How about C? How about Pyrex? Writing most of the product in Python and just implementing the performance-critical parts in Pyrex or C is much easier than writing it with C++ and creating bindings for Python.
It would also be just as fast, because almost always, less than 10% of the code runs more than 90% of the time.
Ok, misunderstood me it is. Pretty much ANY non-embedded CPU that provides memory protection provides memory segments (actually, they offload it to the MMU), where memory segments are defined as an area for data or code. Early systems had little memory to spare (Or was it because VM hadn't been invented yet? I forget), so they used segment extents to define large chunks of memory. Intel took this route with the 286/386, but made the scheme brain damaged in an attempt to be backward compatible with previous versions.
Really? Can you name 3 or 4 processors that commonly use MMU's with segmentation support? MMU's classically only support simple page tables.
In fact, most OSes don't support segmentation also because of the fact it is not available on other systems and thus not portable.
Later system designs had more memory available to them and thus integrated Paging and Segmentation to function along the same lines. i.e. Each memory page is told what type of segment it is so that data cannot overwrite code. This scheme is far simpler to use than the extent scheme, but provides protection against data overwriting code.
What the heck does having more memory have to do with the introduction of page tables? Page tables were introduced, because:
Segmentation is far too complex to use to be useful
Swapping out whole segments is a limited way to implement virtual memory, and paging is a great way to do it
Page tables give you all of the functionality of segmentation, without all the complexity, and with much finer granularity
The problem is that x86 OSes ALWAYS map the code and data segments over top of each other. Guess what happens then?
The reason is that making segmentation transparent to the user-space applications is difficult, as it would require one of two things.
Option A: All pointers become far pointers automatically, and break several assumptions:
A lot of code assumes its ability to use pointer arithmetic in a linear way
The sizeof(void *) is often assumed to be sizeof(int)
(Option A thus unusable as it would break too much code).
Option B: Lets always point to code in CS and always point to data in DS, and distinguish between pointers into the data segment and pointers into the code segment in the source code itself (function pointers always point into CS, data pointers always point into DS).
This is a problem with existing binaries and compilers:
Many put read-only strings in the code segment.
Many make assumptions about the flat memory model that would break (having relative offsets instead of simple offsets)
It is much much simpler to use proper page tables that support all of the access control attributes we want on each page.
BSD licenses are do not allow users more freedoms. The users have the same freedoms with both the GPL and BSD. Those freedoms are: To view or change the source, redistribute it and use it in any way they please.
The difference is not in the freedoms the license gives the users, but the powers. There is a meaningful distinction. The GPL disallows the powers of taking away the very freedoms both of the licenses originally give.
Stallman came against the ceasing of Freedom of software. He does not claim to have invented it. He claims that it was being endangered by the move to closed-source control over the entire computing world, and created a Free alternative upon which a huge body of Free Software builds.
"Freeware" without the source is destined to rot and disappear. Free Software with source, is maintained as long as there is interest, and its code and ideas can be reused in other projects.
Stallman replaced "Freeware" with the huge amounts of Free Software nowdays. And now, the users have the freedom, too. Even if you are arguing that more software was free of charge back then, you are being rediculous.
Many would agree that Richard Stallman, with his GNU manifest, has in fact initiated the Free Software movement, that later also yielded the opensource movement. He has also inspired Linus to make Linux free, if you note the fact Linus has used Richard's GPL license on Linux.
In other words, without Richard, we'd be stuck in the 80's or early 90's where all software is commercial crap, shareware crap, and all of the power over computer users would belong to big companies - forever locking them in and controlling their computer usage.
I'd say he changed the world more than say, a random prime minister of some country did.
That's why it is a threat, and not a different kind of problem, yet.
The American administration would probably be nice enough to not to use the authoritorian powers granted to it by law unjustfully, but that does not mean a future administration will not exploit those. That's what "threat" basically means here.
Well, if browsers get too fancy and usable, then they might serve as a portable application platform.
A portable application platform is a Windows killer. Microsoft does not like Windows killers, so it wants to make sure it controls what browser people use, to disallow it to become a portable application platform.
How often do you encounter something out of the central repository?
