No company would attempt FDA approval in those circumstances. We'd have a million fantastic basic-biotech observations a year, but no resources to turn them into real tools of medicine. Applied biotechnology research would cease to exist.
The only prevention of this calamity of non-innovation is the promise of a limited-time patent on such drugs. Pfizer will pay for a hundred FDA approval trials because it gets exclusive access to the drug market for a short window. - And that window is, indeed, short: by the time a drug hits the market, probably 15 years of its patent have already passed.
There's another model that could work. Since drugs are so tightly controlled by the FDA, don't rely on patenting them, rely on FDA permission to manufacture and market them.
An example of this can be seen in aviation. An aircraft is certified by the FAA, but that certification is very strict. You basically can not modify the aircraft or the way it is flown from the way it was approvied (which is however the manufacturer got it certified by testing and proving it to be safe). If you do want to make modifications, you can get approval from the FAA to do testing, and if you can prove that your changes are safe then YOU (and only you) can authorize others to do the same modification, and you can charge whatever you want for that authorization. For example, many aircraft can be safely flown on ordinary unleaded automobile gasp;ome; however, they are rarely certified to use it. If you purchase an "STC" (Supplemental Type Certificate) from someone who has done FAA-approved testing and shown it to be safe, you can now use auto gas as well. The only thing you get when you pay your fee (usually based on horsepower of the engine) is some paperwork and a sticker you post next to the gas cap that says auto gas is allowed.
ANYONE else can do the same exact testing, and get the same authorization. Thus, the price is limited on what you can charge based on how much that testing will cost someone else. However, the length of time is forever, and often whoever did the original modification did so for their own purpose, and being able to sell that authorization to others is a bonus.
Now, it is true that knowing ahead of time that a certain drug indeed works can eliminate false starts, which can save significant money. However, it seems to me that the original company still gets a good time advantage, plus if it never runs out, they can compete with any newcomers by lowering the price closer and closer to production margin. In addition, it is valuable to have re-testing of drugs, as the original studies are sometimes found later to have been flawed.
Finally, a large amount of the cost basis for drugs is from the marketing of that drug, not the actual development and testing. If, indeed, such marketing is so effective as to justify the consumer being forced to pay for it (as part of the price of the drug), then the companies spending that money shouldn't have to worry so much about an upstart for a while. If it isn't all that effective, then they should stop spending so much on it and lower the price of the drugs!
I'm not arguing that drugs should or should not be patentable. I'm arguing that there are other models that can be used to achieve the same goal of rewarding innovation that the patent process is supposed to do.
Turning on VM allows the program to start up faster, as it uses demand paging from the executable file. Without it, the entire executable must be loaded into memory. That's why it takes more memory as well. You can enable VM with as little as 1MB more than actual RAM. This only applies up to MacOS9, MacOSX uses more conventional swapping.
Programs like RAMDoubler enabled you to turn on VM, but without turning on swap. The main function of RD was, however, to compress memory instead of swapping it out. For a fast processor and a slow disk, that can be faster than swapping.
Of course, there are other ways of doing that. One is to turn on overcommiting memory (which was the default once-upon-a-time). The other is to use vfork(), which basically says "I'm just gonna exec(), so don't bother allocating for me all of my parent's memory space". I know at one time in Linux, vfork() was the same as fork(), but at least in current kernels/glibc, it definitely is different.
Now the problem with overcommit is if you actually try to use memory that was speculatively allocated to you, but now is not available. In that case, a random process can get killed, which is usually undesireable. But there's not much downside to vfork(), as long as you do in fact exec() right afterwards.
When I know they're not going to accept an answer, I'll just ask them what answer they want to hear; or, if they ask a yes/no question ("Are you running Windows 95?") the response will be "Sure, why not?", even though I'm running Linux or Mac OS or whatever. Then I just have to simulate what Windows 95 would do in response to their questions and directions until we get to the point where they accept that there's a problem on their end. This can even work if you've already admitted you're using something they don't want to deal with. Just say "Poof. As if by magic, I'm now running Windows 95. Now let's start over in your script."
Ever since modems went higher than 2400 bps, with various protocols for compression and reliability built in, the actual data transmission (over the phone line) has not included start/stop bits. The transmission between the modem and the computer does, of course, when using asynch transmission modes. Combined with compression, that's why you want the data speed on the serial line to be higher than the transmission rate over the phone line, combined with a flow control mechanism (x-on/off or rts/dts).
Things would have been so much simpler if Hayes hadn't been so successful and modem control lines had been used. In particular, if synchronous transmission (specifically, SDLC) along with a variable clock rate, had become standardized, all of the garbage of trying to packetize frames over SLIP/PPP, all of the headaches (including patents) of +++, all of the hassle of trying to figure out interface speeds by looking at the bit pattern of A and T, and not noticing that a connection had dropped because the "CARRIER DROPPED" came out in the middle of a packet, would have been eliminated. Transmit clock, receive clock, RTS, CTS, DCD, and use DTR to signal between data and sending configuration commands. Combine with RS-422 signalling for better noise resistance and Ethernet might never have needed to be invented. Just using SDLC with a self-clocking protocol would have been a major win, as frames are checksummed and start/stop bits don't need to be sent (the overhead of flags and beginning/end of frame is irrelevant as when the amount of data goes up, the overhead drops as low as necessary). It works fine as an "asynchronous" protocol, i.e. interactive typing.
