The problem is that there are idiots on both sides of the transaction. P.T. Barnum would be so proud. If said idiot believes he can get rich quick by hiring a spammer to send out 100 million emails for $100 on his behalf, it doesn't matter if there is 0 response. The spams have already gone out, and even if he doesn't try again, there are thousands of other idiots willing to take his place.
For the spammers who are trying to use misspellings to get around filters, I wonder if soundex could fix that problem quickly. That is, instead of doing the Bayesian calculations on the raw tokens, calculate probabilities based on the soundex values of the the tokens. You might need to teach soundex that the number one sounds like I, and other leet-speek-like things, but this might solve the problem quickly and easily.
actually recognise that people do not want to watch a postage stamp sized weather report video.
My favorite example of this is a Nortel ad that was running frequently last year. It had a guy who was going to be speaking at some big meeting, but forget his speech at the office. He used his mobile videophone to have his office assistant read it to him quietly, and he repeated it to the unknowing audience, with the phone sitting hidden on the lectern.
Now, what use is the live video in this case? I can get the same functionality today with my plain-old 2G phone (no video, of course). If you just need to repeat what someone is telling you over the phone, you sure don't need the live video. If this gee-whiz, look-how-cool-the-future-is example, unconstrained by reality, is the best 3G can do, isn't it in a whole heap of trouble?
Perl-style regexps tend to be used on things that post-date perl.
True, but things get tricky quickly -- plain-old Unix awk predates perl. But GNU-awk (gawk) does not, so it has some perl-style regexp features, like \w, which are missing from Unix awk.
But what drives me nuts about using regexps is how they differ slightly from implementation to implementation. Even though the perl regexp's tend to be the de-facto standard, the perl people are frequently adding stuff to their regexps. Some regexp implementations require you to escape open-paren to get the special meaning, and not escaped to match an open paren. Others require just the opposite. Madness!
The data in this article claims that 1% of all corporate mail servers in the UK allow open relaying, down from 91% in 1997. For all we know, the total number of corporate e-mail servers has grown by a factor of 100 (or more) in the last six year, meaning that perhaps there are more open relays now.
The article also doesn't measure the amount of spam coming through those relays. Even if there are only 10 open relays in the UK at any one time, it still might be possible for all of the spam to be coming through them.
Certainly, closing down open relays is a good thing, but lowering the percentage of open relays doesn't prove anything about the source of spam
Exactly! I've been in a lot of rather expensive hotels with guilded lobbies, liveried doorman, etc., but had rooms that were not soundproofed from road noise, the elevator, or the ice machine around the corner. Several times, I've unplugged the ice machine in the hallway in the middle of the night just to get some sleep. Unfortatenly, unplugging the elevator is a bit more difficult. Ding!
I'm surprised that the NDA isn't under NDA itself.
Not the usual anti-patent rant
on
Steal This Idea
·
· Score: 4, Interesting
We've all seen and mocked stupid patents, like the patent on swinging sideways on a playground swing. However, I don't think pointing out random bad patents is a useful way to critique the current US patent system. We all know there are a huge number of patents, and with any huge collection, there are outliers.
However, when searching for the mythical Novell Unix patent a the patent office I was really struck by how bad every software patent was.
For example, when searching for patents assigned to Novell (search criteria AN/Novell), the very first patent returned is number 6,567,873, which is a patent having to do with spinlocks in an SMP kernel. Basically, the patent covers the idea of exponential backoff for a contented resource. This is something which ethernet has done for 30 years, and I'm sure there's even further prior art.
Another Novell patent involves resizing FAT file partitions on the fly, and involves no real insight at all.
But it's not these two patents. Almost every single patent is either just this obvious, or just this derivative of prior work. Check it out yourself -- pretty much every computer program ever written must violate hundreds of patents.
Re:Questions I'd like the experts to answer
on
Today's SCO News
·
· Score: 1
Consider that any expert that sees the "secret evidence" may then be unable to do any future work on the Linux kernel. This means the expert almost certainly cannot be someone familiar with the Linux kernel.
This is why I'd like to see the questions I outlined above answered, as it gives non-experts better leverage to think about the issues.
However, I disagree with your assertion. There are a lot of people, especially people who have signed NDAs concerning the original Unix source code, who may not have contributed anything to Linux, but will certainly be able to understand the kernel and the history. I don't think current intimate knowledge of the Linux kernel is that important. Rather, understanding how things like the various Unix standards get written and mapped into code is important. I'm thinking of people like the original Bell Labs team, the BSD CSRG folks, many professors who used to teach Unix internals, kernel hackers from places like Sun, HP, IBM (oops, maybe they'd be excluded), SGI, etc. I don't know who wrote standard documents like the Unix 95 spec, and POSIX, and things like that, but they'd probably be good sources, too.
