Imperative won. Functional and declarative lost. Give it up.
I've written sizable programs in LISP and Prolog. I've used formal specification languages. That era is over.
Even "makefiles", which were supposed to be a declarative language, have become an imperative language with some declarations. YACC and Bison are true declarative languages, as is SQL. But general-purpose declarative programming doesn't work.
Google is in the process of moving product-related sites over to Froogle. Froogle isn't built by looking at web pages; merchants upload product lists.
Those product lists contain prices, and Froogle, of course, can rank by price. So "search engine optimization" for Froogle consists of offering the lowest price.
Once Froogle gets rolling, online retailing is going to be about price competition.
The Swipe Project should sign up for Accurint, so that when you put in your card, you get your whole dossier. That would show people how much is known about them.
I have a Motorola phone with voice dialing.
It can be operated without looking at it, which is a good thing.
You shouldn't have to look at the phone to see what state it is in. Especially when driving.
There's a horrible tendency to put way too much state in the controls of portable devices. BMW, in a monumental fit of stupidity, did this to their car controls with their "i-Drive" system. They put a joystick-based GUI on most of the vehicle functions. This leads to what pilots call "too much head-down time".
It took years for the VCR/DVD player people to get this figured out, and now the cell phone people are botching it.
Tomorrow, the judge rules on whether SCO has "identified with specificity" the alleged infringements.
Magistrate Notice of Hearing
Motion hearing set for 10:00 2/6/04 for all pending motions:...
To be held before Judge Wells.
SCO has already been ordered by the judge to comply. That happened back in December. The order is below. Notice item 4. Tomorrow, the judge rules on whether they did comply.
Order granting [68-1] motion to compel discovery, granting [44-1] motion to compel Discovery.
The SCO Group is hereby ORDERED:
1) To respond fully and in detail to Interrogatory Nos. 1-9 as stated in IBM's First Set of Interrogatories.
2) To respond fully and in detail to Interrogatory Nos 12 and 13 as stated in IBM's Second Set of Interrogatories.
3) IBM is to provide SCO a list of requested documents as stated in IBM's First and Second Requests for the Production of Documents and SCO is to produce all requested documents.
4) To identify and state with specificity the source code(s) that SCO is claiming form the basis of their action against IBM. This is to include identification of all Bates numbered documents previously provided.
5) To the extent IBM's requests call for the production of documents or are met by documents SCO has already provided, SCO is to identify with specificity the location of responsive answers including identification of the Bates numbered documents previously provided if applicable.
6) If SCO does not have sufficient information in its possession, custody, or control to specifically answer any of IBM's requests that are the subject of this order, SCO shall provide an affidavit setting forth the full nature of its efforts, by whom they were taken, what further efforts it intends to utilize in order to comply, and the expected date of compliance. SCO is required to provide such answers and documents within thirty days from the date of this order. All other discovery, including SCO's Motion to Compel is hereby STAYED until this Court determines that SCO has fully complied with this Order.
In that one line of boldface above, the judge captured the key issue. No amount of PR spin control will help SCO in court tomorrow.
The bottom has fallen out of the encyclopedia market. Encyclopaedia Brittanica on DVD is only $14.95 at Fry's, after a mail-in rebate. Encarta has driven pricing through the floor.
I have an interest in the law, but no desire to be a lawyer. I know many lawyers, and I hire them on occasion. Other than trial lawyers, who are like fighter pilots, lawyers tend to either be boring or bored with their work.
There have been implementations with support for 8 windows, and that clearly wasn't enough. If you had support for, say, 1024, that probably would be enough, even if every icon took up a window slot. The transistor count for this isn't a big deal any more.
The point here is that we're tied to some architectural decisions from an era when transistors were more expensive, and those decisions are worth a new look.
Read the Electronic Communications Privacy Act. This may raise some questions, but sending a copy of this section of the ECPA back to the company is likely to result in some serious thinking about the issue. The ECPA only allows disclosure to the "addressee or intended recipient", or the "subscriber, in the case of remote computing service". Who's the subscriber here?
Clearly, though, you can obtain consent from the original addressee and then disclose.
Back in 1952, Collier's Magazine published a six-part series later called the Collier's Space Program. That series is credited with inspiring the US space program.
Those pictures are famous, and there's even an animated Disney documentary from the period.