Why? Is it awful when Microsoft is required to turn over internal memos and e-mail describing how they're going to exploit their monopoly position? If someone at SCO sent e-mail to Bill Gates asking for money to "fight the Linux blight", do you think IBM should be able to compel Microsoft to disclose that e-mail?
From the wording in the subpoena, I don't see why staff time couldn't be charged to the SCO lawyers, as long as they were reasonable rates (i.e. NOT like lawyers charging $1000/hr, just their usual salary). If not, then challenge the subpoena as being too much of a burden. At least, that's what the subpoena says:
(1) A party or an attorney responsible for the issuance and service of a subpoena shall take reasonable steps to avoid imposing undue burden or expense on a person subject to that subpoena. The court on behalf of which the subpoena was issued shall enforce this duty and impose upon the party or attorney in breach of this duty an appropriate sanction which may include, but is not limited to, lost earnings and reasonable attorney's fee.
It then goes on to say that you can object to it within 14 days, and in such a case the subpoena is blocked unless the court issues an order, and
such an order to comply production shall protect any person who is not a party or an officer of a party from significant expense resulting from the inspection and copying commanded.
The court "shall quash or modify the subpoena if it... subjects a person to undue burden". If they still really really want the information, the court will require that reasonable compensation be made. That doesn't include "just copying costs" (although I'd think that "copying costs" would reasonably include wear & tear on the machine (cost of copier divided by expected lifetime), ink/toner, paper, and salary of the person doing the copying).
For someone like the FSF, formally objecting to a subpoena should be pretty easy and inexpensive (couple hundred bucks). All you have to do is object, you don't have to lay out a big legal argument, just make "a written objection", and if you claim that the subpoena is unreasonable on the face of it, and the court agrees, you can get attorney's fees and lost earnings back.
They're not even trying to do that. Didn't anyone RTFA? They say that if what their patent covers becomes a part of an IETF standard, that anyone can use it as long as they aren't suing Cisco for a patent violation.
If this standard is adopted, Cisco will not assert any patents owned or controlled by Cisco against any party for making, using, selling, importing
or offering for sale a product that implements the standard, provided,
however that Cisco retains the right to assert its patents (including the
right to claim past royalties) against any party that asserts a patent it
owns or controls (either directly or indirectly) against Cisco or any of
Cisco's affiliates or successors in title for an implementation of any IETF
standard; and Cisco retains the right to assert its patents against any
product or portion thereof that is not necessary for compliance with the
standard.
But they aren't patenting translucent windows. They are patenting using translucent windows in a very specific way.
Now, I happen to think that patents like this are still too broad (although this one is close), but at least get it straight what it is they are trying to patent!
So if that works, then obviously Microsoft made a mistake selling it for $10,000. They should have sold it in the first place for $14.95 if there's that much of a market for it at that price. They'd make more money that way. Remember, there is plenty of value to customers to getting it from the source, and getting support from the source. A repackager is going to mostly only get those customers who wouldn't have gotten the product in the first place, or would have simply "pirated" it, or the repackager is going to have to put significant resources into supporting it, at which point they can't so easily undercut the price. GPL makes a free market "more free".
If Microsoft decides that by not allowing others to freely redistribute it they can make more money (less the cost of developing those things they would have gotten for no cost by using GPL'ed software as a base), then they obviously should go with a proprietary solution instead of GPL. Of course, if they go that route and find they can't sell it anyway because others have provided a freely available solution and are sharing enhancements, then they lose. As they should.
The author of the paper makes a reasonable case to consider the GPL as a binding an enforceable contract.
The only "reasonable case" the author makes is that
According to Linuxplanet.com columnist Dennis E.Powell,"software licenses are generally considered to be contracts."
However, he makes no argument as to whether it should be considered a contract or a license, nor what the difference would be.
Describing the consideration required in a contract, he says
In the case of the GPL, the users of the software gain consideration in the form of rights that are granted to them by the copyright holder of the software,while the copyright holder gains a promise from its users to follow the terms and conditions set by the GPL.
This has two problems. One, the granting of those exclusive rights that someone holds under copyright is a license, that's one of the rights of a copyright holder. Two, merely doing what you agreed to do in a contract is not consideration. The specifics of what you agreed to do determines whether it is compensation or not. Is the act of distributing any changes you make, and not restricting the rights of people you distribute to, "compensation"? Since you had no right to distribute those changes in the first place, you aren't giving up anything when they allow you to do so. Your changes have no value whatsoever (at least for the purposes covered by the GPL) unless they allow you to distribute them. I don't see how that could be considered compensation. In addition, you aren't even providing it to whoever is granting you rights, you are providing it to whoever you distribute it to,
As a contract, it has some problems. As a license, it works just fine. The author refers to it throughout as a license. It calls itself a license. The only argument you can make that it isn't a license is if it fails to meet the requirements of a license.
Just because a court hasn't yet ruled on the GPL doesn't make it unenforceable, nor does attempting to enforce it under those conditions make it FUD. It is also not the case that the court not ruling on it makes all arguments as to what it is equally valid. I think we can all agree that it isn't a brick of green cheese, yet the court hasn't yet ruled that it isn't. Weight the arguments on their merits.
Clear-cutting business wins in the short term, even if the government doesn't subsidize it. The forested lands are not "valuable" for anything other than clear-cutting, at least by short-sighted values. The price of new forest land won't go up until most of it is gone, because it has little "economic value". The clear-cutting lumber company doesn't care, they're making their profits now. Eventually they may be forced into a new model, but that doesn't have anything to do with whether they are adversely affected by the results of their clear-cutting. The problem is that their costs are decoupled from the "true" costs. Adam Smith's invisible hand is supposed to take into account hidden costs, but only when there is a closed loop. Open up the loop and it doesn't lead to an optimal solution. This is more related to game theory's "tragedy of the commons".
