Re:Microsoft can't legally bundle GPL software
on
Linux Users Are Spoiled
·
· Score: 2, Interesting
Under your interpretation, exactly how would *any* GPL Windows executable be legal to run? If it can be installed by the end user, why can't MS distribute it?
For that matter, under your interpretation, how can *any* GPLed Windows app legally exist, excepting of course the case of the non-dependent app? If the author of a GPLed program links to a non-GPL system library, then distributes it, aren't they violating the GPL themselves?
I'm certainly no lawyer, but my reading of the clause in question definitely doesn't include that clause extending the GPL onto software not otherwise covered. If GPLed Windows apps can legally exist, then I can't see how MS distributing them becomes any different than the author distributing them - that's one of the nice things about the GPL. If party A *legally* distributes application A as package A, then party B can legally distribute package A, no matter what.
What that exemption says is, you have to distribute all source code - except, if some of the source code replicates a component normally distributed within the operating system, you're exempted from being required to provide the source code to that component, unless you're distributing the component as well. However, since distributing a GPL Windows executable with the system is generally going to count as mere aggregation, you were never under obligation to distribute source to the non-GPL sections of your product, and as such the exemption and the revocation of the exemption don't apply to it.
Basically, the clause is there to prevent you from having to distribute source to things that are shared with the OS. It can't extend the GPL beyond the core tenets of the GPL - unless the program links closely enough with Windows to invoke the clauses of the GPL on Windows, Windows itself remains seperate, and thus this clause doesn't apply.
It started them on the browser, but anyone who's worked in electrical CAD applications could tell you they've been around a lot longer than that, and I wouldn't be surprised to find out the mouse gestures existed before ECAD picked it up.
I know you meant this to be funny, but seriously - long distance driving SUCKS, and I would be all kinds of into a system like this.
Small vehicles (maybe, say, electric-powered?) that can only do 40-60 miles on a charge, and a who-cares-how-its-powered system those small vehicles can piggyback onto that does long hauls.
Banks, universities, and credit card companies might not let you get away with it, but Blockbuster will absolutely not require it if you complain. Since most video stores require your account to be credit-card backed anyway, there's no reason for them to ask for your SSN, and no reason for you to give it to them.
Whatever. It's possible you understand info theory, in which case you have no idea how to write. I don't know which one you're a failure at, so I'll leave it at that.
Of course, the Atlas V main stage uses a Russian-designed motor, the RD-180.
As to failures... well, let's not talk about Shuttle vs. Soyuz or Delta III.
The Russians are very, very good rocket engineers. Generally, Russian rocket designs are considered to be the most reliable; however, their manufacturing and maintenance isn't always up to snuff with their design.
Let's say, for theory's sake, that this netgear antenna plugged into a wall is where we are.
Who provides their upstream? Is it Comcast? SBC? Verizon? Qwest? Whoever it is, I bet the upstream has a SMTP server. Almost all of them will accept and route mail from clients on their network.
The provider of the pipe is the ISP. Not the antenna you're attached to, necessarily - in fact, not usually the antenna you're attached to. The provider of the antenna's pipe.
c) Until you've run it on the thousands of computers, how do you know it would work? As far as I'm concerned, you don't deserve the patent until you've proven it works.
Actually, I'm fine with the concept of eliminating all IP law, which would be most fair to the little guy in todays day and age. But if we have to keep it, requiring working models and/or proof of operation isn't so bad.
You claim that coding methods require a good SNR, which isn't true; most of the more clever coding methods are designed to work at low SNR. The statement I object to is "clever methods that get the most bitrate out of the bandwidth require good SNR". It's false. Some codes actually operate worse at high SNR compared to an uncoded channel than at low SNR. Basically, there's a quantity called coding gain - for most codes, this is a function of bit-error-rate and describes how much lower your SNR can be compared to the uncoded channel for the same BER. There are codes where coding gain *decreases* with increasing SNR after a certain point - in fact, most turbo codes are of this type. Even more complex - most turbo codes display a negative coding gain at extremely low SNRs - they're *worse* than an uncoded channel! - quickly catch up to and pass the uncoded channel, so their coding gain increases, and then when they hit the 'error floor' the coding gain decreases again as the uncoded channel begins to catch up. At a certain point of SNR increase, the turbo code is no longer attractive, as its coding gain per its computational complexity ceases to be worthwhile.
You claim that bitrate is dependent solely on bandwidth, when its dependent on bandwidth and SNR jointly. "If you want to transfer a few megabits per second, you're going to end up with bandwidth on the order of a few megahertz." As I proved in my calculations, no, you're not. Want me to do it again?
Transferring 3 megabits per second over a 20kHz bandwidth.
