Never trust a compiler that eats your source file:
105. The statement has been truncated.
120. An identifier or constant has been truncated.
149. Text was deleted from this statement.
Ah, I remember writing PL/I extensively on IBM mainframes. IIRC, the language could not be represented using any of the grammar types available at that time (as an example of the difficulty, there were no reserved keywords, "if" could be a variable name, and some people used it that way). The ad hoc parser in IBM's compiler did not like to give up, so when it got desperate it would delete text until it reduced the offending statement to something that would parse.
You had to declare all your variables in advance, and if the statement that got truncated happened to be a declaration, it would typically be followed by dozens of pages of error messages about using undeclared variables.
I did get my 15 minutes of fame when I found an error in the optimizer that caused it to emit illegal op codes...
Unfortunately, don't neglect the fact that just up the street are dozens of vendors selling other attractive goodies (let's call them cookies and cake, I guess) that many people depend on, but that don't work unless you have a glass of Bill's lemonade in hand.
In the antitrust case, this was called the "application barrier to entry" and was one of the main reasons that MS was declared a monopolist.
Maybe I'm overly paranoid, but putting 1000W of power through a 1/4 inch phono plug for sustained periods of time would make me nervous. The AC connectors were designed with lots of power and >100V in mind-- they have big screw connectors that are well seperated, etc.
The red and green wires are, respectively, the ring and tip for the primary line. The black and yellow are the tip and ring for the secondary line, but there isn't any universal rule for which is used for what on the second pair.
In the early days of touch-tone pads, it made a difference. If you reversed the tip and ring, the touch-tone pad didn't work because it was powered off the DC voltage on the line and the polarity was wrong. This design error was eventually corrected by putting a rectifier in front of the touch-tone pad.
I would certainly assert that the RIAA and MPAA believe (whether stupidly or not) that armed with a single file format and a player that requires the use of patented algorithms, they can trivially win court cases. You have a DivX copy of on your drive? Under current law you're obviously guilty, that'll be a $1000 fine and your equipment is confiscated. Wrote and distributed a piece of software that decodes the file format without licensing the patented algorithms? That'll be $100,000 in damages. If they win a few well-publicized cases, they can reduce "piracy" to levels that they can live with.
The World Wide Web was invented by academic types who were trying to solve the problem of how to share information quickly, easily, and freely. Mainstream media is unfortunately controlled by companies that don't want to share at all.
I'll agree with others that the idea for both MS and HP is, as others have suggested, to make friends with the RIAA and MPAA, and get those content companies to trust the way MS and HP control access and copying. The unfortunate answer to the question
Who the hell would be dumb enough to buy one of these?
would appear to be, in the long term, anyone who wants to play mainstream audio or video content on their PC. Having a copy in some format other than those that are "properly" controlled will be de facto evidence of both copyright violation under the DMCA (because it wasn't released by the owner in that format, so your copy must be unauthorized) and patent violation under the damned software patent system (some aspects of the Windows Media formats are covered by patents, so any reader is illegal unless licensed by MS, and licensed readers will have to honor the protection scheme).
I am not pleased with the way the future of digital media is unfolding...
This seemed to be fairly common practice 25 years ago when I was around people who played in small bands. As it was explained to me, AC power cords could handle reasonable amounts of power at relatively low frequencies and were cheap and readily available. There's apparently a lot to be said for being able to replace a broken 1000W speaker cable with a quick trip to the local home supply store.
Like O'Reilly, I am opposed to a requirement that the government use only open-source applications. However, I am strongly in favor of the government using only file formats and associated algorithms which are effectively in the public domain, and have been public for long enough that open-source apps can be developed.
Public algorithms are important. Consider the case of audio and video. Technically, any MPEG player must implement algorithms which are covered by a number of patents held by individuals, companies, and universities. At least in the US, the Patent Office has allowed patenting of so many coding algorithms that it is difficult to implement a file format and player that does not infringe on some of them. The Oog people have been developing what they believe to be unencumbered algorithms, but it may take years (and multiple court cases) to establish that for a fact.