The "Collier's space program" was far more ambitious than what's been done to date, or even what Cameron had drawn. The Collier's program had a big rotating space station in Earth orbit, a Mars rocket under construction in orbit, and heavy industrial traffic to and from orbit. Cameron has much lower ambitions.
There's considerable military interest in small, cheap sensor platforms, from robots to "smart dust". Most of this stuff doesn't work, and much of it founders on the problems of how to power the gadgets. But someday it probably will work.
Quantity has a quality all its own. Even if the stuff isn't all that effective on a per unit basis, it may become possible to overwhelm an enemy with sheer production power. We can't yet release millions of little robots in Afghanistan, all looking for bin Laden. But the first kills by robotic air vehicles have already happened there.
The future of war in cities and jungles may involve huge flocks of robotic birds. Most just watch. Some kill. All report back and work together.
The $4.50 "Comprehensive Report" includes
"Address Summary, Others
using SSN, Date/Locations where SSN Issued, Census Data, Bankruptcy Indicator,
Property Indicator and Corporate Affiliations Indicator, Bankruptcy, UCC Filings, Corporate Affiliations, Driver's Licenses,
Vehicle Registrations, Property, Merchant Vessels, FAA Pilots, FAA Aircraft,
Professional Licenses, Florida Accidents, Voter Registration, Hunting/Fishing
Permits, Concealed Weapons Permits, Associates, Relatives (3 Degrees), Neighbors,
Criminal Convictions and Sexual Offenders."
More advanced searches include
arrest data, gun licenses, property ownership, Internet domain name ownership, and a "Patriot Act Search".
Order now, and get the facts on anyone.
Much of this information has been available for some time, but never before has it beeen assembled into one convenient package available to anyone at a low price.
See product reviews, including "You can't hide from Accurint" and "No Place to Hide".
Now with XML support, a batch interface for bulk users, and 24 hour tech support!
Channelized I/O is probably a good idea, but it's either going to cost you some bandwidth (route all IO through a expanded version of current MMUs), or be expensive (a seperate MMU for IO).
It shouldn't hurt bandwidth. The problem with MMUs is latency, and adding a few hundred nanoseconds to I/O latency isn't going to hurt. I/O accesses have far more coherency than regular memory accesses, so you don't need that much cacheing within the I/O MMU.
The original Apollo Domain machines had an MMU between the CPU and the Multibus card cage for add-on peripherals. Apollo lost out to Sun for other reasons, their choice of a propretary token ring networking system being one of them.
Depends on the goal. Here's an architecture for reliability. If vendors had to pay whenever a program crashed, we would have seen this years ago.
Channelized I/O With current peripheral bus to memory interfaces, peripherals can store anywhere in memory. So drivers impact system stability. It doesn't have to be that way. IBM got this right in mainframe design in the 1960s.
You want an MMU between the peripherals and memory. Drivers then become non-privileged programs.
Existing peripherals don't even need to know there's an MMU in the middle, just as programs don't.
High-speed copy. Copying data in memory should be really fast, so fast that it's almost free, even if it takes copy-on-write hardware in the cache to do it. Why? Because then the temptation to put everything in one big address space decreases. With good interprocess communication (think QNX messaging, not CORBA, or horrors, SOAP), building programs out of components can actually work. This includes the OS. File systems, networking, and drivers should all be user programs.
The neat hardware implementation of this would be to make all MOV instructions take nearly the same time, regardless of the amount of data moved.
A MOV should result in a remapping of the source and destination memory in the cache system.
Even if this were just implemented for aligned moves, it would be a big help. When your application's 8K buffer needs to be copied to the file system, that copy should be done by updating cache control info, not by really doing it.
Graphics MMUs Get rid of the "window damage" approach, and have real hardware support for overlapping windows. All that's needed are big sprites. Then programs don't have to know or care which window is on top. "Overlay planes" do some of this, but it's not general enough.
With this, windowing becomes far simpler. Each window is maintained locally. Shared window management is reduced to screen space allocation, which is done by commanding the window MMU.
Looks like General Dynamics Robotics is back in business with this new contract. I was looking for a LIDAR for the DARPA Grand Challenge early last year, and the people at General Dynamics Robotics who did their last one had left the company.
