The "Wiring" board is kind of hard to get, but take a look at this $54 board. AtMega128, 2-line LCD, four buttons, relay, LED, buzzer, and connectors for most of the micro controller's interfaces. If that can be made to work with Wiring,
you'd have a nice little self-contained board that students can learn on.
I have not followed Arduino lately and don't know whether it already supports interrupts.
They've made some progress. The IDE now supports interrupts, and Wiring has moved up to the AtMega 128. Actually, they support most of the Atmel line. You can use their IDE with Atmel boards other than the Wiring board, but you may have to make configuration files that describe the other board's pinouts.
It's a classic open source problem. The applications he needed were there. But they didn't work. They almost worked. That's typical of open source projects, where there's a strong tendency to get stuck at "version 0.9x". The big pieces work, but nobody has done the grunt work of cleaning up all the known defects, making the code and documentation agree, and testing for usability problems. Open source projects, even major ones, can be stuck at that point for years.
Now this is where all those companies that "resell" open source code, like Red Hat and Novell, should be working. The community should beat on them to finish the job. That's the price they pay for reselling free software.
Take a look at Wiring. This is a microcontroller development system for artists. There's a board with an Atmel microcontroller and I/O interfaces. There's an integrated development environment with a simplified programming language. All open source. IDE runs on Linux, MacOS, or Windows.
The neat thing about this system is that it uses a modern microcontroller, the Atmel ATMega128, with 128K of memory and a 16MHz clock. This is a substantially more powerful machine than the PIC or the Basic STAMP. The development environment is really GCC, generating machine code; it's not an interpreter. So you can actually do non-trivial work on the thing.
All this is packaged up with an IDE and a set of documentation designed to be used standalone; all you need is their big download to install the IDE, and their web site to tell you how to use the board. You don't have to dump a set of Atmel manuals and a GCC manual on the user.
Yeah, clueless. It's been a long time since either a phone number or an IP address was like a street address.
The distinction is how "locative" an ID is. Seat numbers in a stadium are locative. But few other IDs are completely locative any more.
At one time, phone numbers really were locative; the first three digits specified the central office, or for larger offices, the switch within the CO, and the last four digits were the line number within the switch. That dates from the era when phone numbers were
read like "PLaza 5-1000". But that was a long time ago. Now there's a lookup in the middle; the number goes into a database and a location comes out. That allows for number portability, and also means all the numbers can be used.
The same thing happened to IP addresses. At one time, you could route by tearing apart IP addresses. In the beginning, network 10 was the ARPANET, and the last byte was the IMP number. Now, it's lookup-based, and routers have huge tables.
This is a continuing struggle with numbering plans, from zip codes to Ethernet addresses to UPC codes. They tend to start out locative, but eventually become a flat, arbitrary space as the demands on the number space increase and things change over time.
This is good. When legitimate businesses realize there are consequences for doing anything that looks like spamming, they'll be more careful.
We're actually winning on spam. There aren't that many different spammers left.
The volume of spam is way up because of botnets, but the number of different incoming spams is way down. Notice how few different spams you get.
The remaining spammers are all committing multiple felonies. Notice that when one of them gets caught, which happens a few times a year now, they go to jail. For years.
Jeremy James - 9 years in prison.
David Lin - 2 years in prison.
Howard Carmack - 3.5 years in prison.
Charles Fry - 1 year in prison (plea bargain).
Jeffery Goodin - Federal prisoner #42837-112, release date unknown.
Peter Francis-Macrae - UK spammer, 6 years.
Nelson Barrero - Federal prisoner #14049-004, served 3 years.
Isn't shared memory intrinsically slow, because access has to be interleaved between all the processors sharing it?
In practice, no. There are clever techniques to get around that, most of which involve caches.
Shared memory between modern multiprocessors is to some extent an illusion, one created by very clever cache hardware design. Each processor has its own cache with local memory, and each cache controller has some information about what the other processors have in their caches. When one processor changes a memory location and another process needs to read it, the cache controllers communicate and update the caches. That slows things down, but it doesn't happen too often. This is called "cache coherency", which you can look up to learn more about the subject. The end result is that it looks to programs like all the processors share the same memory, but most of the time they're really accessing local copies of main memory.
Re:Need a radar reflector in shipping lanes
on
Jim Gray Is Missing
·
· Score: 1
He might not have radar, but he certainly should have a radar reflector.
Radar visibility helps, but the big container ships and oil tankers have such limited maneuverability that they can't be expected to try to avoid agile little sailboats.
