"Linux is a Unix clone written from scratch by Linus Torvalds with assistance from a loosely-knit team of hackers across the Net. It aims towards POSIX compliance."
(So the POSIX spec comes close to a product spec.)
Where's the Design Document?
I guess there isn't one. No design document does not, however, imply no design. And having a design document is no guarantee of a good design. There are plenty of books and docs explaining the kernel's architecture, even if it's not written by the original designers.
Did qualified experts sign off on it?
Yes, Linus or the maintainer of the particular stable kernel series has to personally review and accept each patch. Particular subsystems often have their own maintainers who will review code before sending it to Linus.
Is there modularity?
I think Linux qualifies as reasonably modular; maybe not to the extent of microkernel systems like the HURD, but it's not bad. (Without modularity, it would be impossible to do any development with such a large number of developers.) Certainly it has loadable modules and defined driver interfaces.
Where do I download the test cases? Can I independently run the test cases?
As you know, test cases for a kernel cannot be just code, since you can't test a driver without the right hardware. But there are plenty of Linux (usually GNU/Linux) distributions to download, most of which make a pretty good stress test for the kernel. I agree this is not the same as a formal regression test, but there are some tests of that kind for filesystems and other 'software' aspects of the kernel.
But you can buy gold jewellery... it may not be so heavily marketed as 'romantic' compared to diamonds, but it can be just as expensive and that's the important thing.
Gold mining is not a cartel, and I don't think artificial gold will be made in large quantities any time soon.
Bittorrent and rsync both use chunks, but rsync can take chunks from an older version of the file that is almost what you want, compare them against the new version, and transfer only what has changed. Getting a file through Bittorrent, AFAIK, means getting all of it, even though you may get chunks out-of-order.
When will someone combine bittorrent and rsync? That sounds like the best way to upgrade from Fedora Core 2 test1 to Fedora Core 2 test2, or to update your Gentoo source tree.
If you run a web server program, and instruct it to serve files from a certain area, and place files into that area, then you have made them public. No matter if you link to them or not. The step in making the files public is in putting them on the web site; the URL is just an address. HTTP does have authentication measures which are easy to set up with servers such as Apache; don't blame others if you choose not to use them.
You can block Google with your firewall but that is no protection against sneakier robots which ignore robots.txt and don't identify themselves in the User-Agent. If you want a document to be non-public, get the web server to put a password on it or PGP encrypt it.
The trouble with any name ending 'fox' is that it makes one think of the awful Linux Fox which fortunately never had a chance of being adopted as the Linux logo. There was a Slashdot story about it at the time, but Slashdot's search box is so useless I can't find it.
Prison, on the whole, does not stop people from reoffending,
It does, while the offender is in prison. At worst, they can make life unpleasant for other prisoners. Short prison sentences may not be effective in preventing reoffending but a life sentence certainly is.
Does KDE 3.2 work well with 16 colours? 4 colours? Black and white?
Traditionally a lot of X11/Unix desktops were high resolution but only 1-bit colour; that's a bit extreme these days but 4-bit colour should be enough for a good screenshot that doesn't eat bandwidth. I fear it would get in the way of all the kewl graduated shading though.
You can't sterilize an astronaut, but if he is wearing a space suit there shouldn't be any difficulty in soaking a sponge with bleach and rubbing it all round the suit before leaving the craft. You have to sterlize the outside of the spacecraft, but that same problem is faced with robots.
ISTR that Digital had some kind of dynamic recompilation 386 emulator for the Alpha, and it was good enough that a 500MHz Alpha performed faster than a 500MHz PC for running PC software. Whatever happened to that?
Never mind the GNOME/KDE fanboy wars, I'm more interested in another holy war: plan text versus HTML mail. The info page for the mailing list Perens wrote to says 'Please use HTML email when sending to the list.' Why on earth would anyone actively want that? But he seems to have broken his own rule, since his message is plain text.