IANAMC (I am not a marriage counselor), but am speaking from the perspective of a 22-year marriage that seems to still be working. This is a decision that needs to be made by two people, and your posting sounds a little one-sided.
As many people have pointed out, it is possible to get a diamond that you know did not come from one of the countries/mines that you object to. If you and she agree that the two of you want to buy a ring, explain that you feel strongly about the source of the stone and limit the choices. If she doesn't understand, to the point that you can't work something out, your marriage is apt to have bigger problems than a chunk of crystalized carbon.
If it's a financial thing and you're trying to avoid spending money that you think is a waste (a position taken by many posters), just keep in mind that it's about to become "our" money, not "yours" and "hers". More marriages fail over how to handle the finances than any other cause. My wife-to-be and I spent a lot of time exploring how each other felt about wealth and money and spending on different kinds of things before we got married.
Years ago I was involved in a study where people were shown two audio/video clips, one with excellent video and somewhat impaired audio and the other with excellent audio and noticably (at least to experts) impaired video. Subjects were told that the video quality would be different, and were asked to rate which one was superior. No mention of the audio difference was made. People consistently rated the bad video with good audio as having superior video quality. If the same audio quality was used for both clips, people rated the video quality properly.
The weekend after we got the results, I went home and rewired things so my VCR and cable audio went through the good amplifier and big speakers...
Bear in mind that what you think of when you hear "1 TFLOP" and what the people touting the chip mean are probably very different things. SSE-like instructions which work on (for example) eight 16-bit values packed into 128-bit words will be counted as eight operations executed, not one.
If there are multiple execution units that can operate in parallel, the hypsters will assume that each unit is actually used every cycle, when in fact there will be many cycles where some units are idle.
Moore's Law (as applied to transisters) makes it feasible to consider that the main processor, an audio processor and one or more video processors could all be put on a single chip.
Consider their statement sort of a peak processing level if everything is perfect. How fast are today's machines if you (a)add together all of the processing done in the P4, the graphics chip, the audio chip, and the MPEG decoder in your DVD controller, and (b)you assume that it's running some hypothetical instruction sequence loaded with parallelism and (c)instruction scheduling is perfect?
Three years ago there were media processor chip designs clocked at 200 MHz and "delivering" 20B ops/sec under such assumptions. If you talked to the actual designers, though, they told you the chips would deliver 1-3B ops/sec over short intervals while you were executing specialized blocks of code. And for arbitrary code, you might get less than 200M ops/sec.
Applies to large corporate customers too?
on
More MS EULA Fun
·
· Score: 1
Does anyone know if these same terms are showing up in service packs delivered to large corporate customers? Our IT organization gets really bent about people not having the "standard" OS and app images on their machines, and their own tools and schedules for upgrades. They would presumably go ballistic if thousands of machines starting getting into odd states because of partial upgrades...
One of the Laws related to Moore's asserts that the cost of a state-of-the-art fab line doubles every three years. The article says that this line is costing $2.5B. IIRC, new lines in 1999 cost about $800M, so this would appear to be pretty close to the prediction. The potentially bad news in this is that by 2011, a new fab line will cost $20B, which is probably more than anyone except large governments (or Microsoft) can afford. By simple calculation, 100M working devices produced over the lifetime of a $20B fab line must cost $200 each just to cover the initial cost of the fab.
I don't really care about Moore's Law itself (transister count doubles every 18 months), but do care about the corollary that says instructions per second per dollar doubles every 18 months. Can we keep that corollary going without Moore's Law itself (and the attendant economic fab limitations)? Asynchronous circuit designs? Parallel processing? Alternate cheaper fabrication for a 1B op/second processor?
Not just computer-illiterate, but illiterate in general. Most of the common PC apps today (games are one clear exception) implement some form of "smart paper", and most files contain data that constitute a smart-paper object. People who are not comfortable with paper and text and drawings and methods like indices and file folders for organizing collections of paper are not going to be comfortable with smart-paper applications, or collections of smart-paper objects, and there isn't any UI that can fix that. It is probably telling that game consoles, allow users to choose and start games by some non-text-oriented means -- plug in the cartridge, put the CD in the tray, etc.