"A vessel of less than 20 meters in length or a sailing vessel shall not impede the safe passage of a power-driven vessel following a traffic lane." (Rule 10(j)). This is the marine equivalent of "no pedestrians on the freeway".
I've met some of the architects of the Cell processor, and they have a "build it and they will come" attitude. They've designed the computer; it's up to others to make it useful. This is probably not going to fly.
The Cell is a non-shared memory multiprocessor with quite limited memory per processor. There's only 256K per processor, which takes us back to before the 640K IBM PC. There are DMA channels to a bigger memory, but no cacheing. Architecturally, it's very retro; it's very similar to the NCube of the mid-1980s. It's not even superscalar. Cell processors are dumb RISC engines, like the old low-end MIPS machines. They clock fast, but not much gets done per clock.
Yes, you get lots of CPUs, but that may not help. On a server, what are you going to run in a Cell? Not your Java or Perl or Python server app; there's not enough memory. No way will an instance of Apache fit. You could put a copy of the TCP/IP stack in a Cell, but that's not where the CPU time goes in a web server. One IBM document suggests putting "XML acceleration" (i.e. XML parsing) in the server, but that's an answer looking for a problem. It might be useful for streaming video or audio; that's a pipelined process. If you need to compress or decompress or transcode or decrypt, the Cell might be useful. But for most web services, those jobs are done once, not during playout. Even MPEG4 compression might be too much for a Cell; you need at least two frames of storage, and it doesn't have enough memory for that.
Now if they had, say, 16MB per CPU, it might be different.
The track record of non-shared memory supercomputers is terrible. There's a long history of dead ends, from the ILLIAC IV to the BBN Butterfly to the NCube to the Connection Machine. They're easy to design and build, but just not that useful for general purpose computing. Some volumetric simulation problems, like weather prediction, structural analysis, and fluid dynamics can be crammed into those machines, so there are jobs for them, but the applications are limited.
Shared-memory microprocessors look much more promising as general purpose computers.
Having eight or sixteen CPUs in a shared-memory multicore configuration is quite useful. That's how SGI servers worked, and they had a good track record. Scaling up today's multicore shared-memory CPUs is repeating that idea, but smaller and cheaper.
At some point, you have to go to non-shared memory, but that doesn't have to happen until you hit maybe 16 CPUs sharing a few gigabytes of memory, which is about when the cache interconnects start to choke and speed of light lag to the far side of the RAM starts to hurt. That might even be pushed harder; there's been talk of 80 CPUs in a shared memory configuration. That's optimistic. But we know 16 will work; SGI had that years ago.
Then you go to a cluster on a chip, which is also well understood.
These things are really x86 PCs, with upwards of 64MB of memory. They're quite capable computers. The Neoware C50 is a desktop Linux system with no hard drive, for $259. The Wyse S50 is another comparable Linux box. Wyse even has a dual screen model. The HP model runs Debian. HP is having a sale - buy 3, get one free.
Neoware even has a thin client notebook computer. It's only useful when it has a WLAN connection. This is promoted as a security feature; if it's stolen, there's not much data in it.
This may be the way Linux comes to the enterprise desktop. To many companies, this is a cheaper and easier conversion than moving to Vista.
It's good to see ICANN doing some cleanup. For the past few years, they've been something of a trade group for domain registrars.
A few more TLDs could go..museum and.aero could be phased out due to lack of interest. The entire list for.museum is a few pages, the domains aren't the top-tier museums, and almost all of them are redirects anyway..aero has an entry for every airport code (try LAX.AERO), but those were put there by the domain registrar to give the illusion of activity and they're not the primary domain name for those sites. ("LAX.AERO" is really "WWW2.LAWA.ORG").
.biz ought to go as slum clearance..info probably wasn't worth creating.
Possibly run down by a larger ship
on
Jim Gray Is Missing
·
· Score: 5, Informative
He went out in a 40-foot C&C 121 yacht. That's a very nice boat, with a epoxy resin laminate hull, carbon fiber reinforcement and masts, Kevlar sails, and a 38HP engine. There hasn't been any weather lately bad enough to give a boat like that any serious trouble. If it ran aground it would probably survive the experience.
But between San Francisco and the Farralon Islands is a major shipping lane. One with fog. Container ships and oil tankers come through there. Sizable fishing boats have been run down and sunk without anyone on a large ship even noticing. There's a USGS Vessel Traffic Service station and established traffic lanes for large ships, but small boats aren't required to check in with traffic control.