Where do you get the figure of 100 gallons of diesel fuel to produce 75 gallons of biodiesel? Or did you mean ethanol?
We had some figures earlier from a farmer that indicated it took around 3-10 gallons of (presumably) diesel to plant, fertilize, and harvest 1 acre of corn, yielding about 125-150 bushels of corn; each bushel yields about 2.5 gallons of ethanol, or a bit over 300 gallons/acre. The "tractor" portion of the energy equation is clearly not 100 gallons of diesel to produce 75 gallons of ethanol.
That same acre of corn can yield around 30 gallons of biodiesel, even without high-oil-yield corn (around 4% by weight). There's some energy cost to producing that as well, but again, the cost for planting and harvesting it is clearly not a huge percentage of the yield.
Even transporting the harvest is not that big a deal. A 10-ton truck will be carrying around 425 bushels, I believe (which can yield about 1000 gallons of ethanol and 100 gallons of biodiesel. How much diesel does it take to travel a few hundred miles in such a truck?
BTW, a big problem with some of the earlier analyses of how much energy it takes to produce ethanol are that they assume you need to vaporize a large quantity of water/alcohol in the distilling process and that you can't recover any of that energy. You can, of course, recover much of that energy, using the steam to pre-heat the incoming stock. That's not even looking at other possibilities, such as using solar power for heating (with much higher efficiency, when compared to using sunlight to either directly produce electricity and use that for heating, or grow corn and extract energy from that).
What we really need, of course, is an organism that takes sunlight and directly produces ethanol (or methanol, or methane, or whaever) from water and CO2 from the air. Run that through a solar still and you'd have a very effective transformation of sunlight into stored energy.
In fact, if you look at the detailed description of the "invention", they even seem to refer to it themselves as only a possible naming convention:
It should be noted that in one or more embodiments of the invention, every member of the group for which the invention is implemented is provided with a website URL and an e-mail address. In one or more embodiments of the invention, website URL's and e-mail addresses are established using a parallel form. For example, in one or more embodiments, e-mail addresses are created in the general form "[first initial][second initial][lastname]@interfaceserver.com", and website URL's are established in the general form "[first initial][second initial][lastname].interfaceserver.com" (this form of URL, i.e. xxx.xxx.com, is sometimes referred to as a "third level domain name"). For example, for an embodiment in which the group for which the invention is implemented comprises doctors admitted to practice medicine in the U.S., and for which a URL for the interface server is "everymd.com", the e-mail address for a doctor whose name is John H. Smith may be established as "jhsmith@everymd.com" and the corresponding website URL may be established as "jhsmith.everymd.com". The e-mail address and URL are thus virtually identical. The only difference is that in the URL the "@" symbol of the e-mail address is replaced with a period ("."). Establishing URL's and e-mail addresses in this manner provides easy to remember URL/e-mail address pairs. In a more general embodiment, the URL's are established for members of the group in the form "name@subdomain.domain", while e-mail addresses are established in the form "name@subdomain.domain".
There's no particular reason that this form is any better or worse than, e.g. name@subdomain.domain and http://www.subdomain.domain/name. Since the system described is specifically intercepting everything you send via e-mail, and re-directing you to websites, the exact form of the naming convention is particularly useless. Besides, to the extent that keeping the two looking almost exactly alike might be useful simply makes it that much more obvious, especially given the format of the contact information in an SOA record in DNS (i.e. there is specific prior art, that one skilled in the art would know, that explicitly presents that exact correspondence - that should satisfy even the ridiculous "obviousness" test used by the patent office and the courts). So, it seems to me it would have to be at least one of useless or obvious, and hence not patentable.
In addition, the patent abstract, summary and details don't seem to teach the claimed invention at all. I have to wonder if this is even a valid patent at all, as the actual claims are only referenced in the above "note". There are only two claims in this patent, right, I'm not looking at a truncated set of claims? I thought the detailed description had to actually describe the invention claimed, and the abstract had to at least have some connection to the claims.
Also interesting is the part near the end. Beyond the usual boilerplate of "other embodiments will be obvious to those skilled in the art", it includes a specific additional embodiment example that is clearly not related to the claime invention in the slightest:
For example, in one or more embodiments, the interface is provided in the form of a voice mail menu system accessed by a user using a telephone instead of a computer or other internet access device.
As the claims specifically refer to URLs (although presented as hostnames) and e-mail addresses, and specifically how the hostname and e-mail address are typographically represented, I don't see how this could apply to using it through a telephone.
What it looks to me happened is they had a whole lot of claims, matching the description and figures that are there, and all of their claims got thrown out; they finally scrounged up their naming convention, and the idea of giving identities to members of a grou
There are some interesting and somewhat difficult puzzles in Final Fantasy. Many of them are the equivalent of state machines with quite a few states, quite a few actions, and only one halting condition.
It would be interesting to see how fast Windows emulation is, once they come out with a version that works on the G5. Even more interesting would be when it emulates Opteron runmomg 64-bit Windows.
It isn't that difficult to move to 64 bits. It is primarily going through code and making sure that long and int are properly used, and that pointers aren't assumed to be the same size as int. Abstracting data types makes this a lot easier.