Windows 3.0 should be no problem - I ran it in software emulation on my 35MHz Archimedes, although the performance was not blazing fast. I think CPUs are now getting to the level where Win95 should be possible, after all the main factor affecting speed is available RAM.
Still, you'd need some rather cunning dynamic recompilation to get anything less than a 10x slowdown in MHz terms.
A decrease in the standard of living for the American (or $COUNTRY_X) programmers, because they can no longer charge such high prices as before. But programmers and software companies are in a tiny minority compared to the users of software and business which need to pay for it. The people who benefit from free trade are not just the Indian (or $COUNTRY_Y) programmers, but the Western businesses who are able to get what they need more cheaply, and the consumers who (assuming decent competition) get lower prices.
Free trade is just bringing together those who have something to sell - the Indian programmers - and those who want to buy - American firms needing software written. I don't see any reason for a third group to whine about this just because they were previously able to get away with charging more.
This is especially hypocritical on a site such as Slashdot, where readers depend on a steady supply of computer hardware often built in countries with lower wages than the West. In stories about video cards or RAM I don't think I have _ever_ seen any complaint about free trade reducing the price of the hardware and the lost job opportunties for Americans caused by building it in the Far East. Or think of the constant RIAA stories - stop trying to get in the way of progress, stop trying to prop up a failing business model, you don't automatically have the right to keep on getting money just because you did in the past. I know this is partly the fallacy of assuming Slashdot readers speak with one voice, but it's still worth noting.
SQLite boasts of being typeless, and having no checking that a column declared of type integer really does store integers, for example. But for me this defeats one of the main reasons to use a relational database - strong type checking and validation. SQLite also doesn't support check constraints so there is really nothing you can do to fix this.
It's not inherently insecure to allow actions to be associated with shortcuts or add other whizzy things. If the user has write access to the filesystem, why shouldn't the user set up these things for his own files?
Consider - the window manager I use (icewm) has a menu file saying what programs appear on the start menu. It is possible for a program (including a worm) running as my uid to change that menu so that clicking on 'xterm' runs something else instead. This does not mean icewm is insecure.
The problem comes from running malware with full user privileges. As long as that happens, the system will be insecure anyway whether or not there are fancy WinFS active content things.
(However, if WinFS is so stupid as to display active content for files which were created by another user or sent over email, and the design is so stupid as to allow the content full access to the machine, there is certainly a problem. The Acorn RISC OS had a severe security bug of this kind - each application directory (like an appdir on NeXT) had a file !Boot which would be executed merely on _viewing_ the directory listing. Of course this was the ideal vector for viruses.)
The point is more that it clearly can be done. As with a security hole in software - you don't ask whether anyone is currently going around exploiting it, you just get on and fix it.
The flaw is signing up to a service such as webmail has a cost in solving the captcha, but this is once per account. Then there is no cost for each message sent out, so you can send tons of messages, or at least enough to make the cost per message tiny.
Spam comes about because *sending a message is too easy*. Nothing you do that doesn't increase the cost of sending a message - whether in money, computing time, or human time - will do much to combat spam in the long run. Having captchas on account signup doesn't increase the costs of sending an individual message by much, unless each account is limited to a low volume of messages to a small number of recipients.
But you have to trust the auctioneer to keep your bid secret. If someone else knew you had entered a bid of $10, he could enter $9.99 just to annoy you (or more likely, because he is acting on behalf of the seller). In a conventional auction there is no need to keep bids secret because the maximum amount you're prepared to pay is kept in your head. Ebay publishes 'bids' but the maximum bid entered by each participant is secret until the auction ends. Or is it?
I don't really see why it makes sense for zip and unzip programs to care about encryption. If you want to encrypt the whole archive, it's simple to use GPG on the whole thing. If you want encryption on a per-file basis - again, use GPG on individual files before or after archiving. This is true on Windows too, using whatever your preferred GUI encryption program might be.
The only reason to stuff both functions into a single program seems to be the perennial problem of installing anything on Windows systems (you can't assume that an encryption tool is available) and marketing - why should users pay $20 twice for two different pieces of tacky shareware when they could pay Winzip $40 for one?