Buy Vista. Put up with all the nonsense. Know that Microsoft will probably Tivo you at some point, taking away some functionality. Expect downtime due to authorization problems.
Keep running Windows 2000 and retain control of your system. No support, not compatible with many new devices, won't play much content, but a solid system.
Switch to a Mac, the other closed system. Everything from Apple works; third party software is kind of thin.
Run Linux on the desktop. It's almost ready for the desktop, like it has been for five years now.
Those are the options. And they all suck.
This is an opportunity for somebody. Probably somebody in China.
I find it amusing that the CSS zealots out there will rant about making sites that are pure CSS
I know. There's such hype about CSS layout. The truth is, the CSS machinery for layout is dumber than the table machinery. With CSS, the basic mechanisms for layout are "float", "clear", explicit sizing, and absolute positioning. Because the first two are weaker than tables, absolute positioning is overused. Then you get things like text appearing on top of other text, something that never happens with tables.
Dreamweaver had table-based layout working great years ago. Most of the complaints about tables come from people writing the tags by hand, but with a proper layout tool, it works very well.
CSS layout should have been constraint-based. You should be able to express concepts like "align bottom edge of this box with bottom edge of previous box" and "align left edge of this box with right edge of previous box". Then you could handle situations like the dreaded "3-column problem" cleanly. The right way to do it would be to allow arbitrary linear constraints, such as "box must have same width as previous box" and "box must have 0.5 width of enclosing box", dump them all into a linear constraint engine, crank the simplex algorithm, and get out a layout that satisfies all the constraints.
If you allow constraints with respect to non-adjacent boxes, you can do grid-based layout, too, and do all the things tables do. You'd want to do layout like this in a GUI tool, not by writing constraints by hand, of course. Drag box corners to other corners and edges and have them snap into lock.
This isn't a new idea; it's called "parametric CAD" in engineering design. But it's something the designers of CSS probably didn't know about.
It can act as usb mass storage and thus be carrying infected files.
Did they export the whole drive, or what? One would expect they'd export an empty file system, onto which one could load music or other content, and provide some means to reset it to empty if it became corrupted.
What is this thing running? Some consumer-grade version of Windows, apparently.
Clearly, not only is the virus "on the hard drive", but it actually gets executed.
You'd think they'd at least run XP Embedded, or Windows CE, with only the necessary components present. I don't expect a read-only image of QNX in ROM in a non-critical consumer infotainment device, but this is lame.
There's considerable unhappiness in San Francisco about Lucasfilm's operation. It's in the Presidio, which used to be a military base and is now a national park. It's the only national park which has to make a profit, due to a Bush Administration deal. Letterman Army Hospital was torn down to make room for the Lucasfilm facility. The San Francisco Bay Guardian complains about this constantly, as they try to keep the Presidio from turning into an industrial park. The Lucasfilm move to the Presidio was something of a dot-com boom excess, when people thought SF was the place to be.
Pixar, in Emeryville, Tippett, in Berkeley, and Dreamworks, in Redwood City, are the innovative animation companies in the Bay Area. And of course, there's EA, SCEA, and some other game companies. Lucasfilm doesn't seem to get much attention.
There are data centers in San Francisco proper with far more storage, too. The Internet Archive has several petabytes of storage. There's a large colocation facility at the 6th St. offramp from I-280.
Every year, I see these "Linux is ready for the desktop" articles. But it never happens.
Back in 2004, WalMart offered a $499 Linux laptop. They don't do that any more. Lenovo, HP, and Dell have fooled around with Linux laptops, but try to order one on line. Search for "linux laptop" on Dell, and you get back "Dell recommends Windows Vista(TM) Business." There are some off-brand Linux laptops available, but they're overpriced.
Linux on the desktop looked closer three years ago than it does now.
Much of this new stuff sounds like features of QNX. QNX has a "sporadic scheduler", for when you need things like 10ms of CPU every 100ms. QNX has had I/O cancellation for years. In QNX, you can set a timeout on any system call that blocks. If you set a 35ms timeout on a write, after 36 milliseconds, you'll have control back. Very useful in real-time systems where you're doing something less important, like logging, that should never take very long but, in some trouble condition, might. QNX has had prioritized I/O for years, too.
It all works, too. I've done compiles on QNX while running a real time program on the same machine, without the real time program missing a deadline.
Everyone assumed that Apple's $499/$599 prices for the iPhone was subsidized by Cingular.
Who thought that? At that price, it had better be profitable.