Its a lot easier to install things via synaptic, than it is to go through dozens of untrusted web pages, download untrusted executables, run them and answer a bunch of annoying questions.
I barely have anything installed not from the repository, and in those cases, there's usually an installer executable or simple installation instructions.
Sure, many of these installers are a lot less convenient than the Windows installers, but for the 99.99% of the programs I install, I do it via apt-get or synaptic and that is incomparably better than Windows - and much safer too.
I also get the added bonus of auto-upgrades to all the software I install, which is something you can only hope for when using Windows.
Copyright and licensing issues serve as barriers to good technical systems. For example, there is no way to make a central repository of upgrades for the entire system. Microsoft is able to do this with its own products (thus Windows update) but the rest of the software has to rot.
Yet another reason that Free Software is required. It is the freedom that allows organizations like Debian to make a central repository of all the available software. This allows the superior technical solution of a central tool (such as apt) that can upgrade everything.
In the non-free world, huge amounts of resources must be wasted and the central upgrade code in apt has to be written all over again for every piece of software separately, not to mention maintaining separate upgrade servers, etc.
Actually a smart Journaling filesystem can rely only on the order of writes to the disk remaining faithful to the order of write commands, in order to avoid corruption.
Functions are very different containers of code than small blocks in a function.
The conventions about interactions between functions are different. Functions have a name. Functions have clearly different scopes with their input/output well-defined.
Functions are reusable. Functions are in general understandable separately while goto blocks have to be understood all at once.
Dont compare Linux with Windows cause Linux is just a kernel.
Compare Debian to Windows and then you realize that there's no fragmentation at all and in fact Debian is far more consistent than Windows (who cares that there are other distributions?:P)
One of the things holding Linux back in general is the splintering in the distros. I can't install software for one distro on another distro without having some technical background. On windows, You put the CD, autorun starts the install, you blindly click yes a few times, and in a couple minutes you have an icon of your new program, game, whatever on your desktop. When linux can acomplish this maybe then I will get my grandmother to upgrade to it. Until then it is more hassle then reward.
As an argument for well engineering - it isn't so. A very badly engineered program can still have a decently engineered UI, which is what a lot of Microsoft applications are.
As for your comparison of installations in Linux and Windows, its quite a bad example... When I can install any program I want by searching for it in my package manager, clicking on it and clicking install (getting the download and dependencies automatically) in Windows, tell me cause I've been doing that in Debian for years.
In Windows, I need to have a CD, a direct URL or such to get an application, which means there's a different way to install every application. It also means I cannot upgrade all my apps automatically as I can with Debian.
Then there's the issue of "This game requires Direct X 8, download it from...". Unmet dependencies are also handled automatically by the "install" button in my package manager.
The purpose of copyrights is to advance science and useful arts, not to reward authors.
If rewarding authors for that purpose is required, then they will be rewarded.
Copyrights on binaries however, reward authors while stifling the progress of science and useful arts.
It encourages people to create secretly-operating software that helps them get revenue but does not inspire new works, does not enter the public domain and does not help anyone else in the long run.
It is rediculous that binaries are copyrightable and the law that allows it is actually quite new (from the late 70's) and should be reverted.
Enough is enough - copyrighting binaries must stop.
Slashdot Mirror
It can, but under some limitations.
I tried doing it, but in Windows, you cannot really output binary to stdout, because character 26 is considered EOF (In 2005!) and breaks the pipe.
Incredible.
(Yes, I used the text pickle format, but it sucked!)
Some C library functions that operate on strings with checking lengths should also not be used:
strncpy - does not guarantee the terminating NULL and fills the dest string with zeroes if it is short (bad performance)
strncat - does not guarantee the terminating NULL and the 'n' argument is very confusing and encourages buggy calls.
Use strlcpy/strlcat instead (oh wait, the GNU libc maintainer refuses to put them in...)
Well, C++ developers usually use some sort of reference counting schemes.
In such a case, they suffer from the same predictability problems of GC systems.
Also, a lot of GC'd languages have stack-based optimizations to recognize they can destroy an object that's only used in a function locally.
Also, a lot of GC'd languages allow you to control exactly when the GC runs, which allows for predictable performance as well.
There is no choice but to use C++.
How about C? How about Pyrex?