We all have regular experiences that demonstrate clearly that literate and illiterate people are going to need different UIs, that there is no one-size-fits-all UI that meets the needs of everyone.
The author makes the assumption that the service that cable modem and other broadband providers
should be selling is interconnection of networks. In that case, he is right that "basic" service consists of (a) giving the subscriber a network address and netmask, and (b) routing their packets. Presumably it would be up to the subscriber to handle domain hosting, mail service, etc.
In practice, there are relatively few people out there who want to buy this type of service. As a result, current broadband service focuses on operating a subnet to which subscribers connect individual hosts. Even in those few situations where the cable operator allows multiple devices, the model is almost always one with multiple independent hosts, not a true subnet.
If the market eventually demands (and is willing to pay for) network connectivity instead of simple host support, the broadband operators will undoubtedly offer it as an option.
How many times do we have to go through this? If you want to provide the source material that historians will use 100 years from now (or at least have a chance at providing it), use the following techniques:
Don't send e-mail or make phone calls, write letters. Use pigment-based ink and acid-free paper (I believe that laser printer "ink" will have excellent long-term characteristics, as it's essentially all pigment). Send them to people who are packrats and will save bundles of your letters in old trunks in the attic. If you turn out to be famous or important, they will eventually turn up.
Don't keep a blog, keep a journal. Use the same technology as for letters. Put them in the old trunk in the attic yourself. Or in the archives at a long-lived company (Lucent's engineering records include notebooks from Alexander Graham Bell). Or leave them to your university when you die.
Take black-and-white photographs. Use film and print chemistries based on silver, not dyes.
Publish in print when you can. Be sure to send free copies of your books to your university library.
IIRC, and it's been a number of years, the overall goal was about 50 minutes of outage per line per year (a little less than three nines). Different failure modes were allocated different parts of that total. Components like the wires, that only took a single line out of service, were allocated the lion's share. Switch components were allocated smaller amounts, depending on how many lines would be out of service. Total system failure on a switch was allocated about 4.5 minutes per year (five nines).
No switching system ever actually made that grade. Probably the ones that came closest were the old electromechanical "steppers". Many small steppers in small towns ran completely unattended, and maintenance consisted of someone driving out once a month to make sure the building was still there and to polish some relay contacts.
All of the computer-controlled switches had dual synchronized processors (ie, each one executing the same op codes at the same time) and duplex memory, with a bunch of extra hardware to detect faults. The single most common cause of total system failure was when a fault had occured, and the system was running "simplex", and a tech pulled a card from the active rather than the failed processor.
It seems completely clean to me to take their
figure into Rhino or 3DS Max and hand fit a mesh
that will conform to their figure, say in order to make a pair of pants that will fit their character. The mesh is mine, the shape is mine as well - it approximates their shape
How is this cleaner than if I scan someone's book, use software to change the fonts and make synonym substitutions? Or read a track from a CD and use software to reverse the LSB in each 16-bit sample? Attempting to claim that those results weren't covered under the original author's copyright would be laughed out of court.
Ten years or so ago, I had to write an event dispatcher for a distributed processing arrangement. Processes on different machines were long-running and had persistant side effects that could cause problems, and it was important that the dispatcher made good choices about which processes saw any given event.
The dispatcher ended up implementing a virtual machine with an instruction set tailored to the particular problem and the data structures that were involved in events and process responses to those events (which were returned through the dispatcher). The instruction set included the usual flow control stuff (branch on various conditions) and some odder things like "adding" two complicated data structures. The VM provided an efficient way to serialize the dispatch process.
Writing programs for the virtual machine was a bitch, so I eventually implemented a high-level declarative language describing how different processes could respond to an event, and relationships like precedence and mutual exclusion that existed between the different processes, and a compiler that converted a set of declarations into procedural byte-code programs for the VM implemented in the dispatcher.
One issue that they would run into with arial plant (ie, stringing it between the poles) are the fees that have to be paid if the poles are owned by some other local utility. For the case I'm most familar with, a cable company attaching its wires (including fiber) to power company poles, the fee is a few dollars per pole attachment per month. If the "community" in this case is an actual incorporated entity (a real town), this might not be an issue since they would have an existing deal with the pole owner, which might include "free" attachments.