Re:Questions I'd like the experts to answer
on
Today's SCO News
·
· Score: 1
SCO said in yesterday's interview and several other recent interviews that the code is in the Linux kernel
If SCO was to be generally believed, then we wouldn't need these experts to back up what they are saying, which is why I'd like the experts to verify that it is indeed the kernel. I've also read that "entire programs" are the same, which would imply userspace.
Questions I'd like the experts to answer
on
Today's SCO News
·
· Score: 4, Insightful
SCO claims that they will soon reveal the alledged copyright infractions
in Linux to a panel of experts. I haven't seen the names of these
experts, so I'm a little worried about how expert they may be. If
someone knew C well, but not the history of Unix, I could show them a
10 line snippet of code, and they might agree that the code had a common
ancestor. But that doesn't mean that Linux copied the code from SCO. Here
are some questions which I'd like to see these experts answer publically.
Where is the code?
SCO has said that for legal reasons, they won't identify the exact code.
But it would be useful to know if the code is in the kernel, or in
user space. If the kernel, at least what directory, or section of the
kernel is it in?
If it is a header file, in the include directory,
similarities are to be expected. For example, the Linux system call
numbers are the some as many other OSes. These number are well known, and
and have been for decades, and can't possibly be a trade secret. POSIX
defines a zillion well-known constants which will show up in most every
POSIX compliant OS, and similarities are also to be expected here. Again,
this doesn't mean that Linux copied from SCO, rather, that they both copied
from POSIX.
If the code is in a device driver, it is very possible that both the SCO
code and the Linux code share a common ancestor -- many device drivers
are written by the device manufacturers, and given to the OS vendors
to include in their system.
If the code is in the CPU-specific part of the kernel, is it something
that there is only one way to do. Intel publishes specs on how to
manipulate low-level registers to do specfic tasks (start second CPUs, enable
memory protection, etc.), and there just aren't that many different
ways to do them.
Is the code is *BSD, or elsewhere on the net?
If this alledgedly infringing code is in *BSD, or on other places on
the net, this would again point to a non-SCO ancestor. For example,
Intel publishes Application Notes on how to use features of their processors.
It wouldn't surprise me to find this code in many different OSes which
use Intel processors, or Intel-compatible processors. Also, there is some
Unix(tm) code which has been given to the public domain, in the interest
of portability. I believe that cpio is one of these programs.
Both SCO and Linux use the X window system, which came from the
X consortium -- I'm sure there's a ton of common X related code in both,
but again, that doesn't mean that Linux copied from SCO.
The last Unix lawsuit resulted in marking all of the BSD code as "clean",
so if this infringing code is also in one of the BSDs, that would help
to indicate that it is also clean.
When did the code first originate?
As many people have pointed out, even if there is similar code in both Unix
and in Linux, that doesn't indicate which direction the code moved. If
such similarities can be found, the origination dates would need to be proved
as well. Apparently, SCO ships Samba, GCC, and other open source code
today, and marks it as such. Perhaps there's more code like this that slipped
into the kernel.
If you use the advanced search, you want to type someting like AN/Novell and Unix. This limits the search to patents which have been assigned to (i.e funded by) Novell, not those which just reference Novell.
According to Novell, the patents are assigned to Novell, and there are a bunch of patents there, though nothing jumps out as an obvious "Unix" patent. I'd just like to know which patent they are talking about. It seems contrary to the whole patent system not to mention which patent you claim someone to be violating.
For a long time when these people mentioned IP, we thought they meant copyright infringement, not patents, but now there's all this talk about patents. However, no one lists which one(s).
There's the famous 4135240 setuid patent, which Bell labs granted to the public domain, and which has expired by now anyway.
Novell gave us a clue, by pointing out that some patents might be in their name. But searching for Novell and Unix on the USPTO web site yields 62 patents. Most of these seem like they came from work on NetWare, but it is hard to tell for sure. Looking through these patents shows how bogus the US patent system is -- I quickly persued several at random, and every one was either an obvious technique, or being violated all over the place, or both. (IANAL).
The first patent returned by the search (6,546,433) lists "PowerBuilder 5 Unleashed!", by Sams publishing as reference material. Frankly, if I were a patent examiner, this would be evidence alone to reject the application.