However, what is difficult is to make a single OS binary that will run on both 32-bit and 64-bit hardware, and on 64-bit hardware support both 32-bit and 64-bit applications (i.e. have a compatibility mode that makes sure not to allocate memory beyond the range of a 32-bit pointer) and take advantage of 64-bit instructions when appropriate.
Much easier is to make sure the OS saves all the registers as 64 bits on context switches, which allows 64-bit programs to run, albeit in a limited memory space, add some code to check register parameters on system calls to make sure they are only 32 bits (i.e. upper 32 bits have same value as sign bit of lower 32 bits). That's minimal to allow the OS to run on 64-bit hardware and allow applications to access 64-bit registers.
Then you can modify the virtual memory system to allow for more physical memory (which in itself means making sure that all kernel references to user-space and buffer memory go through 64-bit-safe subroutines, and that drivers don't do stupid things with memory mapped i/o and such). Does Panther even allow for more than 4GB physical memory yet?
Once the whole OS is 64-bit safe, it should be pretty easy to add in a 32-bit compatibility mode, which would modify virtual memory mapping to keep everything down low, and allow 64-bit programs to use the full address space (16 exabytes, I believe, or exibytes if you prefer). Making the rest of the kernel take advantage of 64-bit processing when appropriate can then begin (although at that point you probably end up with completely separate kernels for 32-bit and 64-bit - i.e. 10.4 might be the last 32-bit version available, with 10.5 being full 64-bit, although 32-bit applications would still be supported).
I have some numbers (coincidentally, using the CDC 6600 mentioned above). Environment is an Alpha system running (30 year old) Cyber code in a production environment. For much of the code, it has been (statically) translated to Alpha native instructions (although the Cyber architecture pretty much requires straight emulations, as code is often self-modifying, including instruction returns). The translated code runs 5-10 times faster than the emulated code. The emulator is very tight, and straight emulation runs about 20 times faster on recent Alpha processors than they did on a Cyber 830.
Incidentally, "code bloat" factor (ratio of Alpha code to original Cyber code) is around 5 (much of it due to handling the difference between 64-bit 2's complement and 60-bit 1's complement). In the other direction, no-op instructions (of various types) are optimized out. Floating point and population count instructions are the only ones that make a subroutine call from translated code (and floating point doesn't use native floating point at all, except using non-portable kludges for normalization).
I don't know what you mean that Replay doesn't have a season pass thing. There are three types of recording slots: Show, Channel, and single-record; each can be either guaranteed or non-guaranteed. Show based recording is a particular show in a particular time slot (can restrict based on day of week). It will automatically shift if the show's time changes and is contiguous with whatever show is actually on at the specified time. Single-record is the same as Show, except it is for a particular date only. You can switch an entry between the two easily. For Show, you can specify the number of shows to record, with the oldest one being overwritten. A Channel is based on a search, and you specify how many hours to record. You can guarantee it, but that is just a space guarantee and boosts priority over non-guaranteed recording.
Guarantee with Show and Single (and Manual) both reserves space and prevents conflicts (unless a show shifts in its timeslot, in which case you could have a conflict with another guaranteed show). Time reserved for guaranteed shows, but not yet used, can be used by non-guaranteed shows.
I don't know if newer units have changed; mine is an old 30-hour Panasonic Showstopper. Got it for $200, with a $100 rebate, with lifetime service. I recently upgraded it to a 100GB drive (and it's MUCH quieter and starts up much faster now!). However, the SELECT button on the remote is causing problems. I wonder if I can get it replaced under the Circuit City extended warranty I got for it.
I wouldn't give kudos to MS for inventing it, it is popular because it is ubiquitous, not because it has a high goodness/simplicity ratio, and of course that popularity is what leads to it being even more popular, due to network effect. An abstract file-system interface for devices like these would probably be much more useful, and let the devices implement the actual file system in a manner suited to the media.
Priority date can be earlier than application date, due to things like continuations. Priority date is what counts as far as invalidating a patent goes - if anyone else published or publically used the invention before the priority date, or if the inventor published or used the invention more than one year before the priority date, that would invalidate the patent.
Also, older patents were valid 17 years from date of issue, not from date of application. Newer patents are valid for 20 years from date of application. This helps prevent repeated multiple continuations and resubmissions dragging the patent process out for many years, then finally being granted after its already been re-invented and everyone is already using it. There are also problems with the way the scope of a patent can be expanded in a continuation, keeping the earlier priority date but including newer ideas (including ideas that other people invented after that priority date!).
I bought a Netgear wireless router from Best Buy, it had three rebates for a total of $50, one from BB, the other two from Netgear; both of the Netgear rebates required the original UPC. Best Buy rebate support told me I could send a copy for one of them. So far, I've received the rebate from BB and one of the ones from Netgear, but not the other, so now I have to call them.
I also bought a D-Link ethernet card from BB, $5 after a $5 rebate. Got a rejection from D-Link, saying "not purchased during offer period". They keypunched in the postmark date for the purchase date (and the postmark date). Called them, they said to send them a copy of all materials, I circled the receipt date and wrote in the original postmark date, got my rebate check 2 weeks later. But why couldn't they check real-time when doing date entry for invalid entries, so the person could verify and correct immediately? Probably because it doesn't pay them to be careful, as many people won't be careful enough to keep materials to re-submit. If there was a law requiring them to triple the rebate if they reject it improperly, maybe they'd be a bit more careful (similar to some laws regarding incorrect prices when scanning at a register).
I also never got a rebate on a Western Digital 100GB hard drive. The rebate had made the difference in which drive to buy. I called WD, they said it had been sent out two weeks earlier, but that I had to wait 40 days before they could re-issue it. I guess it's time to call them again.
