Or lucky... Or just careful. Remember, the failure rate is an average... Some people will keep their drives well cooled, some will let them overheat badly... Once a device gets into consumer's hands it's fate is in their hands, so individual cases are not really worth looking at, you need a large sampling to work out the average.
I'm surprised the failure rate of mobile phones isn't a lot higher, people generally carry them around all day every day.. They get dropped, they get wet, they get sat on, exposed to shocks etc... A mobile phone is probably subject to much worse treatment than almost any electronic gadget these days... I try to take care of mine, but i've still managed to drop it a few times.
Err, so your sun servers run a non sun OS and you complain that sun don't support the framebuffers? I'm sure if you put Solaris on your sun servers it would support the framebuffers just fine. What type of framebuffers are you having issues with?
On the other hand, they are servers, why the hell would you want to use a framebuffer on them? Serial is the only way to manage a server since then you don't need to set foot inside the datacenter. All of my servers run from serial console, and most don't have any kind of framebuffer hardware fitted at all. The less time i have to spend in an environment designed for machines the better.
Infact, i wouldn't consider any machine that required local physical interaction to install or recover an OS as suitable for use as a server.
Yes, i agree with the lack of drivers... But on Sparc systems the opposite is true, Solaris supports everything out of the box with zero hassle. In that respect it's a bit like OSX, you *can* run it on commodity hardware, but don't expect the smooth experience you get on hardware designed to run it.
Well you could say the same about Linux, but Linux can be tuned heavily at kernel build time to suit your needs, it's possible to turn off desktop oriented features like sound, 3d graphics etc and enable server features like hotplug cpu/mem, support for large numbers of cpus and huge quantities of ram, serial consoles, scsi disks and raid controllers etc...
Kernels compiled for a server oriented distribution typically support more cpus and/or more ram because they are more likely to need it... Enabling that support on a desktop system where it won't be used will just decrease performance, so it's off by default. There's nothing stopping you turning it back on if you want it.
Apple provide an OS which is highly compatible with unix like systems such as linux, bsd and solaris... But if that's all it was, it would be pointless...
What differentiates OSX is the whole package, you get premium hardware which runs an OS and set of apps specifically designed and thoroughly tested to work together on the exact hardware you're using.
What differentiates MS is compatibility with the existing applications and other MS products you may be tied in to...
If MS were to embrace a unix system, then they would have nothing to force people onto their software instead of one of the many free unix systems out there, and each small step towards standardisation chips away at the artificial lock-in they have spent so long creating. MS just aren't geared up for competing on a level field based on quality and value. Competition would massively shrink their margins, they would have to curb their aggressive and costly expansion into other markets to cut costs, and they would have to invest a lot more in improving their products... And their costs for product improvement would be a lot higher than their competitors, due to the haphazard way in which many of their products are engineered.
Commodore also created a GUI that ran on hardware which was considerably cheaper than anything MS or Apple ran on... And their whole system was better, the GUI ran well with less memory, multitasked properly and was considerably quicker.
Is there anywhere with a good compatibility database... That is, a big list of different devices somewhere that reports compatibility with various OS's, as well as the level of compatibility (ie some features not working) and any hoops you have to jump through to get it working...
And most importantly, a facility to search and find the most compatible hardware for the software you intend to run.
That way, things like MSI motherboards where you cannot update the BIOS using Linux, and the standalone versions are out of date (and make you jump through hoops like making dos boot floppies), can be avoided by people planning to use Linux...
Interestingly, the default Linux kernel includes a driver for updating the BIOS on Dell machines, I haven't tried it but it looks like a good idea.
But both Gecko and Webkit are LGPL, which allows linking to a library... The frontend to Safari is not GPL, but the actual rendering engine is LGPL. Nothing to stop MS using Webkit in exactly the same way Apple does. There's also nothing stopping you creating your own frontend to Webkit...
Or you know, a package manager like virtually every unix system has these days... Select what you want from a list, hit install and just let it do the work for you.
Or do it the Linux way... Provide a minimal OS, and let third parties create distributions of it with third party applications added, like OEMs do now but extend it to retail versions too.
Most of the Netscape code actually got junked, and Gecko was written pretty much from scratch.
IE can be traced back to the original Mosaic (of which Netscape was also a fork)...
Most of the big commercial products that got open sourced have been quite messy, if you don't intend anyone else to see your code and your working to a deadline there's very little incentive to keep it clean, just make it work before the deadline and preferably without having to work out of hours. Similarly, there will be a small number of fixed build environments, rather than a need to be portable to a huge number of slightly different systems. I seem to remember StarOffice was quite a mess when first opened up too...