The real problem with imaging LIDAR devices is that you can't make any money building them. Five companies have exited the field in the last decade. There are commercial markets for single-point rangefinders, and for line scanners, but true 3D devices to date have almost all been one-offs.
Most of the existing 3D devices contain rotating machinery. Some have two moving mirrors. This results in a bulky unit, because you need an inch or two of light-collecting aperture on the receive side, implying big moving mirrors inside. The General Dynamics Robotics prototype was a small linear array of laser rangerfinders fronted by a rotating prism, which got them down to one moving part. But it was still a big unit. The mechanisms used to date look too much like the ones used for mechanically scanned television in the 1930s. That's a dead end.
Flash LIDAR devices exist,
but have a basic problem. They must illuminate the whole field of vision, so the optical power requirement goes up as the fourth power of the range. (For point beams, it's only the square of the range.) So either they only work at night, like the Sea Lynx, they have very limited range, like the one from EFPL Zurich with a seven-meter limit, or they are not eye-safe, like the next-generation air-to-air missile seeker head currently in development for the USAF.
Despite this, we'll probably see a good solution in the next few years. It will take custom IC development. Then we'll have true 3D cameras.
And still nobody has managed to write a decent compiler for it.
That's because writing a good compiler for it is very, very hard. A few years ago, the HP team working on the Itanium compiler spoke at Stanford, and they were really discouraged. The compiler has to decide which instructions execute in parallel, which leads to very hard scheduling problems. Either the compiler has to look at profiling data, which is a pain for developers, or it has to be really good at guessing right on which branches get taken most often.
This is an example of the "Our program is Really Important" syndrome. The complicated install, with advertising. The installation of a new program group, several desktop icons, an entry in the system tray, and a background service. Advertising popups when the program runs ("There's more to Acrobat than the Reader"), at startup
("New updates are available for download"), and at random times ("You have used Advanced Turbo Accelerated Plus Pro Extreme for Windows XP for 189 days").
Usually, this nonsense only hits the user who installs the program, but anti-virus programs take it to the next level.
A nice goal for Open Source desktop software should be to eliminate this stuff. Open Source programs don't need this drivel; they're not selling anything.
I'm taking a break from filing 0.02" off a piece of aluminum. Next time, I allow a bigger tolerance.
There has a planet-sized virtual world, which the staff and users are slowly filling in. The system works fine, but they need more interesting places in their world. Lots more. There is hiring.
A fundamental question for American business, and their political servants, is this: Where does the buying power come from? As disposable income decreases, business activity will slow.
This is yet another vaporware "digital ink" product.
The contrast is lousy, it's low greyscale, the cost is high, the updated time is almost a second, the lifetime is short, it's expensive to make, and it's not really in production.
It's great that research is proceeding in this area, but the "real soon now" press releases from the "digital ink" people put them in line for the vaporware awards.
I've bought a number of machines and installed QNX on them, and always buy them with no OS installed. It's not that hard to get no-OS systems. I
get them from Solid Electric, a small systems house in Silicon Valley. They offfer "Fedora Linux" as a free install, or you can get a true no-OS PC with a blank hard drive.
I'm looking forward to the new release of QNX. More drivers, a modern browser, and other good stuff is coming. Windows CE customers are getting fed up with Microsoft and switching to QNX. (I've heard some amusing stories about Microsoft vs. Detroit: "No, you don't get to display the Microsoft logo every time the car starts".)
It's not a general purpose desktop OS yet, but for embedded applications, it's way ahead of everything else. (Windows CE is too flakey, VxWorks is too low-level, RTLinux isn't protected mode, and the user-level "real time" Linux variants aren't hard real time.)
The Coleman AirGen fuel cell power source, heavily hyped in 2000-2001, seems to have sunk without a trace. Coleman blames an unnamed supplier. Ballard claims to be selling the thing for industrial use but links to a distributor that doesn't mention it. Everything Ballard has seems to be a prototype.
There are people selling things that claim to be fuel cells, but are use-once devices. "Zinc air fuel cells" for cell phones are basically batteries.
Nobody really seems to have large fuel cells in production.
The hype in this area is becoming severe. Truly stupid ideas are being proposed, like using electricity to crack water to get hydrogen to be shipped by pipeline to fixed generating plants to be burned to make electricity. Sounds like something Enron would come up with.
Slashdot Mirror
I've written sizable programs in LISP and Prolog. I've used formal specification languages. That era is over.