There was an old, old startup program for the original Macs that gave a greeting based on the date and/or time ("Welcome to Macintosh, but you should be in bed"). It was fully customizable with a text file, and used MacInTalk (or whatever the text-to-speech software was). I still have an old MacPlus startup diskette with a cool startup screen, plays a bit of Money For Nothing, then runs the greeting program.
There was also a game called "The Colony" which had a standard decoration throughout of old dried-up sticks in tubs. On Christmas, they turned into Christmas trees. That was a great game, by the way, I'd love to see it on something other than a MacPlus (even running on a Mac II the enemies ran too fast, no real-time based motion). There was supposed to be a color version coming out, I sent in my $5, but never heard of it again.
Someone already mentione Hit & Run, but Simpson's Road Rage also gives you a special car/driver - got it for Christmas, and there's a Christmas Apu. Probably other dates, but haven't checked.
The GPL says "Accompany it with the information you received as to the offer to distribute corresponding source code." Information can not be copyrighted.
Not just to protect against malicious code, but to allow you to modify the way the program works. For example, I have a Linksys Wireless-G Router, $69 after rebate. They also make a Wireless-G Bridge, for $169. It has less hardware, and I see no reason why the Router hardware couldn't perform the same function (connecting to an access point, rather than being an access point). I'm definitely interested in looking at the source for both devices to see how it can be modified.
Slashdot Mirror
There's another model that could work. Since drugs are so tightly controlled by the FDA, don't rely on patenting them, rely on FDA permission to manufacture and market them.
An example of this can be seen in aviation. An aircraft is certified by the FAA, but that certification is very strict. You basically can not modify the aircraft or the way it is flown from the way it was approvied (which is however the manufacturer got it certified by testing and proving it to be safe). If you do want to make modifications, you can get approval from the FAA to do testing, and if you can prove that your changes are safe then YOU (and only you) can authorize others to do the same modification, and you can charge whatever you want for that authorization. For example, many aircraft can be safely flown on ordinary unleaded automobile gasp;ome; however, they are rarely certified to use it. If you purchase an "STC" (Supplemental Type Certificate) from someone who has done FAA-approved testing and shown it to be safe, you can now use auto gas as well. The only thing you get when you pay your fee (usually based on horsepower of the engine) is some paperwork and a sticker you post next to the gas cap that says auto gas is allowed.
ANYONE else can do the same exact testing, and get the same authorization. Thus, the price is limited on what you can charge based on how much that testing will cost someone else. However, the length of time is forever, and often whoever did the original modification did so for their own purpose, and being able to sell that authorization to others is a bonus.
Now, it is true that knowing ahead of time that a certain drug indeed works can eliminate false starts, which can save significant money. However, it seems to me that the original company still gets a good time advantage, plus if it never runs out, they can compete with any newcomers by lowering the price closer and closer to production margin. In addition, it is valuable to have re-testing of drugs, as the original studies are sometimes found later to have been flawed.
Finally, a large amount of the cost basis for drugs is from the marketing of that drug, not the actual development and testing. If, indeed, such marketing is so effective as to justify the consumer being forced to pay for it (as part of the price of the drug), then the companies spending that money shouldn't have to worry so much about an upstart for a while. If it isn't all that effective, then they should stop spending so much on it and lower the price of the drugs!
I'm not arguing that drugs should or should not be patentable. I'm arguing that there are other models that can be used to achieve the same goal of rewarding innovation that the patent process is supposed to do.
Turning on VM allows the program to start up faster, as it uses demand paging from the executable file. Without it, the entire executable must be loaded into memory. That's why it takes more memory as well. You can enable VM with as little as 1MB more than actual RAM. This only applies up to MacOS9, MacOSX uses more conventional swapping.
Programs like RAMDoubler enabled you to turn on VM, but without turning on swap. The main function of RD was, however, to compress memory instead of swapping it out. For a fast processor and a slow disk, that can be faster than swapping.
Of course, there are other ways of doing that. One is to turn on overcommiting memory (which was the default once-upon-a-time). The other is to use vfork(), which basically says "I'm just gonna exec(), so don't bother allocating for me all of my parent's memory space". I know at one time in Linux, vfork() was the same as fork(), but at least in current kernels/glibc, it definitely is different.
Now the problem with overcommit is if you actually try to use memory that was speculatively allocated to you, but now is not available. In that case, a random process can get killed, which is usually undesireable. But there's not much downside to vfork(), as long as you do in fact exec() right afterwards.
When I know they're not going to accept an answer, I'll just ask them what answer they want to hear; or, if they ask a yes/no question ("Are you running Windows 95?") the response will be "Sure, why not?", even though I'm running Linux or Mac OS or whatever. Then I just have to simulate what Windows 95 would do in response to their questions and directions until we get to the point where they accept that there's a problem on their end. This can even work if you've already admitted you're using something they don't want to deal with. Just say "Poof. As if by magic, I'm now running Windows 95. Now let's start over in your script."
Ever since modems went higher than 2400 bps, with various protocols for compression and reliability built in, the actual data transmission (over the phone line) has not included start/stop bits. The transmission between the modem and the computer does, of course, when using asynch transmission modes. Combined with compression, that's why you want the data speed on the serial line to be higher than the transmission rate over the phone line, combined with a flow control mechanism (x-on/off or rts/dts).
