Applying updates directly on production ?!
This should not be done on whatever OS. It's not the "Windows way", it's the "stupid way".
That said, even under the hypothesis that the argument is not flawed then doing things the proper way would have taken too much time, leading Linux to a greater TCO anyway.
What's missing from the entry is the only meaningful study : that there is no absolute best operative system, and every single case is a different story which should be deeply analyzed by itself.
Re:"Several posts" on a few boards = "very" unstab
on
Xbox 360 Very Unstable
·
· Score: 1
No reason to do otherwise. If your program tramples memory or whatever, it crashes even in user mode. Having applications in user mode is to protect them from *other* applications behaving badly, not to protect from themselves behaving badly.
Since consoles run a single application at a time, there is pratically no reason to separate user and kernel mode. Also in comparison with other consoles (on the PS2 for example you directly program DMA transfer and packets!) at least the xbox shines in hardware abstraction.
Apart from the reasons you'll find in the other replies, here are more :
- people needing (or thinking they need) ActiveX support
- people using IE only intranets
- people afraid/lazy/uninformed/etc to try FF just because they *THINK* their intranet is IE only or has ActiveX
Then they would enforce 64bit desktop software.
They are only enforcing 64bits for *future* *server* software. This hardly can be blamed considering those software will be released at least in 2007 and all x86 server processors since almost an year are x64 compatible.
Sure, Athlon XPs, Pentium M, some Pentium 4 and Semprons do not run x64. I really can't see a reason to run Exchange 12 on an Athlon XP in 2007.
And that percentage of organization in the kernel changelog is good to kernel development, because a professional paid for his work by an organization will do what he is said to do, particularly those kind of works noone else will do because they're boring.
A problem with most (small, not the kernel of course;)) open source projects is that no developers will do some of the essential, but boring , not exciting and unpopular tasks. Then, since someone has to do those tasks, the admin does all of them, doing none of the exciting tasks. The admin burns out and the project gets dropped to another developer. Often this spirals down to the death of the project.
And what other people like you fail to acknowledge is that the whole set of your programming needs are something like 0.001% of the programming needs of the world.
Yes you can build a library of handy functions in C just like you can in C++ or Java. I say more, for a small library C is the best language. For programs heavy in user interface, for programs going over the million lines of code with 100 developers working on them, for programs which operates heavily on DBs, C is not the best language. Eventually, realize that almost all of the rest of the world is not programming libraries for embedded devices but financial applications, graphic engines, complex database report apps, web applications, network applications with security issues, plugins for other applications, internal tools, etc. Yes you can do them all in C. In assembler too.
Also saying that C++/Java/C# are not improvements over C is plainly wrong. You can say that they are not improvements over C *for you*. I, for example, produce roughly double the code with considerably less errors when working in C++ rather than C - and whether you like it or not, it all boils down to productivity; with the time you saved you can optimize the program where it really matters, or get into the market a year before everyone else, or simply save a bunch of money, your choice.
6. Feeding generations of inept developers. check. [re: C#, anything.net, VB,...]... I wonder why anything.net feeds generations of inept developers and python or perl or java or any other language don't.
Oh let's return to the good old days where programmers had two big keys with 0 and 1 written on them and programmed opcodes like playing bongos..
>> Obviously not, if a third party change to the registry can break IE, IE has some major problems...
Uh then Linux has major problems! If a third party randomly changes the etc/passwd file I cannot logon anymore.. Apache too! it seems to dislike random bytes in httpd.conf..
Let's take that overwriting the book takes, let's say 5 weeks. Writing a text on a new paper takes, instead 3 weeks. Let be X the time to create paper from sheeps' skins.
If X is less than 2 weeks, there is no reason at all to overwrite. Even if he wanted to destroy Arhimedes work, he would have burned it in less than an hour, than created new skin and wrote whatever he wrote. Why didn't he burned it down ? Probably because X is >= 2 weeks.