Writing most of the product in Python and just implementing the performance-critical parts in Pyrex or C is much easier than writing it with C++ and creating bindings for Python.
It would also be just as fast, because almost always, less than 10% of the code runs more than 90% of the time.
Really? Can you name 3 or 4 processors that commonly use MMU's with segmentation support? MMU's classically only support simple page tables.
In fact, most OSes don't support segmentation also because of the fact it is not available on other systems and thus not portable.
Later system designs had more memory available to them and thus integrated Paging and Segmentation to function along the same lines. i.e. Each memory page is told what type of segment it is so that data cannot overwrite code. This scheme is far simpler to use than the extent scheme, but provides protection against data overwriting code.
What the heck does having more memory have to do with the introduction of page tables? Page tables were introduced, because:
Segmentation is far too complex to use to be useful
Swapping out whole segments is a limited way to implement virtual memory, and paging is a great way to do it
Page tables give you all of the functionality of segmentation, without all the complexity, and with much finer granularity
The problem is that x86 OSes ALWAYS map the code and data segments over top of each other. Guess what happens then?
The reason is that making segmentation transparent to the user-space applications is difficult, as it would require one of two things.
Option A: All pointers become far pointers automatically, and break several assumptions:
A lot of code assumes its ability to use pointer arithmetic in a linear way
The sizeof(void *) is often assumed to be sizeof(int)
(Option A thus unusable as it would break too much code).
Option B: Lets always point to code in CS and always point to data in DS, and distinguish between pointers into the data segment and pointers into the code segment in the source code itself (function pointers always point into CS, data pointers always point into DS).
This is a problem with existing binaries and compilers:
Many put read-only strings in the code segment.
Many make assumptions about the flat memory model that would break (having relative offsets instead of simple offsets)
It is much much simpler to use proper page tables that support all of the access control attributes we want on each page.
Using a certain meta-format does not make it open. Using an open format does. We still have to wait and see what format they use.
For those with difficulty understanding what I mean, consider their new format is:
WordFile.xml:
<WordFile>
big-hex-dump-of-old-word-format
</WordFile>
But copyright is in itself a restriction of freedom.
The BSD allows use of copyright.
The GPL does not allow it.
Thus the GPL cancels restriction!
BSD licenses are do not allow users more freedoms. The users have the same freedoms with both the GPL and BSD. Those freedoms are: To view or change the source, redistribute it and use it in any way they please.
The difference is not in the freedoms the license gives the users, but the powers. There is a meaningful distinction. The GPL disallows the powers of taking away the very freedoms both of the licenses originally give.
So how is GNU/Linux not "truly free"?
Stallman came against the ceasing of Freedom of software. He does not claim to have invented it. He claims that it was being endangered by the move to closed-source control over the entire computing world, and created a Free alternative upon which a huge body of Free Software builds.
Back then, pretty much anything you downloaded was *really* Free (not any of the GPL restrictions.)
The GPL restrictions are restrictions of powers, not freedoms. Thus, the GPL is *really* free.
"Freeware" without the source is destined to rot and disappear.
Free Software with source, is maintained as long as there is interest, and its code and ideas can be reused in other projects.
Stallman replaced "Freeware" with the huge amounts of Free Software nowdays. And now, the users have the freedom, too.
Even if you are arguing that more software was free of charge back then, you are being rediculous.
Many would agree that Richard Stallman, with his GNU manifest, has in fact initiated the Free Software movement, that later also yielded the opensource movement. He has also inspired Linus to make Linux free, if you note the fact Linus has used Richard's GPL license on Linux.
In other words, without Richard, we'd be stuck in the 80's or early 90's where all software is commercial crap, shareware crap, and all of the power over computer users would belong to big companies - forever locking them in and controlling their computer usage.
I'd say he changed the world more than say, a random prime minister of some country did.
"Stallman very pleased with Linux remaining GPL"
"Stallman very pleased with his dinner"
"Stallman took a crap"
"Stallman very pleased with his love life"
(Okay, the last one is newsworthy)
That's why it is a threat, and not a different kind of problem, yet.
The American administration would probably be nice enough to not to use the authoritorian powers granted to it by law unjustfully, but that does not mean a future administration will not exploit those. That's what "threat" basically means here.