Yes, that's a problem, which the acid tests are trying to address (but don't go far enough)... We need a thorough set of tests at various baselines so we can assure compatibility, and browsers can advertise which baseline they support.
Then the customers need to be more vigilant and forward thinking, if the majority of customers start demanding open specs then companies supplying hardware will fulfil their customers needs, or risk losing out to someone else who will.
You don't need to keep physical copies in stock, you just provide it for download as a set of floppy images and let the end user worry about transferring it to real floppies.
The color LCD screens let them bombard you with advertisements while you wait for the sluggish ATM to dispense your money... I also prefer the old style machines, they just work better and they're tried and tested.
On the other hand, you spent $200,000 on a piece of proprietary equipment that depends on a single piece of proprietary software? Being forced to keep win3.1 (you'l never be able to get a new copy legitimately) and a stockpile of antiquated hardware to run it on is the price you pay... Perhaps you should press for open specs and open code next time you buy a $200,000 piece of hardware, if your spending that much the company should bend over backwards to provide what you need... Or was it just a lack of forward thinking? Never thought about long term support for the equipment you bought?
Yes, using linux to save money moreso than any other reason... They also try to save money on the hardware too, so its quite often not hardware thats designed to be in a vibrating environment, and will crash during turbulence etc.
I think the flicker of mode switching comes from switching between kernel mode drivers (console) and user land drivers (X11).. On the other hand i don't seem to notice any issues, the console screen goes blank and comes back with the X11 screen, and it resyncs the new display mode if using a CRT... Tho i have in the past seen brief flashes of screen corruption when it initializes a new screenmode and some of the display data from the previous one is still in the framebuffer memory.
That, and the fact many apps would need to be rewritten... All the alternatives i've seen were considerably slower than X, probably due to the lack of accelerated drivers as you pointed out tho.
They won't care about fines... The reason the government and their contractors don't care about losing data and money is because it's not their money... The whole system is corrupt and screwed up because there is no incentive to fix it. If you give a civil servant the choice between spending 50k for a poor quality supplier who will take him out to dinner, or 10k for a better quality supplier who will do a good job but not provide the free lunch, guess who that civil servant will choose.
Slashdot Mirror
Or lucky... Or just careful.
Remember, the failure rate is an average... Some people will keep their drives well cooled, some will let them overheat badly... Once a device gets into consumer's hands it's fate is in their hands, so individual cases are not really worth looking at, you need a large sampling to work out the average.
It doesn't explain why they fail...
I'm surprised the failure rate of mobile phones isn't a lot higher, people generally carry them around all day every day.. They get dropped, they get wet, they get sat on, exposed to shocks etc... A mobile phone is probably subject to much worse treatment than almost any electronic gadget these days... I try to take care of mine, but i've still managed to drop it a few times.
The ipod touch has wifi...
Err, so your sun servers run a non sun OS and you complain that sun don't support the framebuffers?
I'm sure if you put Solaris on your sun servers it would support the framebuffers just fine. What type of framebuffers are you having issues with?
On the other hand, they are servers, why the hell would you want to use a framebuffer on them? Serial is the only way to manage a server since then you don't need to set foot inside the datacenter.
All of my servers run from serial console, and most don't have any kind of framebuffer hardware fitted at all. The less time i have to spend in an environment designed for machines the better.
Infact, i wouldn't consider any machine that required local physical interaction to install or recover an OS as suitable for use as a server.
Yes, i agree with the lack of drivers...
But on Sparc systems the opposite is true, Solaris supports everything out of the box with zero hassle. In that respect it's a bit like OSX, you *can* run it on commodity hardware, but don't expect the smooth experience you get on hardware designed to run it.
Well you could say the same about Linux, but Linux can be tuned heavily at kernel build time to suit your needs, it's possible to turn off desktop oriented features like sound, 3d graphics etc and enable server features like hotplug cpu/mem, support for large numbers of cpus and huge quantities of ram, serial consoles, scsi disks and raid controllers etc...
Kernels compiled for a server oriented distribution typically support more cpus and/or more ram because they are more likely to need it...
Enabling that support on a desktop system where it won't be used will just decrease performance, so it's off by default. There's nothing stopping you turning it back on if you want it.
Apple provide an OS which is highly compatible with unix like systems such as linux, bsd and solaris... But if that's all it was, it would be pointless...
What differentiates OSX is the whole package, you get premium hardware which runs an OS and set of apps specifically designed and thoroughly tested to work together on the exact hardware you're using.
What differentiates MS is compatibility with the existing applications and other MS products you may be tied in to...