How much do you want to bet the mailserver provided by the ISP of the moment - the wifi provider - works?
Your ISP is the person providing your pipe, not the person you pay at home. If you aren't at home, you might need to use a different server - how hard is that to understand?
If you're a software developer and have a business case for direct port 25, here are your choices:
Convince Comcast to open port 25 for you. If you have a legit reason, they may. Of course, if you're trying to do business on a residential connection, they might not.
SSH or VPN to a corporate machine, where presumably you have port 25 unblocked, and can send that way.
Send it through Comcast's server.
I have no issue with there being limits, so long as those limits are not universal. If its something you need, pay for the access in one way or another (cash or time spent convincing Comcast to let you through).
I was saying that in general, not on a power line. A line has appreciable TL characteristics at a given frequency starting from when its length hits lambda/10, roughly. Since BPL uses VHF frequencies, of course it has TL characteristics on multi-mile power lines. However, if you transmit over a very short length of power line, say, a 1cm chunk, it will more or less *not* act as a transmission line. This is because the AC waveform doesn't have the 'space' to change appreciably from one end of the wire to the other (horrible explanation, I know, but workable). The point was that length is a factor in whether something behaves like a transmission line at a given frequency, not just the characteristics of the line.
As to the rest of it - you don't understand info theory. There is a tradeoff between bitrate (capacity), bandwidth, and power/SNR. There is a tradeoff between symbol rate and bits per symbol, within a limit. There is a tradeoff between achievable bits per symbol at a given error rate and SNR. Clever systems trade on these such that whichever quantities may be fixed (usually SNR and BW), they can achieve the best capacity, or alternately for a fixed capacity can minimize SNR or BW. Given a sufficiently powerful transmitter, you can achieve an arbitrary capacity on a powerline, even using a fixed frequency and bandwidth.
Hell, even a 50 Hz bandwidth can transmit gigabits of data - C=BW*log2(1+S/N). You want 1 megabit per second on a 50 Hz channel? You need SNR=2^(10^6/50)-1 (not in dB, just a ratio), or 120411 dB. Which is ridiculous for many reasons, but possible in theory. The problem that this would cause, which is what the hams are complaining about with BPL, is that you'll need to use huge amounts of power, which means you will spew huge amounts of radiated interference. Since BPL operates partially within ham frequencies, the interference screws them up.
Yes, I want to avoid patents on things which haven't been created. What benefit does patenting things that haven't been proven to work give that only allowing patents on working examples doesn't provide?
The sole exception are things which are simply too expensive for even a large government to build; obviously a working model is difficult to provide there. And I would argue that holding the patent on something like that is totally valueless, since no one can utilize the patent knowledge anyway.
If you invent a working solar sail, and have it in space, there are ways to verify it without putting it directly in the patent office. I'm not saying you need to give the patent office a prototype; rather, you need to provide them with proof of the prototype's existence, operation, and a way to verify that it does in fact do what you say it does, the way you say it does it.
Nanomachines? Come on. Anyone with the gear to get a working, patentable nanomachine also has the gear to properly secure it such that its easy to hand over to someone else to verify.
Deadly bioagent is easy. Put bioagent in glass vial. Give patent agent live monkey, glass box, glass vial. Carefully. Put live monkey in glass box. Put glass vial in glass box. Shake glass box until glass vial breaks. Monkey died? Congratulations, it works.
Killer computer virus - I assume you mean "destroys the computer system it runs on" and not "kills a person". If you're patenting a physically deadly computer virus, I can think of ways to test it, but none of them are nice (especially if the virus is only toxic to humans). If it simply destroys the computer it runs on, then that's easy - give them the virus and a spare computer. Also, I'd question the patentability of either one, though the physically deadly virus might have some merit in terms of how it affects the human.
Brain implant - if you've gotten to the point where you have a working brain implant, it should be no problem to provide a working example to the patent office, complete with attached live being.
I don't think the problems are as stiff as you think they are.
Slashdot Mirror
Under your interpretation, exactly how would *any* GPL Windows executable be legal to run? If it can be installed by the end user, why can't MS distribute it?
For that matter, under your interpretation, how can *any* GPLed Windows app legally exist, excepting of course the case of the non-dependent app? If the author of a GPLed program links to a non-GPL system library, then distributes it, aren't they violating the GPL themselves?
I'm certainly no lawyer, but my reading of the clause in question definitely doesn't include that clause extending the GPL onto software not otherwise covered. If GPLed Windows apps can legally exist, then I can't see how MS distributing them becomes any different than the author distributing them - that's one of the nice things about the GPL. If party A *legally* distributes application A as package A, then party B can legally distribute package A, no matter what.
Dear god.