If X is more than 2 weeks, you don't have a point.
If X is exactly 2 weeks, and the monk figured that this method was effectively faster saving him an hour, then the genius is the monk, not Archimedes:)
Protected CDs have quite a few problems on computer drives, MP3 cd players and they absolutely won't work on many car audio systems.
Oddly enough, some people are still buying them. And above all, so many people aren't aware of this fact, until they try to put that CD in their car audio player..
And you too completely ignore the whole point. good linux zealot.
The point is I strongly believe it's wrong for a Linux distro to have so much (if any!) bundled applications.
Think of the implications:
1) Adobe will have much less chance of making money on the Linux market, because the Gimp is preinstalled on so many machines. This breaks the 3rd party market on Linux machines. This also has the implication that 3rd party developers (e.g. Adobe/MacroMedia, discreet, etc) have less reasons to support free OSs and more and more reasons to support MS only and eventually Apple.
2) MS will never be able to compete with that just because (rightly) it cannot bundle applications in the OS.
3) Contradicting #2, MS could eventually bundle a good graphics application in the OS. Whenever Adobe sue them, they'll go in court and say : "hey every other OS have a graphics app in the OS.. this GIMP thingy.. Why they can and we cannot ?"
--
If you doubt bundling is a problem, answer this question..
How many people buy Opera on Linux when they have Firefox, Konqueror, Galeon, Mozilla and another 5 o 6 browsers all preinstalled ? At least, on Windows, IE is so fu**ed up that someone will buy it anyway..
Given infinite money, all software is free.. Hell if I really want 3d graphics support for Windows 3.0 and pine as the default email client and bash as the only shell, I'd just buy Microsoft:)
Actually you are free to use OpenOffice.Org under Windows too. And Firefox and Opera. And Thunderbird, and Pine.
Application choice has nothing to do with the OS.
>> As a first, they could make Win 95 true multiuser system
Well, it's true that Win95 was indeed crappy, however it had to run with the best possible compatibility on a 386 with 4mb (or 8 ? anyway..) of ram to be competitive. Also, having that kind of user management on Win95 would have trampled NT 3.5 / NT4 market which at the time were weak. And really at the time there was almost no reason at all to do that.
For your other example we are on the same opinion. Microsoft is to blame because they don't enforce (or at least suggest!) that normal users run as.. well normal users. In part, it's also to blame because having this policy of suggesting the "always admin" approach, third party programmers feel the right to do what they want with the registry, processes and other reserved resources.
Truth to be said, 99% of the programs which require admin rights to be run, are copy protected (and the copy protection is the part which requires that rights). For example Armadillo protection stores activation times, expiration data, etc as a fake OCX in the registry and in some key in HKEY_LOCAL_MACHINE. Both are admin only (and risky!) operations, used for the most stupid of the purposes.
Because XPSP2 recv Buffers are limited to 8KB. Every OS has a size for those buffers, you have just discovered the XPSP2 size, congratulations.
Every other OS has a limit on that buffer, and I guess for every OS it is configurable in some way (in Windows there is some remote key in the registry).
Wether this is a good or bad choice, it shouldn't bother you if you are writing internet applications, since only a few apps really need raw socket access.
Slashdot Mirror
Applying updates directly on production ?! This should not be done on whatever OS. It's not the "Windows way", it's the "stupid way". That said, even under the hypothesis that the argument is not flawed then doing things the proper way would have taken too much time, leading Linux to a greater TCO anyway. What's missing from the entry is the only meaningful study : that there is no absolute best operative system, and every single case is a different story which should be deeply analyzed by itself.
No reason to do otherwise. If your program tramples memory or whatever, it crashes even in user mode. Having applications in user mode is to protect them from *other* applications behaving badly, not to protect from themselves behaving badly.
Since consoles run a single application at a time, there is pratically no reason to separate user and kernel mode. Also in comparison with other consoles (on the PS2 for example you directly program DMA transfer and packets!) at least the xbox shines in hardware abstraction.