If the local phone lines are underground and in ductwork of some sort (fairly common in new construction), the state may require that the phone company share the duct space for something like this, but there would probably still be a monthly fee.
This judge is going to end up with the same basic remedy that Judge Jackson had. She's going to toss both the DOJ/MS proposal and that of the dissenting states. She has to deal with all of the things that the Appeals Court upheld as being violations of the antitrust law-- code mingling, non-removable applications, application placement, middleware, etc. She can take into account MS's behavior, including behavior after the guilty verdict. Her conclusion will be, that in order to address those issues and restore competition, she needs to split MS (either literally, or through fairly extreme conduct restrictions) into an OS company and an application company. The OS APIs will have to be thoroughly documented, to the extent that someone else can duplicate it and run Windows apps. She will require source code disclosure only if it turns out that that's the only sufficient way to "document" the APIs.
The Appeals Court didn't overturn Jackson's remedy because it wasn't proper, they overturned it because he didn't follow an appropriate process for imposing such a severe remedy.
There have been a rash of security flaws announced recently by MS. Does anyone know how many of these are being found by outside parties, and how many by MS internally? If the five-month-old security initiative is finding the errors, good for them! There was a boatload of code to cover, and it was bound to take time. If the majority of these are still being found by outside people who don't have access to the source, then BillG needs to smack his security czar upside the head.
The people selling these things want to make money, which means they want to give people whatever it is that they want to pay the most for.
The studios and labels do not view their business this way, they want to dictate where and when a person can watch/listen, and to dictate the price as well. So they fight against various kinds of software technology, and try to make reverse engineering illegal, and ignore the fact that the real answer is in economics.
Given two products (a crappy VCD version of a movie and a good DVD) with the same price, the consumer will almost always choose the higher quality one. The cost of the VCD is the time, effort, cost of blanks, cost of network access (not just the bandwidth itself, but what you might have done with that bandwidth if you weren't downloading the movie). The cost of the DVD is whatever your local store charges. If the studios would price their product properly, casual piracy would be reduced to a minor inconvenience.
Good approach. For most users, a 3GB monthly limit is simply not going to come into play (most SlashDot readers are probably atypical). As you point out, seven songs per day is about 1GB. At 16kbps, playing a networked online first-person-shooter game for two hours per day adds up to about 430MB. At 24kbps, streamed audio for two hours per day adds up to about 650 MB. On this scale, e-mail and Web surfing are almost incidental. File sharing involving ISO images and VCDs will get you to the limit fairly quickly.
For sound technical reasons, upstream bandwidth in a cable plant is more difficult to provide hence a more limited resource. It makes sense that the limits are lower and "easier" to reach.
Speaking as a planner from the inside of the business, there are about 2% of our cable modem customers that I would really, really like to see move to DSL. Let 'em clog up the telco's network instead of mine!
This is a reasonable point, and the license under which Berkeley released their code satisfied what I believe was their primary goal at that time, which was to get a proper TCP/IP stack into lots of computers. One could certainly argue that this would be a good license for many pieces of software developed by the government or an educational institute who is seeking wide-spread adoption across many platforms.
On the other hand, assuming that Berkeley intended to remain in the support business for their TCP/IP stack, the license has the weakness that other people who start from the Berkeley source and improve it (including bug fixes) are not obligated to reveal those changes. Thus MS (and Pyramid and HP and any number of companies that adopted the Berkeley code) benefit directly from the original code. But if one of them fixes a bug and keeps it secret, only they benefit-- not even Berkeley, the original author, benefits from that fix.
If the Berkeley stack had been released under the GPL, I'm sure that vendors who wanted to use it could have designed their system around the limitation so that they could release the source for their version of the stack without revealing the proprietary portions of their OS. The FSF would be hard-pressed to argue that your file system or memory management code is derivative from the TCP/IP stack.
Slashdot Mirror
You had to declare all your variables in advance, and if the statement that got truncated happened to be a declaration, it would typically be followed by dozens of pages of error messages about using undeclared variables.
I did get my 15 minutes of fame when I found an error in the optimizer that caused it to emit illegal op codes...
