Which is why when a PM or engineer chooses to deploy an open source product part of your job is to select something with well written, maintainable code, an active developer community and large user base. It's no different than a proprietary solution where you have to be careful to choose a vendor that's likely to be around for a while.
The big difference is in the worst case scenario, wiht open source you can hire a developer to maintain it in house. With a proprietary solution you are screwed if the vendor goes away.
I am sorry if you bought hardware from a manufacturer who does not support Linux, but you will have to take this up with the vendor - there's nothing the Linux kernel developers can do if your vendor won't release the necessary information to write a driver.
Fortunately there are plenty of modern machines that Linux fully supports. If you want to make Linux a better OS, evan if you never plan to run it, you can help by purchasing from a Linux friendly vendor. You can ask on any Linux user web forum or mailing list how well a given laptop is supported, before you buy.
In general, avoid Nvidia video cards, and Broadcom and Atheros wireless - these work but may require a binary only driver which is a real pain, can destabilise your system, and generally negates the benefits of an open source OS (eg you won't be able to run 64bit linux if the vendor only releases 32bit drivers).
Given that ACPI barely works on it and is very buggy, therefore most people usually avoid doing anything fancy such as configuring power saving functionalities...
s/ACPI barely works on it/ACPI barely works/
Seriously, read all the comments from Windows users in the thread - Linux is hardly the only OS that suffers from ACPI issues. At least with Linux you can report the bug in bugzilla and track its progress, as opposed to being helpless until Microsoft or your vendor fixes it.
The problem with ACPI is that vendors don't bother to fully debug their implementation - as soon as it boots XP, they ship it, and work around any BIOS bugs that surface later in the drivers - unfortunately for Linux, they don't document these bugs so it's often a lengthy trial and error process.
I read the Supreme Court ruling and it seems like you are only required to tell the cop who you are - they can't require you to produce ID because you're not required to have one.
Actually that website does not even say what the outcome of the case was - all the text on the site was clearly written before the decision, and the PDF link to the decision is broken.
So what happens if a cop asks for my ID and I don't have one? There's no law requiring me to have a passport or a drivers license, so how can I be jailed for refusing to produce one?
I don't think the cops in most places are allowed to pick you up off the street for no reason and throw you in jail if you can't produce ID. If this is allowed we already have a "Papers please" society.
The telecoms seem to want to evolve Linux in the direction of Solaris, which has traditionally been the platform of choice for many of these apps. Lots of it is realtime stuff - robust mutexes, priority inheritance, NPTL, reliable RT scheduling, fully preemptible kernel, etc which also benefit multimedia and games.
Recently Linux has become a better soft RT platform than Windows and is creeping up on OSX, Irix, and Solaris.
With the work being done in the -rt patchset Linux is quickly developing into a world class RTOS. Today Linux runs the desktops, the next generation of rovers could run Linux for the control systems.
Re: the stability of 2.6, a lot of the increased churn is necessary if Linux wants to be viable on the desktop. Lots of key features that desktop users expect to Just Work are still not 100%, like wireless and suspend on laptops.
Servers are boring, if you want a rock-solid server just run Linux 2.4.
"Drivers should be in the kernel if A. at least one of their primary clients exists in the kernel, e.g. disk controller drivers, B. they service a large number of clients directly (e.g./dev/random), C. real-time performance is critical to the correct operation of the device (e.g. audio/video)."
I disagree with C. It's a common misconception that you can't do realtime in userspace but this is really a Windows limitation. This leads to horrors like vendors implementing an audio synthesizer as a kernel driver. A decent OS like Linux or MacOS can do realtime in userspace just fine.
The development versions of Audacity actually are compatible with JACK. The issue is that Audacity relies on PortAudio, and PortAudio v19 which adds JACK support is not released yet. So this will be ready real soon now.
But if you go a few blocks away, you see scenes like I did [amazing.com]. All these pictures were taken on what would be prime real estate in any other country, a block or less from the Malecon, the giant seawall that faces the ocean and is a major gathering spot for Cubans.
Obviously you don't get out of the suburbs much. There are worse slums than that in Philly or NYC.
I'll take the complete lack of any followup to mean that no one on slashdot reads linux-kernel anymore. The hard realtime (yes it is hard RT, check the list archives) capabilities that have been developed for Linux lately are a much more interesting story than this. In addition to the usual suspects (pro audio users, Montavista etc), people doing defense, physics, and high end simulation work have expressed interest and contributed code or testing.
No one in their right mind would deploy the RT kernel for anything mission critical yet, but it's coming along nicely. It can (deterministically!) do about 50 usecs from interrupt to running user code (RTLinux can do 15 or so).
WaveRT is a pretty cool Longhorn feature:
http://www.microsoft.com/whdc/device/audio/wavertp ort.mspx
This will supersede the 5 different low latency audio APIs that people unfortunate enough to use Windows have to deal with.
Of course Linux users already know we have a better system with ALSA and JACK.
Slashdot Mirror
If you want to be taken seriously, please use correct grammar. The adjectival form of "Democrat" is "Democratic".
Oh man, would that be a fun bug report to file.
Software version: Phalanx 5.6
Expected behavior: Helicopter lands on ship, phalanx does not activate
Actual behavior: Phalanx tries to shoot down helicopter, crew needs new underwear
Etc...
It's still better than the worst case scenario for proprietary software.
Which is why when a PM or engineer chooses to deploy an open source product part of your job is to select something with well written, maintainable code, an active developer community and large user base. It's no different than a proprietary solution where you have to be careful to choose a vendor that's likely to be around for a while.
The big difference is in the worst case scenario, wiht open source you can hire a developer to maintain it in house. With a proprietary solution you are screwed if the vendor goes away.