Things would have been so much simpler if Hayes hadn't been so successful and modem control lines had been used. In particular, if synchronous transmission (specifically, SDLC) along with a variable clock rate, had become standardized, all of the garbage of trying to packetize frames over SLIP/PPP, all of the headaches (including patents) of +++, all of the hassle of trying to figure out interface speeds by looking at the bit pattern of A and T, and not noticing that a connection had dropped because the "CARRIER DROPPED" came out in the middle of a packet, would have been eliminated. Transmit clock, receive clock, RTS, CTS, DCD, and use DTR to signal between data and sending configuration commands. Combine with RS-422 signalling for better noise resistance and Ethernet might never have needed to be invented. Just using SDLC with a self-clocking protocol would have been a major win, as frames are checksummed and start/stop bits don't need to be sent (the overhead of flags and beginning/end of frame is irrelevant as when the amount of data goes up, the overhead drops as low as necessary). It works fine as an "asynchronous" protocol, i.e. interactive typing.
Why? Is it awful when Microsoft is required to turn over internal memos and e-mail describing how they're going to exploit their monopoly position? If someone at SCO sent e-mail to Bill Gates asking for money to "fight the Linux blight", do you think IBM should be able to compel Microsoft to disclose that e-mail?
From the wording in the subpoena, I don't see why staff time couldn't be charged to the SCO lawyers, as long as they were reasonable rates (i.e. NOT like lawyers charging $1000/hr, just their usual salary). If not, then challenge the subpoena as being too much of a burden. At least, that's what the subpoena says:
It then goes on to say that you can object to it within 14 days, and in such a case the subpoena is blocked unless the court issues an order, and
The court "shall quash or modify the subpoena if it ... subjects a person to undue burden". If they still really really want the information, the court will require that reasonable compensation be made. That doesn't include "just copying costs" (although I'd think that "copying costs" would reasonably include wear & tear on the machine (cost of copier divided by expected lifetime), ink/toner, paper, and salary of the person doing the copying).
For someone like the FSF, formally objecting to a subpoena should be pretty easy and inexpensive (couple hundred bucks). All you have to do is object, you don't have to lay out a big legal argument, just make "a written objection", and if you claim that the subpoena is unreasonable on the face of it, and the court agrees, you can get attorney's fees and lost earnings back.
Where did it say they were banned from talking about it?
They're not even trying to do that. Didn't anyone RTFA? They say that if what their patent covers becomes a part of an IETF standard, that anyone can use it as long as they aren't suing Cisco for a patent violation.
But they aren't patenting translucent windows. They are patenting using translucent windows in a very specific way.
Now, I happen to think that patents like this are still too broad (although this one is close), but at least get it straight what it is they are trying to patent!
So if that works, then obviously Microsoft made a mistake selling it for $10,000. They should have sold it in the first place for $14.95 if there's that much of a market for it at that price. They'd make more money that way. Remember, there is plenty of value to customers to getting it from the source, and getting support from the source. A repackager is going to mostly only get those customers who wouldn't have gotten the product in the first place, or would have simply "pirated" it, or the repackager is going to have to put significant resources into supporting it, at which point they can't so easily undercut the price. GPL makes a free market "more free".
If Microsoft decides that by not allowing others to freely redistribute it they can make more money (less the cost of developing those things they would have gotten for no cost by using GPL'ed software as a base), then they obviously should go with a proprietary solution instead of GPL. Of course, if they go that route and find they can't sell it anyway because others have provided a freely available solution and are sharing enhancements, then they lose. As they should.
The only "reasonable case" the author makes is that
However, he makes no argument as to whether it should be considered a contract or a license, nor what the difference would be.Describing the consideration required in a contract, he says
This has two problems. One, the granting of those exclusive rights that someone holds under copyright is a license, that's one of the rights of a copyright holder. Two, merely doing what you agreed to do in a contract is not consideration. The specifics of what you agreed to do determines whether it is compensation or not. Is the act of distributing any changes you make, and not restricting the rights of people you distribute to, "compensation"? Since you had no right to distribute those changes in the first place, you aren't giving up anything when they allow you to do so. Your changes have no value whatsoever (at least for the purposes covered by the GPL) unless they allow you to distribute them. I don't see how that could be considered compensation. In addition, you aren't even providing it to whoever is granting you rights, you are providing it to whoever you distribute it to,As a contract, it has some problems. As a license, it works just fine. The author refers to it throughout as a license. It calls itself a license. The only argument you can make that it isn't a license is if it fails to meet the requirements of a license.
Just because a court hasn't yet ruled on the GPL doesn't make it unenforceable, nor does attempting to enforce it under those conditions make it FUD. It is also not the case that the court not ruling on it makes all arguments as to what it is equally valid. I think we can all agree that it isn't a brick of green cheese, yet the court hasn't yet ruled that it isn't. Weight the arguments on their merits.
Clear-cutting business wins in the short term, even if the government doesn't subsidize it. The forested lands are not "valuable" for anything other than clear-cutting, at least by short-sighted values. The price of new forest land won't go up until most of it is gone, because it has little "economic value". The clear-cutting lumber company doesn't care, they're making their profits now. Eventually they may be forced into a new model, but that doesn't have anything to do with whether they are adversely affected by the results of their clear-cutting. The problem is that their costs are decoupled from the "true" costs. Adam Smith's invisible hand is supposed to take into account hidden costs, but only when there is a closed loop. Open up the loop and it doesn't lead to an optimal solution. This is more related to game theory's "tragedy of the commons".
Where do you get the figure of 100 gallons of diesel fuel to produce 75 gallons of biodiesel? Or did you mean ethanol?