Unfortunately, don't neglect the fact that just up the street are dozens of vendors selling other attractive goodies (let's call them cookies and cake, I guess) that many people depend on, but that don't work unless you have a glass of Bill's lemonade in hand.
In the antitrust case, this was called the "application barrier to entry" and was one of the main reasons that MS was declared a monopolist.
Maybe I'm overly paranoid, but putting 1000W of power through a 1/4 inch phono plug for sustained periods of time would make me nervous. The AC connectors were designed with lots of power and >100V in mind-- they have big screw connectors that are well seperated, etc.
In the early days of touch-tone pads, it made a difference. If you reversed the tip and ring, the touch-tone pad didn't work because it was powered off the DC voltage on the line and the polarity was wrong. This design error was eventually corrected by putting a rectifier in front of the touch-tone pad.
The World Wide Web was invented by academic types who were trying to solve the problem of how to share information quickly, easily, and freely. Mainstream media is unfortunately controlled by companies that don't want to share at all.
I am not pleased with the way the future of digital media is unfolding...
This seemed to be fairly common practice 25 years ago when I was around people who played in small bands. As it was explained to me, AC power cords could handle reasonable amounts of power at relatively low frequencies and were cheap and readily available. There's apparently a lot to be said for being able to replace a broken 1000W speaker cable with a quick trip to the local home supply store.
Public algorithms are important. Consider the case of audio and video. Technically, any MPEG player must implement algorithms which are covered by a number of patents held by individuals, companies, and universities. At least in the US, the Patent Office has allowed patenting of so many coding algorithms that it is difficult to implement a file format and player that does not infringe on some of them. The Oog people have been developing what they believe to be unencumbered algorithms, but it may take years (and multiple court cases) to establish that for a fact.
IANAMC (I am not a marriage counselor), but am speaking from the perspective of a 22-year marriage that seems to still be working. This is a decision that needs to be made by two people, and your posting sounds a little one-sided.
As many people have pointed out, it is possible to get a diamond that you know did not come from one of the countries/mines that you object to. If you and she agree that the two of you want to buy a ring, explain that you feel strongly about the source of the stone and limit the choices. If she doesn't understand, to the point that you can't work something out, your marriage is apt to have bigger problems than a chunk of crystalized carbon.
If it's a financial thing and you're trying to avoid spending money that you think is a waste (a position taken by many posters), just keep in mind that it's about to become "our" money, not "yours" and "hers". More marriages fail over how to handle the finances than any other cause. My wife-to-be and I spent a lot of time exploring how each other felt about wealth and money and spending on different kinds of things before we got married.
Years ago I was involved in a study where people were shown two audio/video clips, one with excellent video and somewhat impaired audio and the other with excellent audio and noticably (at least to experts) impaired video. Subjects were told that the video quality would be different, and were asked to rate which one was superior. No mention of the audio difference was made. People consistently rated the bad video with good audio as having superior video quality. If the same audio quality was used for both clips, people rated the video quality properly.
The weekend after we got the results, I went home and rewired things so my VCR and cable audio went through the good amplifier and big speakers...
Bear in mind that what you think of when you hear "1 TFLOP" and what the people touting the chip mean are probably very different things. SSE-like instructions which work on (for example) eight 16-bit values packed into 128-bit words will be counted as eight operations executed, not one. If there are multiple execution units that can operate in parallel, the hypsters will assume that each unit is actually used every cycle, when in fact there will be many cycles where some units are idle. Moore's Law (as applied to transisters) makes it feasible to consider that the main processor, an audio processor and one or more video processors could all be put on a single chip.
Consider their statement sort of a peak processing level if everything is perfect. How fast are today's machines if you (a)add together all of the processing done in the P4, the graphics chip, the audio chip, and the MPEG decoder in your DVD controller, and (b)you assume that it's running some hypothetical instruction sequence loaded with parallelism and (c)instruction scheduling is perfect?
Three years ago there were media processor chip designs clocked at 200 MHz and "delivering" 20B ops/sec under such assumptions. If you talked to the actual designers, though, they told you the chips would deliver 1-3B ops/sec over short intervals while you were executing specialized blocks of code. And for arbitrary code, you might get less than 200M ops/sec.