Well, if browsers get too fancy and usable, then they might serve as a portable application platform.
A portable application platform is a Windows killer. Microsoft does not like Windows killers, so it wants to make sure it controls what browser people use, to disallow it to become a portable application platform.
That's one theory anyhow.
How often do you encounter something out of the central repository?
Its a lot easier to install things via synaptic, than it is to go through dozens of untrusted web pages, download untrusted executables, run them and answer a bunch of annoying questions.
I barely have anything installed not from the repository, and in those cases, there's usually an installer executable or simple installation instructions.
Sure, many of these installers are a lot less convenient than the Windows installers, but for the 99.99% of the programs I install, I do it via apt-get or synaptic and that is incomparably better than Windows - and much safer too.
I also get the added bonus of auto-upgrades to all the software I install, which is something you can only hope for when using Windows.
Copyright and licensing issues serve as barriers to good technical systems. For example, there is no way to make a central repository of upgrades for the entire system. Microsoft is able to do this with its own products (thus Windows update) but the rest of the software has to rot.
Yet another reason that Free Software is required. It is the freedom that allows organizations like Debian to make a central repository of all the available software. This allows the superior technical solution of a central tool (such as apt) that can upgrade everything.
In the non-free world, huge amounts of resources must be wasted and the central upgrade code in apt has to be written all over again for every piece of software separately, not to mention maintaining separate upgrade servers, etc.
Self-upgrading programs are a waste of resources.
Actually a smart Journaling filesystem can rely only on the order of writes to the disk remaining faithful to the order of write commands, in order to avoid corruption.
Allowing sync commands helps but is not required.
Functions are very different containers of code than small blocks in a function.
The conventions about interactions between functions are different. Functions have a name. Functions have clearly different scopes with their input/output well-defined.
Functions are reusable. Functions are in general understandable separately while goto blocks have to be understood all at once.
and he is now paying the price.
Stallman is right.
or pressing an 'accept' button
Does pressing the 'accept' button have any legal bearing?
I think it doesn't and I think it is impressive that by using it in so many places they managed to convince people that it does.
The distribution stuff is silly
:P)
Dont compare Linux with Windows cause Linux is just a kernel.
Compare Debian to Windows and then you realize that there's no fragmentation at all and in fact Debian is far more consistent than Windows (who cares that there are other distributions?
One of the things holding Linux back in general is the splintering in the distros. I can't install software for one distro on another distro without having some technical background. On windows, You put the CD, autorun starts the install, you blindly click yes a few times, and in a couple minutes you have an icon of your new program, game, whatever on your desktop. When linux can acomplish this maybe then I will get my grandmother to upgrade to it. Until then it is more hassle then reward.
...". Unmet dependencies are also handled automatically by the "install" button in my package manager.
As an argument for well engineering - it isn't so. A very badly engineered program can still have a decently engineered UI, which is what a lot of Microsoft applications are.
As for your comparison of installations in Linux and Windows, its quite a bad example... When I can install any program I want by searching for it in my package manager, clicking on it and clicking install (getting the download and dependencies automatically) in Windows, tell me cause I've been doing that in Debian for years.
In Windows, I need to have a CD, a direct URL or such to get an application, which means there's a different way to install every application. It also means I cannot upgrade all my apps automatically as I can with Debian.
Then there's the issue of "This game requires Direct X 8, download it from
So.. bad example.
The purpose of copyrights is to advance science and useful arts, not to reward authors.
If rewarding authors for that purpose is required, then they will be rewarded.
Copyrights on binaries however, reward authors while stifling the progress of science and useful arts.
It encourages people to create secretly-operating software that helps them get revenue but does not inspire new works, does not enter the public domain and does not help anyone else in the long run.
It is rediculous that binaries are copyrightable and the law that allows it is actually quite new (from the late 70's) and should be reverted.
Enough is enough - copyrighting binaries must stop.
Microsoft's well-engineered products
Which ones? I mean, is there a single well-engineered product from Microsoft?
the commoditization of their core markets and the parasites eating their way in from the internets
Parasites? who?
I personally look forward to the inevitable death of copyright as it became a barrier to progress instead of an incentive towards it.