Even "makefiles", which were supposed to be a declarative language, have become an imperative language with some declarations. YACC and Bison are true declarative languages, as is SQL. But general-purpose declarative programming doesn't work.
Those product lists contain prices, and Froogle, of course, can rank by price. So "search engine optimization" for Froogle consists of offering the lowest price. Once Froogle gets rolling, online retailing is going to be about price competition.
The Swipe Project should sign up for Accurint, so that when you put in your card, you get your whole dossier. That would show people how much is known about them.
Posted article too soon. Sorry. It's Friday, not Wednesday.
There's a horrible tendency to put way too much state in the controls of portable devices. BMW, in a monumental fit of stupidity, did this to their car controls with their "i-Drive" system. They put a joystick-based GUI on most of the vehicle functions. This leads to what pilots call "too much head-down time".
It took years for the VCR/DVD player people to get this figured out, and now the cell phone people are botching it.
-
Magistrate Notice of Hearing
...
To be held before Judge Wells.
SCO has already been ordered by the judge to comply. That happened back in December. The order is below. Notice item 4. Tomorrow, the judge rules on whether they did comply.Motion hearing set for 10:00 2/6/04 for all pending motions:
The SCO Group is hereby ORDERED:
In that one line of boldface above, the judge captured the key issue. No amount of PR spin control will help SCO in court tomorrow.
Why develop a free one?
So if you have a free access point, it's OK, but if you charge, you're a "CyberCafe", and have to have a security guard and video surveillance.
Now find every WAP in Garden Grove that bills, and insist on strict enforcement.
I have an interest in the law, but no desire to be a lawyer. I know many lawyers, and I hire them on occasion. Other than trial lawyers, who are like fighter pilots, lawyers tend to either be boring or bored with their work.
The point here is that we're tied to some architectural decisions from an era when transistors were more expensive, and those decisions are worth a new look.
Clearly, though, you can obtain consent from the original addressee and then disclose.
Those pictures are famous, and there's even an animated Disney documentary from the period.
The "Collier's space program" was far more ambitious than what's been done to date, or even what Cameron had drawn. The Collier's program had a big rotating space station in Earth orbit, a Mars rocket under construction in orbit, and heavy industrial traffic to and from orbit. Cameron has much lower ambitions.
There's considerable military interest in small, cheap sensor platforms, from robots to "smart dust". Most of this stuff doesn't work, and much of it founders on the problems of how to power the gadgets. But someday it probably will work.
Quantity has a quality all its own. Even if the stuff isn't all that effective on a per unit basis, it may become possible to overwhelm an enemy with sheer production power. We can't yet release millions of little robots in Afghanistan, all looking for bin Laden. But the first kills by robotic air vehicles have already happened there.
The future of war in cities and jungles may involve huge flocks of robotic birds. Most just watch. Some kill. All report back and work together.
The $4.50 "Comprehensive Report" includes "Address Summary, Others using SSN, Date/Locations where SSN Issued, Census Data, Bankruptcy Indicator, Property Indicator and Corporate Affiliations Indicator, Bankruptcy, UCC Filings, Corporate Affiliations, Driver's Licenses, Vehicle Registrations, Property, Merchant Vessels, FAA Pilots, FAA Aircraft, Professional Licenses, Florida Accidents, Voter Registration, Hunting/Fishing Permits, Concealed Weapons Permits, Associates, Relatives (3 Degrees), Neighbors, Criminal Convictions and Sexual Offenders." More advanced searches include arrest data, gun licenses, property ownership, Internet domain name ownership, and a "Patriot Act Search".
Order now, and get the facts on anyone.
Much of this information has been available for some time, but never before has it beeen assembled into one convenient package available to anyone at a low price. See product reviews, including "You can't hide from Accurint" and "No Place to Hide".
Now with XML support, a batch interface for bulk users, and 24 hour tech support!
If you have a problem with that, tough.
It shouldn't hurt bandwidth. The problem with MMUs is latency, and adding a few hundred nanoseconds to I/O latency isn't going to hurt. I/O accesses have far more coherency than regular memory accesses, so you don't need that much cacheing within the I/O MMU.
The original Apollo Domain machines had an MMU between the CPU and the Multibus card cage for add-on peripherals. Apollo lost out to Sun for other reasons, their choice of a propretary token ring networking system being one of them.