Does anyone know if these same terms are showing up in service packs delivered to large corporate customers? Our IT organization gets really bent about people not having the "standard" OS and app images on their machines, and their own tools and schedules for upgrades. They would presumably go ballistic if thousands of machines starting getting into odd states because of partial upgrades...
One of the Laws related to Moore's asserts that the cost of a state-of-the-art fab line doubles every three years. The article says that this line is costing $2.5B. IIRC, new lines in 1999 cost about $800M, so this would appear to be pretty close to the prediction. The potentially bad news in this is that by 2011, a new fab line will cost $20B, which is probably more than anyone except large governments (or Microsoft) can afford. By simple calculation, 100M working devices produced over the lifetime of a $20B fab line must cost $200 each just to cover the initial cost of the fab.
I don't really care about Moore's Law itself (transister count doubles every 18 months), but do care about the corollary that says instructions per second per dollar doubles every 18 months. Can we keep that corollary going without Moore's Law itself (and the attendant economic fab limitations)? Asynchronous circuit designs? Parallel processing? Alternate cheaper fabrication for a 1B op/second processor?
Not just computer-illiterate, but illiterate in general. Most of the common PC apps today (games are one clear exception) implement some form of "smart paper", and most files contain data that constitute a smart-paper object. People who are not comfortable with paper and text and drawings and methods like indices and file folders for organizing collections of paper are not going to be comfortable with smart-paper applications, or collections of smart-paper objects, and there isn't any UI that can fix that. It is probably telling that game consoles, allow users to choose and start games by some non-text-oriented means -- plug in the cartridge, put the CD in the tray, etc.
We all have regular experiences that demonstrate clearly that literate and illiterate people are going to need different UIs, that there is no one-size-fits-all UI that meets the needs of everyone.
The author makes the assumption that the service that cable modem and other broadband providers should be selling is interconnection of networks. In that case, he is right that "basic" service consists of (a) giving the subscriber a network address and netmask, and (b) routing their packets. Presumably it would be up to the subscriber to handle domain hosting, mail service, etc.
In practice, there are relatively few people out there who want to buy this type of service. As a result, current broadband service focuses on operating a subnet to which subscribers connect individual hosts. Even in those few situations where the cable operator allows multiple devices, the model is almost always one with multiple independent hosts, not a true subnet.
If the market eventually demands (and is willing to pay for) network connectivity instead of simple host support, the broadband operators will undoubtedly offer it as an option.
IIRC, and it's been a number of years, the overall goal was about 50 minutes of outage per line per year (a little less than three nines). Different failure modes were allocated different parts of that total. Components like the wires, that only took a single line out of service, were allocated the lion's share. Switch components were allocated smaller amounts, depending on how many lines would be out of service. Total system failure on a switch was allocated about 4.5 minutes per year (five nines).
No switching system ever actually made that grade. Probably the ones that came closest were the old electromechanical "steppers". Many small steppers in small towns ran completely unattended, and maintenance consisted of someone driving out once a month to make sure the building was still there and to polish some relay contacts.
All of the computer-controlled switches had dual synchronized processors (ie, each one executing the same op codes at the same time) and duplex memory, with a bunch of extra hardware to detect faults. The single most common cause of total system failure was when a fault had occured, and the system was running "simplex", and a tech pulled a card from the active rather than the failed processor.
The dispatcher ended up implementing a virtual machine with an instruction set tailored to the particular problem and the data structures that were involved in events and process responses to those events (which were returned through the dispatcher). The instruction set included the usual flow control stuff (branch on various conditions) and some odder things like "adding" two complicated data structures. The VM provided an efficient way to serialize the dispatch process.
Writing programs for the virtual machine was a bitch, so I eventually implemented a high-level declarative language describing how different processes could respond to an event, and relationships like precedence and mutual exclusion that existed between the different processes, and a compiler that converted a set of declarations into procedural byte-code programs for the VM implemented in the dispatcher.