Apart from the reasons you'll find in the other replies, here are more : - people needing (or thinking they need) ActiveX support - people using IE only intranets - people afraid/lazy/uninformed/etc to try FF just because they *THINK* their intranet is IE only or has ActiveX
Then they would enforce 64bit desktop software. They are only enforcing 64bits for *future* *server* software. This hardly can be blamed considering those software will be released at least in 2007 and all x86 server processors since almost an year are x64 compatible. Sure, Athlon XPs, Pentium M, some Pentium 4 and Semprons do not run x64. I really can't see a reason to run Exchange 12 on an Athlon XP in 2007.
Ok so every EULA will contain "this software may not even do what's meant for and may contain bugs" (which is essentially what they already say).
Now that the candy has the poison warning on it, is anything changed ?
If sources are enough to save OSS from liability, EULAs are even better.
The grandparent says "Windows will get hacked for sure" and is modded flamebait.
The parent says "If patched Windows will not get hacked" and is modded flamebait.
Maybe they are both just opinions ?
And that percentage of organization in the kernel changelog is good to kernel development, because a professional paid for his work by an organization will do what he is said to do, particularly those kind of works noone else will do because they're boring.
;)) open source projects is that no developers will do some of the essential, but boring , not exciting and unpopular tasks. Then, since someone has to do those tasks, the admin does all of them, doing none of the exciting tasks. The admin burns out and the project gets dropped to another developer. Often this spirals down to the death of the project.
A problem with most (small, not the kernel of course
And what other people like you fail to acknowledge is that the whole set of your programming needs are something like 0.001% of the programming needs of the world.
Yes you can build a library of handy functions in C just like you can in C++ or Java. I say more, for a small library C is the best language. For programs heavy in user interface, for programs going over the million lines of code with 100 developers working on them, for programs which operates heavily on DBs, C is not the best language. Eventually, realize that almost all of the rest of the world is not programming libraries for embedded devices but financial applications, graphic engines, complex database report apps, web applications, network applications with security issues, plugins for other applications, internal tools, etc. Yes you can do them all in C. In assembler too.
Also saying that C++/Java/C# are not improvements over C is plainly wrong. You can say that they are not improvements over C *for you*. I, for example, produce roughly double the code with considerably less errors when working in C++ rather than C - and whether you like it or not, it all boils down to productivity; with the time you saved you can optimize the program where it really matters, or get into the market a year before everyone else, or simply save a bunch of money, your choice.
6. Feeding generations of inept developers. check. [re: C#, anything .net, VB, ...] ... I wonder why anything .net feeds generations of inept developers and python or perl or java or any other language don't.
Oh let's return to the good old days where programmers had two big keys with 0 and 1 written on them and programmed opcodes like playing bongos..
>> Obviously not, if a third party change to the registry can break IE, IE has some major problems...
Uh then Linux has major problems! If a third party randomly changes the etc/passwd file I cannot logon anymore.. Apache too! it seems to dislike random bytes in httpd.conf..
The key NS8 touches is
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Internet Explorer\Plugins\Extension
It seems to me it's reserved to IE and if a third party application (NS8) breaks IE changing it, it's the third party fault.
Then, of course, IE has major problems. But this is not the case.
Let's take that overwriting the book takes, let's say 5 weeks. Writing a text on a new paper takes, instead 3 weeks. Let be X the time to create paper from sheeps' skins. If X is less than 2 weeks, there is no reason at all to overwrite. Even if he wanted to destroy Arhimedes work, he would have burned it in less than an hour, than created new skin and wrote whatever he wrote. Why didn't he burned it down ? Probably because X is >= 2 weeks. If X is more than 2 weeks, you don't have a point. If X is exactly 2 weeks, and the monk figured that this method was effectively faster saving him an hour, then the genius is the monk, not Archimedes :)
Protected CDs have quite a few problems on computer drives, MP3 cd players and they absolutely won't work on many car audio systems.