Do yourself a favor - copy the Beta tapes over to some other format and sell the Beta as a collector's item. It's just ridiculous.
I think you're misinterpreting.
What that exemption says is, you have to distribute all source code - except, if some of the source code replicates a component normally distributed within the operating system, you're exempted from being required to provide the source code to that component, unless you're distributing the component as well. However, since distributing a GPL Windows executable with the system is generally going to count as mere aggregation, you were never under obligation to distribute source to the non-GPL sections of your product, and as such the exemption and the revocation of the exemption don't apply to it.
Basically, the clause is there to prevent you from having to distribute source to things that are shared with the OS. It can't extend the GPL beyond the core tenets of the GPL - unless the program links closely enough with Windows to invoke the clauses of the GPL on Windows, Windows itself remains seperate, and thus this clause doesn't apply.
Having lived with more than a couple of hippies, I can honestly say the following:
Most of them are good and kind. And smelly. God, are they smelly.
(Look, there's nothing wrong with them, really, but you have to allow the rest of us to make fun of them - they smell like patchouli, for chrissake!)
Hell, I'd have sold you my "secret menu" Apex for $100. Just turn that Macrovision shit off!
That you're still a dirty, filthy hippie?
It goes down past skin deep, my friend. You can't shower off your patchouli-stink.
Sure you can. It's called a Pismo, and with enough memory does just fine on OS X as well.
And if you happen to run into turbulence, think of the entertainment value you could receive by staggering into people!
No it didn't.
It started them on the browser, but anyone who's worked in electrical CAD applications could tell you they've been around a lot longer than that, and I wouldn't be surprised to find out the mouse gestures existed before ECAD picked it up.
Because IE crashed more than enough *without* unsupported addins, thank you.
Amtrak doesn't leave me with a vehicle at the other end.
No, the year digit is the sole issue here, since we're due for a rollover at the end of the decade.
Technically speaking, if we used a 4 digit year field, we'd still get a rollover, it would just be roughly 8000 years away.
I know you meant this to be funny, but seriously - long distance driving SUCKS, and I would be all kinds of into a system like this.
Small vehicles (maybe, say, electric-powered?) that can only do 40-60 miles on a charge, and a who-cares-how-its-powered system those small vehicles can piggyback onto that does long hauls.
Banks, universities, and credit card companies might not let you get away with it, but Blockbuster will absolutely not require it if you complain. Since most video stores require your account to be credit-card backed anyway, there's no reason for them to ask for your SSN, and no reason for you to give it to them.
Whatever. It's possible you understand info theory, in which case you have no idea how to write. I don't know which one you're a failure at, so I'll leave it at that.
Of course, the Atlas V main stage uses a Russian-designed motor, the RD-180.
As to failures... well, let's not talk about Shuttle vs. Soyuz or Delta III.
The Russians are very, very good rocket engineers. Generally, Russian rocket designs are considered to be the most reliable; however, their manufacturing and maintenance isn't always up to snuff with their design.
Considering that new American rockets are using Russian-designed rocket motors...
No, no it shouldn't.
Let's say, for theory's sake, that this netgear antenna plugged into a wall is where we are.
Who provides their upstream? Is it Comcast? SBC? Verizon? Qwest? Whoever it is, I bet the upstream has a SMTP server. Almost all of them will accept and route mail from clients on their network.
The provider of the pipe is the ISP. Not the antenna you're attached to, necessarily - in fact, not usually the antenna you're attached to. The provider of the antenna's pipe.
a) Take out a loan.
b) Find an investor.
c) Until you've run it on the thousands of computers, how do you know it would work? As far as I'm concerned, you don't deserve the patent until you've proven it works.
Actually, I'm fine with the concept of eliminating all IP law, which would be most fair to the little guy in todays day and age. But if we have to keep it, requiring working models and/or proof of operation isn't so bad.
You claim that coding methods require a good SNR, which isn't true; most of the more clever coding methods are designed to work at low SNR. The statement I object to is "clever methods that get the most bitrate out of the bandwidth require good SNR". It's false. Some codes actually operate worse at high SNR compared to an uncoded channel than at low SNR. Basically, there's a quantity called coding gain - for most codes, this is a function of bit-error-rate and describes how much lower your SNR can be compared to the uncoded channel for the same BER. There are codes where coding gain *decreases* with increasing SNR after a certain point - in fact, most turbo codes are of this type. Even more complex - most turbo codes display a negative coding gain at extremely low SNRs - they're *worse* than an uncoded channel! - quickly catch up to and pass the uncoded channel, so their coding gain increases, and then when they hit the 'error floor' the coding gain decreases again as the uncoded channel begins to catch up. At a certain point of SNR increase, the turbo code is no longer attractive, as its coding gain per its computational complexity ceases to be worthwhile.