So, does anyone want to make a prediction for ping time across 3,000 miles, and grid only knows how many hops? Does anyone know the record for most routers from one end of an IP network to the other today?
That's the point -- there was a point in the book where one was used where the other should have been (I forget which). Any reasonable editor would have caught the error, but it looked to me like the book was just run through a spell-checker and sent to the printer.
I don't expect much in the way of correct spelling, good grammar, and typos here on slashdot, and I make plenty of these mistakes myself. But when I'm paying north of $20 for a hardback book, like Cryptonomicon, I really expect to see the work of a professional editor. This book was filled with typos and even spell-checker kinds of errors (e.g. cannon vs canon). Never mind the perl code in the book which lost all newlines. It appeared that the manuscript had just been run through a spell checker, then sent to the printer. Can we expect better for this go around?
Slashdot Mirror
And it better be free!
The problem is that there are idiots on both sides of the transaction. P.T. Barnum would be so proud. If said idiot believes he can get rich quick by hiring a spammer to send out 100 million emails for $100 on his behalf, it doesn't matter if there is 0 response. The spams have already gone out, and even if he doesn't try again, there are thousands of other idiots willing to take his place.
My favorite example of this is a Nortel ad that was running frequently last year. It had a guy who was going to be speaking at some big meeting, but forget his speech at the office. He used his mobile videophone to have his office assistant read it to him quietly, and he repeated it to the unknowing audience, with the phone sitting hidden on the lectern.
Now, what use is the live video in this case? I can get the same functionality today with my plain-old 2G phone (no video, of course). If you just need to repeat what someone is telling you over the phone, you sure don't need the live video. If this gee-whiz, look-how-cool-the-future-is example, unconstrained by reality, is the best 3G can do, isn't it in a whole heap of trouble?
True, but things get tricky quickly -- plain-old Unix awk predates perl. But GNU-awk (gawk) does not, so it has some perl-style regexp features, like \w, which are missing from Unix awk.
The privacy folks worry mostly about RFID tags in cash.
But what drives me nuts about using regexps is how they differ slightly from implementation to implementation. Even though the perl regexp's tend to be the de-facto standard, the perl people are frequently adding stuff to their regexps. Some regexp implementations require you to escape open-paren to get the special meaning, and not escaped to match an open paren. Others require just the opposite. Madness!
The article also doesn't measure the amount of spam coming through those relays. Even if there are only 10 open relays in the UK at any one time, it still might be possible for all of the spam to be coming through them.
Certainly, closing down open relays is a good thing, but lowering the percentage of open relays doesn't prove anything about the source of spam
I've never found a pair that was both effective and comfortable. Do you have any recommendations?
Exactly! I've been in a lot of rather expensive hotels with guilded lobbies, liveried doorman, etc., but had rooms that were not soundproofed from road noise, the elevator, or the ice machine around the corner. Several times, I've unplugged the ice machine in the hallway in the middle of the night just to get some sleep. Unfortatenly, unplugging the elevator is a bit more difficult. Ding!
I'm surprised that the NDA isn't under NDA itself.
However, when searching for the mythical Novell Unix patent a the patent office I was really struck by how bad every software patent was.
For example, when searching for patents assigned to Novell (search criteria AN/Novell), the very first patent returned is number 6,567,873, which is a patent having to do with spinlocks in an SMP kernel. Basically, the patent covers the idea of exponential backoff for a contented resource. This is something which ethernet has done for 30 years, and I'm sure there's even further prior art.
Another Novell patent involves resizing FAT file partitions on the fly, and involves no real insight at all.
But it's not these two patents. Almost every single patent is either just this obvious, or just this derivative of prior work. Check it out yourself -- pretty much every computer program ever written must violate hundreds of patents.
This is why I'd like to see the questions I outlined above answered, as it gives non-experts better leverage to think about the issues.
However, I disagree with your assertion. There are a lot of people, especially people who have signed NDAs concerning the original Unix source code, who may not have contributed anything to Linux, but will certainly be able to understand the kernel and the history. I don't think current intimate knowledge of the Linux kernel is that important. Rather, understanding how things like the various Unix standards get written and mapped into code is important. I'm thinking of people like the original Bell Labs team, the BSD CSRG folks, many professors who used to teach Unix internals, kernel hackers from places like Sun, HP, IBM (oops, maybe they'd be excluded), SGI, etc. I don't know who wrote standard documents like the Unix 95 spec, and POSIX, and things like that, but they'd probably be good sources, too.