I am sorry if you bought hardware from a manufacturer who does not support Linux, but you will have to take this up with the vendor - there's nothing the Linux kernel developers can do if your vendor won't release the necessary information to write a driver.
Fortunately there are plenty of modern machines that Linux fully supports. If you want to make Linux a better OS, evan if you never plan to run it, you can help by purchasing from a Linux friendly vendor. You can ask on any Linux user web forum or mailing list how well a given laptop is supported, before you buy.
In general, avoid Nvidia video cards, and Broadcom and Atheros wireless - these work but may require a binary only driver which is a real pain, can destabilise your system, and generally negates the benefits of an open source OS (eg you won't be able to run 64bit linux if the vendor only releases 32bit drivers).
Asian Americans Defying Traditional Stereotypes
More and More Are Lazy, Stupid
The Onion rules...
Ha! Good luck getting processors C-states working in Linux. Have you seen the ACPI code in the linux kernel?
Yeah, it sure is crap, especially when you compare it to the ACPI code in Microsoft or Apple's kernel...
Given that ACPI barely works on it and is very buggy, therefore most people usually avoid doing anything fancy such as configuring power saving functionalities...
s/ACPI barely works on it/ACPI barely works/
Seriously, read all the comments from Windows users in the thread - Linux is hardly the only OS that suffers from ACPI issues. At least with Linux you can report the bug in bugzilla and track its progress, as opposed to being helpless until Microsoft or your vendor fixes it.
The problem with ACPI is that vendors don't bother to fully debug their implementation - as soon as it boots XP, they ship it, and work around any BIOS bugs that surface later in the drivers - unfortunately for Linux, they don't document these bugs so it's often a lengthy trial and error process.
I read the Supreme Court ruling and it seems like you are only required to tell the cop who you are - they can't require you to produce ID because you're not required to have one.
Actually that website does not even say what the outcome of the case was - all the text on the site was clearly written before the decision, and the PDF link to the decision is broken. So what happens if a cop asks for my ID and I don't have one? There's no law requiring me to have a passport or a drivers license, so how can I be jailed for refusing to produce one?
I don't think the cops in most places are allowed to pick you up off the street for no reason and throw you in jail if you can't produce ID. If this is allowed we already have a "Papers please" society.
The best advice is to just look at /usr/src/linux/drivers/net/wireless/Kconfig, and pick a device with one of the chipsets listed in there.
The telecoms seem to want to evolve Linux in the direction of Solaris, which has traditionally been the platform of choice for many of these apps. Lots of it is realtime stuff - robust mutexes, priority inheritance, NPTL, reliable RT scheduling, fully preemptible kernel, etc which also benefit multimedia and games.
Recently Linux has become a better soft RT platform than Windows and is creeping up on OSX, Irix, and Solaris.
The Coverity checker is quite useful and has been used to fix dozens, possibly hundreds of bugs in the Linux kernel.
With the work being done in the -rt patchset Linux is quickly developing into a world class RTOS. Today Linux runs the desktops, the next generation of rovers could run Linux for the control systems.
Re: the stability of 2.6, a lot of the increased churn is necessary if Linux wants to be viable on the desktop. Lots of key features that desktop users expect to Just Work are still not 100%, like wireless and suspend on laptops.
Servers are boring, if you want a rock-solid server just run Linux 2.4.
While they are more miss than hit these days, come on, you have to admit the "Future Drama" episode was brilliant.
Stupid flounders...
So it sounds like the Linux Bluetooth stack (and all references on the page) just need to be replaced with "IEEE 801.15" (or ieee80115 or whatever).
"Drivers should be in the kernel if A. at least one of their primary clients exists in the kernel, e.g. disk controller drivers, B. they service a large number of clients directly (e.g. /dev/random), C. real-time performance is critical to the correct operation of the device (e.g. audio/video)."
I disagree with C. It's a common misconception that you can't do realtime in userspace but this is really a Windows limitation. This leads to horrors like vendors implementing an audio synthesizer as a kernel driver. A decent OS like Linux or MacOS can do realtime in userspace just fine.
The development versions of Audacity actually are compatible with JACK. The issue is that Audacity relies on PortAudio, and PortAudio v19 which adds JACK support is not released yet. So this will be ready real soon now.
Filthy ice rats...
This doesn't stop spam, but it makes sure that no one can forge an address from your domain, unless it was really sent from your domain.
Better known as a joe job. Damn, if this succeeds, my domain name won't be funny for too long... Lee
But if you go a few blocks away, you see scenes like I did [amazing.com]. All these pictures were taken on what would be prime real estate in any other country, a block or less from the Malecon, the giant seawall that faces the ocean and is a major gathering spot for Cubans.
Obviously you don't get out of the suburbs much. There are worse slums than that in Philly or NYC.
I'm one of those testers AM talks about. Please see this. I haven't been paid squat.
If someone wants to help, rather than just talk about it, I accept PayPal donations via rlrevell at joe-job.com.
I'll take the complete lack of any followup to mean that no one on slashdot reads linux-kernel anymore. The hard realtime (yes it is hard RT, check the list archives) capabilities that have been developed for Linux lately are a much more interesting story than this. In addition to the usual suspects (pro audio users, Montavista etc), people doing defense, physics, and high end simulation work have expressed interest and contributed code or testing.
No one in their right mind would deploy the RT kernel for anything mission critical yet, but it's coming along nicely. It can (deterministically!) do about 50 usecs from interrupt to running user code (RTLinux can do 15 or so).
WaveRT is a pretty cool Longhorn feature: http://www.microsoft.com/whdc/device/audio/wavertp ort.mspx
This will supersede the 5 different low latency audio APIs that people unfortunate enough to use Windows have to deal with.
Of course Linux users already know we have a better system with ALSA and JACK.