Everyone knows about Cygwin which is a POSIX layer and distribution of GNU software for Windows. You can compile and run a lot of packages, including X applications.
Line takes this a step further by running unmodified Linux binaries on Windows, but it's not currently an active project.
ISTR that one of the 32-bit DOS extenders (EMX perhaps?) had some wacky scheme to run the same binaries on Linux, Windows and OS/2.
Slashdot Mirror
But you can buy gold jewellery... it may not be so heavily marketed as 'romantic' compared to diamonds, but it can be just as expensive and that's the important thing.
Gold mining is not a cartel, and I don't think artificial gold will be made in large quantities any time soon.
Bittorrent and rsync both use chunks, but rsync can take chunks from an older version of the file that is almost what you want, compare them against the new version, and transfer only what has changed. Getting a file through Bittorrent, AFAIK, means getting all of it, even though you may get chunks out-of-order.
When will someone combine bittorrent and rsync? That sounds like the best way to upgrade from Fedora Core 2 test1 to Fedora Core 2 test2, or to update your Gentoo source tree.
If you run a web server program, and instruct it to serve files from a certain area, and place files into that area, then you have made them public. No matter if you link to them or not. The step in making the files public is in putting them on the web site; the URL is just an address. HTTP does have authentication measures which are easy to set up with servers such as Apache; don't blame others if you choose not to use them.
You can block Google with your firewall but that is no protection against sneakier robots which ignore robots.txt and don't identify themselves in the User-Agent. If you want a document to be non-public, get the web server to put a password on it or PGP encrypt it.
The trouble with any name ending 'fox' is that it makes one think of the awful Linux Fox which fortunately never had a chance of being adopted as the Linux logo. There was a Slashdot story about it at the time, but Slashdot's search box is so useless I can't find it.
Does KDE 3.2 work well with 16 colours? 4 colours? Black and white?
Traditionally a lot of X11/Unix desktops were high resolution but only 1-bit colour; that's a bit extreme these days but 4-bit colour should be enough for a good screenshot that doesn't eat bandwidth. I fear it would get in the way of all the kewl graduated shading though.
You can't sterilize an astronaut, but if he is wearing a space suit there shouldn't be any difficulty in soaking a sponge with bleach and rubbing it all round the suit before leaving the craft. You have to sterlize the outside of the spacecraft, but that same problem is faced with robots.
Could someone suggest a review site that doesn't split every article across 20 web pages?
ISTR that Digital had some kind of dynamic recompilation 386 emulator for the Alpha, and it was good enough that a 500MHz Alpha performed faster than a 500MHz PC for running PC software. Whatever happened to that?
Never mind the GNOME/KDE fanboy wars, I'm more interested in another holy war: plan text versus HTML mail. The info page for the mailing list Perens wrote to says 'Please use HTML email when sending to the list.' Why on earth would anyone actively want that? But he seems to have broken his own rule, since his message is plain text.
Windows 3.0 should be no problem - I ran it in software emulation on my 35MHz Archimedes, although the performance was not blazing fast. I think CPUs are now getting to the level where Win95 should be possible, after all the main factor affecting speed is available RAM.
Still, you'd need some rather cunning dynamic recompilation to get anything less than a 10x slowdown in MHz terms.
A decrease in the standard of living for the American (or $COUNTRY_X) programmers, because they can no longer charge such high prices as before. But programmers and software companies are in a tiny minority compared to the users of software and business which need to pay for it. The people who benefit from free trade are not just the Indian (or $COUNTRY_Y) programmers, but the Western businesses who are able to get what they need more cheaply, and the consumers who (assuming decent competition) get lower prices.
Free trade is just bringing together those who have something to sell - the Indian programmers - and those who want to buy - American firms needing software written. I don't see any reason for a third group to whine about this just because they were previously able to get away with charging more.