If SCO was to be generally believed, then we wouldn't need these experts to back up what they are saying, which is why I'd like the experts to verify that it is indeed the kernel. I've also read that "entire programs" are the same, which would imply userspace.
Where is the code?
SCO has said that for legal reasons, they won't identify the exact code. But it would be useful to know if the code is in the kernel, or in user space. If the kernel, at least what directory, or section of the kernel is it in?
If it is a header file, in the include directory, similarities are to be expected. For example, the Linux system call numbers are the some as many other OSes. These number are well known, and and have been for decades, and can't possibly be a trade secret. POSIX defines a zillion well-known constants which will show up in most every POSIX compliant OS, and similarities are also to be expected here. Again, this doesn't mean that Linux copied from SCO, rather, that they both copied from POSIX.
If the code is in a device driver, it is very possible that both the SCO code and the Linux code share a common ancestor -- many device drivers are written by the device manufacturers, and given to the OS vendors to include in their system.
If the code is in the CPU-specific part of the kernel, is it something that there is only one way to do. Intel publishes specs on how to manipulate low-level registers to do specfic tasks (start second CPUs, enable memory protection, etc.), and there just aren't that many different ways to do them.
Is the code is *BSD, or elsewhere on the net?
If this alledgedly infringing code is in *BSD, or on other places on the net, this would again point to a non-SCO ancestor. For example, Intel publishes Application Notes on how to use features of their processors. It wouldn't surprise me to find this code in many different OSes which use Intel processors, or Intel-compatible processors. Also, there is some Unix(tm) code which has been given to the public domain, in the interest of portability. I believe that cpio is one of these programs.
Both SCO and Linux use the X window system, which came from the X consortium -- I'm sure there's a ton of common X related code in both, but again, that doesn't mean that Linux copied from SCO.
The last Unix lawsuit resulted in marking all of the BSD code as "clean", so if this infringing code is also in one of the BSDs, that would help to indicate that it is also clean.
When did the code first originate?
As many people have pointed out, even if there is similar code in both Unix and in Linux, that doesn't indicate which direction the code moved. If such similarities can be found, the origination dates would need to be proved as well. Apparently, SCO ships Samba, GCC, and other open source code today, and marks it as such. Perhaps there's more code like this that slipped into the kernel.
If you use the advanced search, you want to type someting like AN/Novell and Unix. This limits the search to patents which have been assigned to (i.e funded by) Novell, not those which just reference Novell.
According to Novell, the patents are assigned to Novell, and there are a bunch of patents there, though nothing jumps out as an obvious "Unix" patent. I'd just like to know which patent they are talking about. It seems contrary to the whole patent system not to mention which patent you claim someone to be violating.
There's the famous 4135240 setuid patent, which Bell labs granted to the public domain, and which has expired by now anyway.
Novell gave us a clue, by pointing out that some patents might be in their name. But searching for Novell and Unix on the USPTO web site yields 62 patents. Most of these seem like they came from work on NetWare, but it is hard to tell for sure. Looking through these patents shows how bogus the US patent system is -- I quickly persued several at random, and every one was either an obvious technique, or being violated all over the place, or both. (IANAL).
The first patent returned by the search (6,546,433) lists "PowerBuilder 5 Unleashed!", by Sams publishing as reference material. Frankly, if I were a patent examiner, this would be evidence alone to reject the application.
Wiping the drive after it is removed from the machine is a pretty neat trick.
If each of these 400 employees required an expensive SGI workstation on their desks, just think of the savings!
So, does anyone want to make a prediction for ping time across 3,000 miles, and grid only knows how many hops? Does anyone know the record for most routers from one end of an IP network to the other today?
That's the point -- there was a point in the book where one was used where the other should have been (I forget which). Any reasonable editor would have caught the error, but it looked to me like the book was just run through a spell-checker and sent to the printer.
I agree. This is why I never create my own functions or methods. Evey program should be just one big function.
I don't expect much in the way of correct spelling, good grammar, and typos here on slashdot, and I make plenty of these mistakes myself. But when I'm paying north of $20 for a hardback book, like Cryptonomicon, I really expect to see the work of a professional editor. This book was filled with typos and even spell-checker kinds of errors (e.g. cannon vs canon). Never mind the perl code in the book which lost all newlines. It appeared that the manuscript had just been run through a spell checker, then sent to the printer. Can we expect better for this go around?