We had some figures earlier from a farmer that indicated it took around 3-10 gallons of (presumably) diesel to plant, fertilize, and harvest 1 acre of corn, yielding about 125-150 bushels of corn; each bushel yields about 2.5 gallons of ethanol, or a bit over 300 gallons/acre. The "tractor" portion of the energy equation is clearly not 100 gallons of diesel to produce 75 gallons of ethanol.
That same acre of corn can yield around 30 gallons of biodiesel, even without high-oil-yield corn (around 4% by weight). There's some energy cost to producing that as well, but again, the cost for planting and harvesting it is clearly not a huge percentage of the yield.
Even transporting the harvest is not that big a deal. A 10-ton truck will be carrying around 425 bushels, I believe (which can yield about 1000 gallons of ethanol and 100 gallons of biodiesel. How much diesel does it take to travel a few hundred miles in such a truck?
For some real numbers, check this 2002 U.S. Department of Agriculture study, it goes into great detail on methodology, assumptions, etc, and includes such things as the energy cost of producing fertilizer. For biodiesel, see An Overview of Biodiesel and Petroleum Diesel Life Cycles.
BTW, a big problem with some of the earlier analyses of how much energy it takes to produce ethanol are that they assume you need to vaporize a large quantity of water/alcohol in the distilling process and that you can't recover any of that energy. You can, of course, recover much of that energy, using the steam to pre-heat the incoming stock. That's not even looking at other possibilities, such as using solar power for heating (with much higher efficiency, when compared to using sunlight to either directly produce electricity and use that for heating, or grow corn and extract energy from that).
What we really need, of course, is an organism that takes sunlight and directly produces ethanol (or methanol, or methane, or whaever) from water and CO2 from the air. Run that through a solar still and you'd have a very effective transformation of sunlight into stored energy.
In fact, if you look at the detailed description of the "invention", they even seem to refer to it themselves as only a possible naming convention:
There's no particular reason that this form is any better or worse than, e.g. name@subdomain.domain and http://www.subdomain.domain/name. Since the system described is specifically intercepting everything you send via e-mail, and re-directing you to websites, the exact form of the naming convention is particularly useless. Besides, to the extent that keeping the two looking almost exactly alike might be useful simply makes it that much more obvious, especially given the format of the contact information in an SOA record in DNS (i.e. there is specific prior art, that one skilled in the art would know, that explicitly presents that exact correspondence - that should satisfy even the ridiculous "obviousness" test used by the patent office and the courts). So, it seems to me it would have to be at least one of useless or obvious, and hence not patentable.
In addition, the patent abstract, summary and details don't seem to teach the claimed invention at all. I have to wonder if this is even a valid patent at all, as the actual claims are only referenced in the above "note". There are only two claims in this patent, right, I'm not looking at a truncated set of claims? I thought the detailed description had to actually describe the invention claimed, and the abstract had to at least have some connection to the claims.
Also interesting is the part near the end. Beyond the usual boilerplate of "other embodiments will be obvious to those skilled in the art", it includes a specific additional embodiment example that is clearly not related to the claime invention in the slightest:
As the claims specifically refer to URLs (although presented as hostnames) and e-mail addresses, and specifically how the hostname and e-mail address are typographically represented, I don't see how this could apply to using it through a telephone.
What it looks to me happened is they had a whole lot of claims, matching the description and figures that are there, and all of their claims got thrown out; they finally scrounged up their naming convention, and the idea of giving identities to members of a grou
There are some interesting and somewhat difficult puzzles in Final Fantasy. Many of them are the equivalent of state machines with quite a few states, quite a few actions, and only one halting condition.
It would be interesting to see how fast Windows emulation is, once they come out with a version that works on the G5. Even more interesting would be when it emulates Opteron runmomg 64-bit Windows.
It isn't that difficult to move to 64 bits. It is primarily going through code and making sure that long and int are properly used, and that pointers aren't assumed to be the same size as int. Abstracting data types makes this a lot easier.
However, what is difficult is to make a single OS binary that will run on both 32-bit and 64-bit hardware, and on 64-bit hardware support both 32-bit and 64-bit applications (i.e. have a compatibility mode that makes sure not to allocate memory beyond the range of a 32-bit pointer) and take advantage of 64-bit instructions when appropriate.
Much easier is to make sure the OS saves all the registers as 64 bits on context switches, which allows 64-bit programs to run, albeit in a limited memory space, add some code to check register parameters on system calls to make sure they are only 32 bits (i.e. upper 32 bits have same value as sign bit of lower 32 bits). That's minimal to allow the OS to run on 64-bit hardware and allow applications to access 64-bit registers.
Then you can modify the virtual memory system to allow for more physical memory (which in itself means making sure that all kernel references to user-space and buffer memory go through 64-bit-safe subroutines, and that drivers don't do stupid things with memory mapped i/o and such). Does Panther even allow for more than 4GB physical memory yet?
Once the whole OS is 64-bit safe, it should be pretty easy to add in a 32-bit compatibility mode, which would modify virtual memory mapping to keep everything down low, and allow 64-bit programs to use the full address space (16 exabytes, I believe, or exibytes if you prefer). Making the rest of the kernel take advantage of 64-bit processing when appropriate can then begin (although at that point you probably end up with completely separate kernels for 32-bit and 64-bit - i.e. 10.4 might be the last 32-bit version available, with 10.5 being full 64-bit, although 32-bit applications would still be supported).