If MS were to embrace a unix system, then they would have nothing to force people onto their software instead of one of the many free unix systems out there, and each small step towards standardisation chips away at the artificial lock-in they have spent so long creating.
MS just aren't geared up for competing on a level field based on quality and value. Competition would massively shrink their margins, they would have to curb their aggressive and costly expansion into other markets to cut costs, and they would have to invest a lot more in improving their products... And their costs for product improvement would be a lot higher than their competitors, due to the haphazard way in which many of their products are engineered.
Commodore also created a GUI that ran on hardware which was considerably cheaper than anything MS or Apple ran on... And their whole system was better, the GUI ran well with less memory, multitasked properly and was considerably quicker.
Is there anywhere with a good compatibility database...
That is, a big list of different devices somewhere that reports compatibility with various OS's, as well as the level of compatibility (ie some features not working) and any hoops you have to jump through to get it working...
And most importantly, a facility to search and find the most compatible hardware for the software you intend to run.
That way, things like MSI motherboards where you cannot update the BIOS using Linux, and the standalone versions are out of date (and make you jump through hoops like making dos boot floppies), can be avoided by people planning to use Linux...
Interestingly, the default Linux kernel includes a driver for updating the BIOS on Dell machines, I haven't tried it but it looks like a good idea.
But both Gecko and Webkit are LGPL, which allows linking to a library...
The frontend to Safari is not GPL, but the actual rendering engine is LGPL. Nothing to stop MS using Webkit in exactly the same way Apple does.
There's also nothing stopping you creating your own frontend to Webkit...
Or you know, a package manager like virtually every unix system has these days...
Select what you want from a list, hit install and just let it do the work for you.
Or do it the Linux way...
Provide a minimal OS, and let third parties create distributions of it with third party applications added, like OEMs do now but extend it to retail versions too.
Most of the Netscape code actually got junked, and Gecko was written pretty much from scratch.
IE can be traced back to the original Mosaic (of which Netscape was also a fork)...
Most of the big commercial products that got open sourced have been quite messy, if you don't intend anyone else to see your code and your working to a deadline there's very little incentive to keep it clean, just make it work before the deadline and preferably without having to work out of hours. Similarly, there will be a small number of fixed build environments, rather than a need to be portable to a huge number of slightly different systems.
I seem to remember StarOffice was quite a mess when first opened up too...
Yes, that's a problem, which the acid tests are trying to address (but don't go far enough)...
We need a thorough set of tests at various baselines so we can assure compatibility, and browsers can advertise which baseline they support.
Then the customers need to be more vigilant and forward thinking, if the majority of customers start demanding open specs then companies supplying hardware will fulfil their customers needs, or risk losing out to someone else who will.
You don't need to keep physical copies in stock, you just provide it for download as a set of floppy images and let the end user worry about transferring it to real floppies.
What does windows 3.1 make of a 30gb hard drive?
Except for XP and newer versions with online activation, which will become unusable for new installs after the activation servers are turned off...
The color LCD screens let them bombard you with advertisements while you wait for the sluggish ATM to dispense your money...
I also prefer the old style machines, they just work better and they're tried and tested.
PCI slots? Don't you mean ISA?
On the other hand, you spent $200,000 on a piece of proprietary equipment that depends on a single piece of proprietary software? Being forced to keep win3.1 (you'l never be able to get a new copy legitimately) and a stockpile of antiquated hardware to run it on is the price you pay... Perhaps you should press for open specs and open code next time you buy a $200,000 piece of hardware, if your spending that much the company should bend over backwards to provide what you need...
Or was it just a lack of forward thinking? Never thought about long term support for the equipment you bought?
Yes, using linux to save money moreso than any other reason...
They also try to save money on the hardware too, so its quite often not hardware thats designed to be in a vibrating environment, and will crash during turbulence etc.
I think the flicker of mode switching comes from switching between kernel mode drivers (console) and user land drivers (X11)..
On the other hand i don't seem to notice any issues, the console screen goes blank and comes back with the X11 screen, and it resyncs the new display mode if using a CRT... Tho i have in the past seen brief flashes of screen corruption when it initializes a new screenmode and some of the display data from the previous one is still in the framebuffer memory.
That, and the fact many apps would need to be rewritten...
All the alternatives i've seen were considerably slower than X, probably due to the lack of accelerated drivers as you pointed out tho.
They won't care about fines...
The reason the government and their contractors don't care about losing data and money is because it's not their money... The whole system is corrupt and screwed up because there is no incentive to fix it.
If you give a civil servant the choice between spending 50k for a poor quality supplier who will take him out to dinner, or 10k for a better quality supplier who will do a good job but not provide the free lunch, guess who that civil servant will choose.