5 0=1+S/N
You claim that bitrate is dependent solely on bandwidth, when its dependent on bandwidth and SNR jointly. "If you want to transfer a few megabits per second, you're going to end up with bandwidth on the order of a few megahertz." As I proved in my calculations, no, you're not. Want me to do it again?
Transferring 3 megabits per second over a 20kHz bandwidth.
C=B*log2(1+S/N)
3000000=20000*log2(1+S/N)
150=log2(1+S/N)
2^1
SNR = 903dB.
Practically speaking, you wouldn't want to try on anything less than a 300kHz channel, anything else would be power-prohibitive. But its possible.
Like I said, I see no evidence you actually understand info theory.
How much do you want to bet the mailserver provided by the ISP of the moment - the wifi provider - works?
Your ISP is the person providing your pipe, not the person you pay at home. If you aren't at home, you might need to use a different server - how hard is that to understand?
If you're a software developer and have a business case for direct port 25, here are your choices:
Convince Comcast to open port 25 for you. If you have a legit reason, they may. Of course, if you're trying to do business on a residential connection, they might not.
SSH or VPN to a corporate machine, where presumably you have port 25 unblocked, and can send that way.
Send it through Comcast's server.
I have no issue with there being limits, so long as those limits are not universal. If its something you need, pay for the access in one way or another (cash or time spent convincing Comcast to let you through).
I was saying that in general, not on a power line. A line has appreciable TL characteristics at a given frequency starting from when its length hits lambda/10, roughly. Since BPL uses VHF frequencies, of course it has TL characteristics on multi-mile power lines. However, if you transmit over a very short length of power line, say, a 1cm chunk, it will more or less *not* act as a transmission line. This is because the AC waveform doesn't have the 'space' to change appreciably from one end of the wire to the other (horrible explanation, I know, but workable). The point was that length is a factor in whether something behaves like a transmission line at a given frequency, not just the characteristics of the line.
As to the rest of it - you don't understand info theory. There is a tradeoff between bitrate (capacity), bandwidth, and power/SNR. There is a tradeoff between symbol rate and bits per symbol, within a limit. There is a tradeoff between achievable bits per symbol at a given error rate and SNR. Clever systems trade on these such that whichever quantities may be fixed (usually SNR and BW), they can achieve the best capacity, or alternately for a fixed capacity can minimize SNR or BW. Given a sufficiently powerful transmitter, you can achieve an arbitrary capacity on a powerline, even using a fixed frequency and bandwidth.
Hell, even a 50 Hz bandwidth can transmit gigabits of data - C=BW*log2(1+S/N). You want 1 megabit per second on a 50 Hz channel? You need SNR=2^(10^6/50)-1 (not in dB, just a ratio), or 120411 dB. Which is ridiculous for many reasons, but possible in theory. The problem that this would cause, which is what the hams are complaining about with BPL, is that you'll need to use huge amounts of power, which means you will spew huge amounts of radiated interference. Since BPL operates partially within ham frequencies, the interference screws them up.
Yes, I want to avoid patents on things which haven't been created. What benefit does patenting things that haven't been proven to work give that only allowing patents on working examples doesn't provide?
The sole exception are things which are simply too expensive for even a large government to build; obviously a working model is difficult to provide there. And I would argue that holding the patent on something like that is totally valueless, since no one can utilize the patent knowledge anyway.
If you invent a working solar sail, and have it in space, there are ways to verify it without putting it directly in the patent office. I'm not saying you need to give the patent office a prototype; rather, you need to provide them with proof of the prototype's existence, operation, and a way to verify that it does in fact do what you say it does, the way you say it does it.
Nanomachines? Come on. Anyone with the gear to get a working, patentable nanomachine also has the gear to properly secure it such that its easy to hand over to someone else to verify.
Deadly bioagent is easy. Put bioagent in glass vial. Give patent agent live monkey, glass box, glass vial. Carefully. Put live monkey in glass box. Put glass vial in glass box. Shake glass box until glass vial breaks. Monkey died? Congratulations, it works.
Killer computer virus - I assume you mean "destroys the computer system it runs on" and not "kills a person". If you're patenting a physically deadly computer virus, I can think of ways to test it, but none of them are nice (especially if the virus is only toxic to humans). If it simply destroys the computer it runs on, then that's easy - give them the virus and a spare computer. Also, I'd question the patentability of either one, though the physically deadly virus might have some merit in terms of how it affects the human.
Brain implant - if you've gotten to the point where you have a working brain implant, it should be no problem to provide a working example to the patent office, complete with attached live being.
I don't think the problems are as stiff as you think they are.