Oddly enough, some people are still buying them. And above all, so many people aren't aware of this fact, until they try to put that CD in their car audio player..
is that Ballmer is
50% Wrong : Google in 2010 will still be there.
50% Right : Google in 2010 will probably be something radically different from today.
[after the obligatory discussion over copyright infringment and theft differences...]
I can rent a car for 50$/day.
So I guess I can steal Mark Cuban's car and if I accidentally get caught after a week I simply owe him 350$, right ?
Infact RIAA will not sue you if you SELL your original CD to me, of course you should not have a copy of it.
Don't know how it is in the USA, but here in Italy they are infact two different crimes, with very different fines.
MSAV is coming back!
Will Longhorn have InterLnk, MSD and DoubleSpace ?
getting wet for electronic devices is a dangerous thing.
And you too completely ignore the whole point. good linux zealot.
:
The point is I strongly believe it's wrong for a Linux distro to have so much (if any!) bundled applications.
Think of the implications
1) Adobe will have much less chance of making money on the Linux market, because the Gimp is preinstalled on so many machines. This breaks the 3rd party market on Linux machines. This also has the implication that 3rd party developers (e.g. Adobe/MacroMedia, discreet, etc) have less reasons to support free OSs and more and more reasons to support MS only and eventually Apple.
2) MS will never be able to compete with that just because (rightly) it cannot bundle applications in the OS.
3) Contradicting #2, MS could eventually bundle a good graphics application in the OS. Whenever Adobe sue them, they'll go in court and say : "hey every other OS have a graphics app in the OS.. this GIMP thingy.. Why they can and we cannot ?"
--
If you doubt bundling is a problem, answer this question..
How many people buy Opera on Linux when they have Firefox, Konqueror, Galeon, Mozilla and another 5 o 6 browsers all preinstalled ?
At least, on Windows, IE is so fu**ed up that someone will buy it anyway..
Because well, it always all come down to money.
:)
Given infinite money, all software is free.. Hell if I really want 3d graphics support for Windows 3.0 and pine as the default email client and bash as the only shell, I'd just buy Microsoft
Actually you are free to use OpenOffice.Org under Windows too. And Firefox and Opera. And Thunderbird, and Pine. Application choice has nothing to do with the OS.
>> As a first, they could make Win 95 true multiuser system
Well, it's true that Win95 was indeed crappy, however it had to run with the best possible compatibility on a 386 with 4mb (or 8 ? anyway..) of ram to be competitive. Also, having that kind of user management on Win95 would have trampled NT 3.5 / NT4 market which at the time were weak. And really at the time there was almost no reason at all to do that.
For your other example we are on the same opinion. Microsoft is to blame because they don't enforce (or at least suggest!) that normal users run as.. well normal users. In part, it's also to blame because having this policy of suggesting the "always admin" approach, third party programmers feel the right to do what they want with the registry, processes and other reserved resources.
Truth to be said, 99% of the programs which require admin rights to be run, are copy protected (and the copy protection is the part which requires that rights).
For example Armadillo protection stores activation times, expiration data, etc as a fake OCX in the registry and in some key in HKEY_LOCAL_MACHINE. Both are admin only (and risky!) operations, used for the most stupid of the purposes.
Because XPSP2 recv Buffers are limited to 8KB.
Every OS has a size for those buffers, you have just discovered the XPSP2 size, congratulations.
Every other OS has a limit on that buffer, and I guess for every OS it is configurable in some way (in Windows there is some remote key in the registry).
If ICQ (or any other app) doesn't run correctly without admin priviledges is not a M$ fault rathern than a Mirabilis (or whoever writes ICQ now) one.
Anyway if you need to lie on these things, good luck with your relationship.
Wether this is a good or bad choice, it shouldn't bother you if you are writing internet applications, since only a few apps really need raw socket access.