This is especially hypocritical on a site such as Slashdot, where readers depend on a steady supply of computer hardware often built in countries with lower wages than the West. In stories about video cards or RAM I don't think I have _ever_ seen any complaint about free trade reducing the price of the hardware and the lost job opportunties for Americans caused by building it in the Far East. Or think of the constant RIAA stories - stop trying to get in the way of progress, stop trying to prop up a failing business model, you don't automatically have the right to keep on getting money just because you did in the past. I know this is partly the fallacy of assuming Slashdot readers speak with one voice, but it's still worth noting.
You mean you *could not* care less about a camera. Think about it.
SQLite boasts of being typeless, and having no checking that a column declared of type integer really does store integers, for example. But for me this defeats one of the main reasons to use a relational database - strong type checking and validation. SQLite also doesn't support check constraints so there is really nothing you can do to fix this.
It's not inherently insecure to allow actions to be associated with shortcuts or add other whizzy things. If the user has write access to the filesystem, why shouldn't the user set up these things for his own files?
Consider - the window manager I use (icewm) has a menu file saying what programs appear on the start menu. It is possible for a program (including a worm) running as my uid to change that menu so that clicking on 'xterm' runs something else instead. This does not mean icewm is insecure.
The problem comes from running malware with full user privileges. As long as that happens, the system will be insecure anyway whether or not there are fancy WinFS active content things.
(However, if WinFS is so stupid as to display active content for files which were created by another user or sent over email, and the design is so stupid as to allow the content full access to the machine, there is certainly a problem. The Acorn RISC OS had a severe security bug of this kind - each application directory (like an appdir on NeXT) had a file !Boot which would be executed merely on _viewing_ the directory listing. Of course this was the ideal vector for viruses.)
The point is more that it clearly can be done. As with a security hole in software - you don't ask whether anyone is currently going around exploiting it, you just get on and fix it.
The flaw is signing up to a service such as webmail has a cost in solving the captcha, but this is once per account. Then there is no cost for each message sent out, so you can send tons of messages, or at least enough to make the cost per message tiny.
Spam comes about because *sending a message is too easy*. Nothing you do that doesn't increase the cost of sending a message - whether in money, computing time, or human time - will do much to combat spam in the long run. Having captchas on account signup doesn't increase the costs of sending an individual message by much, unless each account is limited to a low volume of messages to a small number of recipients.
But you have to trust the auctioneer to keep your bid secret. If someone else knew you had entered a bid of $10, he could enter $9.99 just to annoy you (or more likely, because he is acting on behalf of the seller). In a conventional auction there is no need to keep bids secret because the maximum amount you're prepared to pay is kept in your head. Ebay publishes 'bids' but the maximum bid entered by each participant is secret until the auction ends. Or is it?
What good reason is there to leave out gcc, if it's simple to make a package of it and there is plenty of disk space?
'if you will' == archaic for 'if you want', 'if you like'.
'They were constructing an enormous barrier, a wall, if you like' == 'one could call it a wall'.
Unless the characters are meant to grow older during the stories (eg Harry Potter - although that's bigger than a trilogy).
Or if you were making a Dr Who trilogy and the Doctor was meant to regenerate...
I don't really see why it makes sense for zip and unzip programs to care about encryption. If you want to encrypt the whole archive, it's simple to use GPG on the whole thing. If you want encryption on a per-file basis - again, use GPG on individual files before or after archiving. This is true on Windows too, using whatever your preferred GUI encryption program might be.
The only reason to stuff both functions into a single program seems to be the perennial problem of installing anything on Windows systems (you can't assume that an encryption tool is available) and marketing - why should users pay $20 twice for two different pieces of tacky shareware when they could pay Winzip $40 for one?
Everyone knows about Cygwin which is a POSIX layer and distribution of GNU software for Windows. You can compile and run a lot of packages, including X applications.
Line
takes this a step further by running unmodified Linux binaries on Windows, but it's not currently an active project.
ISTR that one of the 32-bit DOS extenders (EMX perhaps?) had some wacky scheme to run the same binaries on Linux, Windows and OS/2.