Wikipedia de 7,964 M 551 M 6,470 M 942 M
    Wikipedia ru 7,467 M 391 M 5,862 M 1,213 M
    Wikipedia fr 6,149 M 259 M 5,219 M 671 M
    Wikipedia ja 5,525 M 424 M 3,472 M 1,629 M
    Wikipedia it 4,668 M 210 M 4,055 M 403 M
    Wikipedia pt 3,270 M 129 M 2,839 M 302 M
    Wikipedia pl 2,290 M 130 M 2,048 M 112 M
    Wikipedia zh 2,101 M 274 M 1,302 M 524 M”

    So, China which has more PCs than USA, is not even in the running for top-8 … What’s with that? Oh, there is a zh.wikipedia.org but few hits by any OS and it only has 500K articles compared to 4 million articles in English.

  25. Meanwhile, here’s some stats from the rest of the world:

    http://en.wikipedia.org/wiki/Usage_share_of_operating_systems

    These are the live statistics for what is connecting to Wikipedia. I guess desktop linux users just don’t visit wikipedia, right?

  26. oiaohm says:

    Thorsten Rahn the reality is only one project I know follows the true release early and release often. This is the wine project itself http://www.winehq.org

    Release cycle of wine is about every 2 weeks. That is about the limit with current setups how fast you can release. This does require serous automated testing frameworks. To the point it annoys the hell out of people submitting patches. Hey I just added a new feature no its not going to get merged upstream until you add a unit test for it.

    By the way Thorsten Rahn you did not need to use a translator that Author publish in English as well.

    http://www.h-online.com/open/features/Kernel-Comment-Release-early-release-often-1696535.html
    http://www.h-online.com is the english site of http://www.heise.de anything from heise check if the author republished it on h-online since this avoid computer translators that can make errors.

    Thorsten Leemhuis does not cover everything. Missed something very important going on.

    https://backports.wiki.kernel.org We have about 4 months before this kicks off. There is more than 1 way to skin the driver problem. Recent years the Linux kernel has been using a lot of automated transformation of source code. Coccinelle SmPL is the common transformation tool used.

    So what they are doing to source compatibleness instead of binary. This is a very different way to skin the cat. Lot of current issues are lack of both binary and source compatibility so forcing the complete kernel to be updated to get new drivers. Question is do we need both or is source compatibility be good enough.

    The source way of skinning the cat means you don’t legacy interfaces around for running drivers so you have less memory foot print and drivers for the new kernel work fast because driver developers cannot use legacy kernel functions because they are not there to be used.

    Now if this backports project works there is every chance it can be applied to the other driver projects.

    This will be a true FOSS invention without question.

    The next big question is if this project works does this mean automated sideways porting of drivers will be possible. Like Linux Kernel drivers ported sideways to FreeBSD and OS X.

    Its the backports project that started some of the recent research into if you need the kernel headers or not to make a driver to work. The answer has been no to that.

    Thorsten Leemhuis is good but he not always up on current going on. Linux kernel developers have been attempting for long-time to find away of doing backwards compatible drivers without ABI.

    Thorsten Rahn I would not group the developers working on the backport project as not interested in end users they are doing it because end users want it.

    Once one project has done it requesting others todo it would be a good thing.

    The backport project is also about reducing maintenance load at kernel.org as well. Supporting all the way back to the oldest long term stable at kernel.org being able to run the tip and the last stable releases drivers could allow a lot of time saving.

    Link backported linux kernel drivers up with dkms framework and you have a god send. Because dkms can tell if you have a newer kernel or not on update of kernel apply if required.

    The backport will allow you to pull a driver from Linux Next that is the development branch for the next kernel. So the waiting bit for drivers will be over once its fully rolled out.

    Thorsten Rahn you really need to be making lists of what ones you want to have automated source backport.

    The Linux world has changed since Android. These changes are only starting to now appear.

    We are under 12 to possible by by to most MS troll arguments against Linux. People have had tunnel vision of we want a binary ABI for drivers. Not thinking its not the binary ABI for drivers it the driver they want. It does not matter really how they get if it binary or source as long as it will work without disrupting things.

    Coccinelle SmPL has been used to forwards port in the Linux kernel world quite a bit. This is why its possible to take a old driver from a prior Linux kernel and make it work in a more modern one most of the time without any human labour. The issue is the process has not been mirrored. The backport project is it being mirrored.

    Once source backporting automatically is proved to work all these waiting for driver issues can end without adding a binary ABI and the security and performance nightmares that brings. Thing is this automated back-porting most likely will be able to be applied to other things as well.

    Thorsten Rahn it will also screw up the comments that FOSS did not invent anything arguments perfectly.

  27. Thorsten Rahn wrote, “FLOSS proves once again that they’re not interested in their users. Is there anything except the Linux kernel in FLOSS land that’s developed in a relatively sane fashion?”

    You mean like

  28. GCompris, an educational software package for youngsters written and designed by teachers who know what they are doing,
  29. VLC which has gazillions of downloads,
  30. audacity which has gazillions of downloads,
  31. LibreOffice which is thriving, and
  32. GIMP which anyone can use, even the little woman, to touch up photos they get wrong,
  33. and thousands of other FLOSS projects widely used and enjoyed? You know, there’s hardly anything one cannot do in software with a basic design with a modular approach. Software which has a well-designed purpose does not need to be bloated like some “commercial” junk I have seen. It’s easier to obtain as FLOSS, and it’s simple and easy to use. Most FLOSS projects are like that and I am greatful to all the developers who cooperate to make the world a better place.