The neat hardware implementation of this would be to make all MOV instructions take nearly the same time, regardless of the amount of data moved. A MOV should result in a remapping of the source and destination memory in the cache system. Even if this were just implemented for aligned moves, it would be a big help. When your application's 8K buffer needs to be copied to the file system, that copy should be done by updating cache control info, not by really doing it.
With this, windowing becomes far simpler. Each window is maintained locally. Shared window management is reduced to screen space allocation, which is done by commanding the window MMU.
The real problem with imaging LIDAR devices is that you can't make any money building them. Five companies have exited the field in the last decade. There are commercial markets for single-point rangefinders, and for line scanners, but true 3D devices to date have almost all been one-offs.
Most of the existing 3D devices contain rotating machinery. Some have two moving mirrors. This results in a bulky unit, because you need an inch or two of light-collecting aperture on the receive side, implying big moving mirrors inside. The General Dynamics Robotics prototype was a small linear array of laser rangerfinders fronted by a rotating prism, which got them down to one moving part. But it was still a big unit. The mechanisms used to date look too much like the ones used for mechanically scanned television in the 1930s. That's a dead end.
Flash LIDAR devices exist, but have a basic problem. They must illuminate the whole field of vision, so the optical power requirement goes up as the fourth power of the range. (For point beams, it's only the square of the range.) So either they only work at night, like the Sea Lynx, they have very limited range, like the one from EFPL Zurich with a seven-meter limit, or they are not eye-safe, like the next-generation air-to-air missile seeker head currently in development for the USAF.
Despite this, we'll probably see a good solution in the next few years. It will take custom IC development. Then we'll have true 3D cameras.
That's because writing a good compiler for it is very, very hard. A few years ago, the HP team working on the Itanium compiler spoke at Stanford, and they were really discouraged. The compiler has to decide which instructions execute in parallel, which leads to very hard scheduling problems. Either the compiler has to look at profiling data, which is a pain for developers, or it has to be really good at guessing right on which branches get taken most often.
Usually, this nonsense only hits the user who installs the program, but anti-virus programs take it to the next level.
A nice goal for Open Source desktop software should be to eliminate this stuff. Open Source programs don't need this drivel; they're not selling anything.
I'm taking a break from filing 0.02" off a piece of aluminum. Next time, I allow a bigger tolerance.
There has a planet-sized virtual world, which the staff and users are slowly filling in. The system works fine, but they need more interesting places in their world. Lots more. There is hiring.
The answer seems to be "downhill". Most job growth is in low-paying sectors. The U.S. Department offers this guide to anticipated job growth through 2010. Top job growth areas are "Combined food preparation and serving workers, including fast food", and "Customer service representatives".
Would you like fries with that?
A fundamental question for American business, and their political servants, is this: Where does the buying power come from? As disposable income decreases, business activity will slow.
Recall that SCO is lobbying Congress for restrictions on open source. This will help their lobbying.
It's great that research is proceeding in this area, but the "real soon now" press releases from the "digital ink" people put them in line for the vaporware awards.
I'm looking forward to the new release of QNX. More drivers, a modern browser, and other good stuff is coming. Windows CE customers are getting fed up with Microsoft and switching to QNX. (I've heard some amusing stories about Microsoft vs. Detroit: "No, you don't get to display the Microsoft logo every time the car starts".) It's not a general purpose desktop OS yet, but for embedded applications, it's way ahead of everything else. (Windows CE is too flakey, VxWorks is too low-level, RTLinux isn't protected mode, and the user-level "real time" Linux variants aren't hard real time.)
The Coleman AirGen fuel cell power source, heavily hyped in 2000-2001, seems to have sunk without a trace. Coleman blames an unnamed supplier. Ballard claims to be selling the thing for industrial use but links to a distributor that doesn't mention it. Everything Ballard has seems to be a prototype.
There are people selling things that claim to be fuel cells, but are use-once devices. "Zinc air fuel cells" for cell phones are basically batteries.
Nobody really seems to have large fuel cells in production.
The hype in this area is becoming severe. Truly stupid ideas are being proposed, like using electricity to crack water to get hydrogen to be shipped by pipeline to fixed generating plants to be burned to make electricity. Sounds like something Enron would come up with.