One issue that they would run into with arial plant (ie, stringing it between the poles) are the fees that have to be paid if the poles are owned by some other local utility. For the case I'm most familar with, a cable company attaching its wires (including fiber) to power company poles, the fee is a few dollars per pole attachment per month. If the "community" in this case is an actual incorporated entity (a real town), this might not be an issue since they would have an existing deal with the pole owner, which might include "free" attachments.
If the local phone lines are underground and in ductwork of some sort (fairly common in new construction), the state may require that the phone company share the duct space for something like this, but there would probably still be a monthly fee.
Remember that you read it here first...
This judge is going to end up with the same basic remedy that Judge Jackson had. She's going to toss both the DOJ/MS proposal and that of the dissenting states. She has to deal with all of the things that the Appeals Court upheld as being violations of the antitrust law-- code mingling, non-removable applications, application placement, middleware, etc. She can take into account MS's behavior, including behavior after the guilty verdict. Her conclusion will be, that in order to address those issues and restore competition, she needs to split MS (either literally, or through fairly extreme conduct restrictions) into an OS company and an application company. The OS APIs will have to be thoroughly documented, to the extent that someone else can duplicate it and run Windows apps. She will require source code disclosure only if it turns out that that's the only sufficient way to "document" the APIs.
The Appeals Court didn't overturn Jackson's remedy because it wasn't proper, they overturned it because he didn't follow an appropriate process for imposing such a severe remedy.
There have been a rash of security flaws announced recently by MS. Does anyone know how many of these are being found by outside parties, and how many by MS internally? If the five-month-old security initiative is finding the errors, good for them! There was a boatload of code to cover, and it was bound to take time. If the majority of these are still being found by outside people who don't have access to the source, then BillG needs to smack his security czar upside the head.
Near the end of the piece, he says:
The studios and labels do not view their business this way, they want to dictate where and when a person can watch/listen, and to dictate the price as well. So they fight against various kinds of software technology, and try to make reverse engineering illegal, and ignore the fact that the real answer is in economics.Given two products (a crappy VCD version of a movie and a good DVD) with the same price, the consumer will almost always choose the higher quality one. The cost of the VCD is the time, effort, cost of blanks, cost of network access (not just the bandwidth itself, but what you might have done with that bandwidth if you weren't downloading the movie). The cost of the DVD is whatever your local store charges. If the studios would price their product properly, casual piracy would be reduced to a minor inconvenience.
Good approach. For most users, a 3GB monthly limit is simply not going to come into play (most SlashDot readers are probably atypical). As you point out, seven songs per day is about 1GB. At 16kbps, playing a networked online first-person-shooter game for two hours per day adds up to about 430MB. At 24kbps, streamed audio for two hours per day adds up to about 650 MB. On this scale, e-mail and Web surfing are almost incidental. File sharing involving ISO images and VCDs will get you to the limit fairly quickly.
For sound technical reasons, upstream bandwidth in a cable plant is more difficult to provide hence a more limited resource. It makes sense that the limits are lower and "easier" to reach.
Speaking as a planner from the inside of the business, there are about 2% of our cable modem customers that I would really, really like to see move to DSL. Let 'em clog up the telco's network instead of mine!
This is a reasonable point, and the license under which Berkeley released their code satisfied what I believe was their primary goal at that time, which was to get a proper TCP/IP stack into lots of computers. One could certainly argue that this would be a good license for many pieces of software developed by the government or an educational institute who is seeking wide-spread adoption across many platforms.
On the other hand, assuming that Berkeley intended to remain in the support business for their TCP/IP stack, the license has the weakness that other people who start from the Berkeley source and improve it (including bug fixes) are not obligated to reveal those changes. Thus MS (and Pyramid and HP and any number of companies that adopted the Berkeley code) benefit directly from the original code. But if one of them fixes a bug and keeps it secret, only they benefit-- not even Berkeley, the original author, benefits from that fix.
If the Berkeley stack had been released under the GPL, I'm sure that vendors who wanted to use it could have designed their system around the limitation so that they could release the source for their version of the stack without revealing the proprietary portions of their OS. The FSF would be hard-pressed to argue that your file system or memory management code is derivative from the TCP/IP stack.