I have some numbers (coincidentally, using the CDC 6600 mentioned above). Environment is an Alpha system running (30 year old) Cyber code in a production environment. For much of the code, it has been (statically) translated to Alpha native instructions (although the Cyber architecture pretty much requires straight emulations, as code is often self-modifying, including instruction returns). The translated code runs 5-10 times faster than the emulated code. The emulator is very tight, and straight emulation runs about 20 times faster on recent Alpha processors than they did on a Cyber 830.
Incidentally, "code bloat" factor (ratio of Alpha code to original Cyber code) is around 5 (much of it due to handling the difference between 64-bit 2's complement and 60-bit 1's complement). In the other direction, no-op instructions (of various types) are optimized out. Floating point and population count instructions are the only ones that make a subroutine call from translated code (and floating point doesn't use native floating point at all, except using non-portable kludges for normalization).
I don't know what you mean that Replay doesn't have a season pass thing. There are three types of recording slots: Show, Channel, and single-record; each can be either guaranteed or non-guaranteed. Show based recording is a particular show in a particular time slot (can restrict based on day of week). It will automatically shift if the show's time changes and is contiguous with whatever show is actually on at the specified time. Single-record is the same as Show, except it is for a particular date only. You can switch an entry between the two easily. For Show, you can specify the number of shows to record, with the oldest one being overwritten. A Channel is based on a search, and you specify how many hours to record. You can guarantee it, but that is just a space guarantee and boosts priority over non-guaranteed recording.
Guarantee with Show and Single (and Manual) both reserves space and prevents conflicts (unless a show shifts in its timeslot, in which case you could have a conflict with another guaranteed show). Time reserved for guaranteed shows, but not yet used, can be used by non-guaranteed shows.
I don't know if newer units have changed; mine is an old 30-hour Panasonic Showstopper. Got it for $200, with a $100 rebate, with lifetime service. I recently upgraded it to a 100GB drive (and it's MUCH quieter and starts up much faster now!). However, the SELECT button on the remote is causing problems. I wonder if I can get it replaced under the Circuit City extended warranty I got for it.
I wouldn't give kudos to MS for inventing it, it is popular because it is ubiquitous, not because it has a high goodness/simplicity ratio, and of course that popularity is what leads to it being even more popular, due to network effect. An abstract file-system interface for devices like these would probably be much more useful, and let the devices implement the actual file system in a manner suited to the media.
Priority date can be earlier than application date, due to things like continuations. Priority date is what counts as far as invalidating a patent goes - if anyone else published or publically used the invention before the priority date, or if the inventor published or used the invention more than one year before the priority date, that would invalidate the patent.
Also, older patents were valid 17 years from date of issue, not from date of application. Newer patents are valid for 20 years from date of application. This helps prevent repeated multiple continuations and resubmissions dragging the patent process out for many years, then finally being granted after its already been re-invented and everyone is already using it. There are also problems with the way the scope of a patent can be expanded in a continuation, keeping the earlier priority date but including newer ideas (including ideas that other people invented after that priority date!).
I bought a Netgear wireless router from Best Buy, it had three rebates for a total of $50, one from BB, the other two from Netgear; both of the Netgear rebates required the original UPC. Best Buy rebate support told me I could send a copy for one of them. So far, I've received the rebate from BB and one of the ones from Netgear, but not the other, so now I have to call them.
I also bought a D-Link ethernet card from BB, $5 after a $5 rebate. Got a rejection from D-Link, saying "not purchased during offer period". They keypunched in the postmark date for the purchase date (and the postmark date). Called them, they said to send them a copy of all materials, I circled the receipt date and wrote in the original postmark date, got my rebate check 2 weeks later. But why couldn't they check real-time when doing date entry for invalid entries, so the person could verify and correct immediately? Probably because it doesn't pay them to be careful, as many people won't be careful enough to keep materials to re-submit. If there was a law requiring them to triple the rebate if they reject it improperly, maybe they'd be a bit more careful (similar to some laws regarding incorrect prices when scanning at a register).
I also never got a rebate on a Western Digital 100GB hard drive. The rebate had made the difference in which drive to buy. I called WD, they said it had been sent out two weeks earlier, but that I had to wait 40 days before they could re-issue it. I guess it's time to call them again.
There was an old, old startup program for the original Macs that gave a greeting based on the date and/or time ("Welcome to Macintosh, but you should be in bed"). It was fully customizable with a text file, and used MacInTalk (or whatever the text-to-speech software was). I still have an old MacPlus startup diskette with a cool startup screen, plays a bit of Money For Nothing, then runs the greeting program.
There was also a game called "The Colony" which had a standard decoration throughout of old dried-up sticks in tubs. On Christmas, they turned into Christmas trees. That was a great game, by the way, I'd love to see it on something other than a MacPlus (even running on a Mac II the enemies ran too fast, no real-time based motion). There was supposed to be a color version coming out, I sent in my $5, but never heard of it again.
Someone already mentione Hit & Run, but Simpson's Road Rage also gives you a special car/driver - got it for Christmas, and there's a Christmas Apu. Probably other dates, but haven't checked.
The GPL says "Accompany it with the information you received as to the offer to distribute corresponding source code." Information can not be copyrighted.
Not just to protect against malicious code, but to allow you to modify the way the program works. For example, I have a Linksys Wireless-G Router, $69 after rebate. They also make a Wireless-G Bridge, for $169. It has less hardware, and I see no reason why the Router hardware couldn't perform the same function (connecting to an access point, rather than being an access point). I'm definitely interested in looking at the source for both devices to see how it can be modified.