Cell phone internals are becoming very cheap. Check out the Texas Instruments "LoCosto" two-chip cell phone. Manufacturing costs are approaching $20. This isn't being reflected in the prices seen at US carrier's retail outlets, though. The handset price is inflated there, then "discounted" to compel users to sign up for "plans".
That's coming from Froogle. Companies that sign up with Froogle or Google Checkout have some additional info about them. Try, say, Super Warehouse, which Google describes as "Online retailer of color laser printers, laptops, hard drives, LCD monitors, and digital cameras". That text isn't from the "www.superwarehouse.com" web page, which starts out "Printers - Scanners - Toner - Monitors - Projectors & More at Super Warehouse".
It worked for the DARPA Grand Challenge, but not in the way most people think.
The prize was the carrot. But there was a stick, too. The Grand Challenge was a real threat to robotics funding at CMU and Stanford, which had been getting DARPA money for decades but were progressing very slowly. Originally, neither university's robotics group intended to enter. But there were apparently hints that if the non-university entries did significantly better than the people DARPA had been funding, the funding for the big robotics labs would be in jeopardy.
At Stanford, the management of the AI Lab had to be replaced to get real results. That was done, and things picked up quite a bit. There was a significant breakthrough in robotic vision for the Stanford vehicle. That was the real payoff in all this.
The CMU vehicles were what someone above called a "point solution" - the searchlight-sized gyro-stabilized gimbal with a line scanning LIDAR was a technological dead end.
Slashdot Mirror
The "Wiring" board is kind of hard to get, but take a look at this $54 board. AtMega128, 2-line LCD, four buttons, relay, LED, buzzer, and connectors for most of the micro controller's interfaces. If that can be made to work with Wiring, you'd have a nice little self-contained board that students can learn on.
They've made some progress. The IDE now supports interrupts, and Wiring has moved up to the AtMega 128. Actually, they support most of the Atmel line. You can use their IDE with Atmel boards other than the Wiring board, but you may have to make configuration files that describe the other board's pinouts.
It's a classic open source problem. The applications he needed were there. But they didn't work. They almost worked. That's typical of open source projects, where there's a strong tendency to get stuck at "version 0.9x". The big pieces work, but nobody has done the grunt work of cleaning up all the known defects, making the code and documentation agree, and testing for usability problems. Open source projects, even major ones, can be stuck at that point for years.
Now this is where all those companies that "resell" open source code, like Red Hat and Novell, should be working. The community should beat on them to finish the job. That's the price they pay for reselling free software.
Take a look at Wiring. This is a microcontroller development system for artists. There's a board with an Atmel microcontroller and I/O interfaces. There's an integrated development environment with a simplified programming language. All open source. IDE runs on Linux, MacOS, or Windows.
The neat thing about this system is that it uses a modern microcontroller, the Atmel ATMega128, with 128K of memory and a 16MHz clock. This is a substantially more powerful machine than the PIC or the Basic STAMP. The development environment is really GCC, generating machine code; it's not an interpreter. So you can actually do non-trivial work on the thing.
All this is packaged up with an IDE and a set of documentation designed to be used standalone; all you need is their big download to install the IDE, and their web site to tell you how to use the board. You don't have to dump a set of Atmel manuals and a GCC manual on the user.
Yeah, clueless. It's been a long time since either a phone number or an IP address was like a street address.
The distinction is how "locative" an ID is. Seat numbers in a stadium are locative. But few other IDs are completely locative any more.
At one time, phone numbers really were locative; the first three digits specified the central office, or for larger offices, the switch within the CO, and the last four digits were the line number within the switch. That dates from the era when phone numbers were read like "PLaza 5-1000". But that was a long time ago. Now there's a lookup in the middle; the number goes into a database and a location comes out. That allows for number portability, and also means all the numbers can be used.
The same thing happened to IP addresses. At one time, you could route by tearing apart IP addresses. In the beginning, network 10 was the ARPANET, and the last byte was the IMP number. Now, it's lookup-based, and routers have huge tables.
This is a continuing struggle with numbering plans, from zip codes to Ethernet addresses to UPC codes. They tend to start out locative, but eventually become a flat, arbitrary space as the demands on the number space increase and things change over time.
This is good. When legitimate businesses realize there are consequences for doing anything that looks like spamming, they'll be more careful.
We're actually winning on spam. There aren't that many different spammers left. The volume of spam is way up because of botnets, but the number of different incoming spams is way down. Notice how few different spams you get.
The remaining spammers are all committing multiple felonies. Notice that when one of them gets caught, which happens a few times a year now, they go to jail. For years.
Progress against spam is slow, but real.
Isn't shared memory intrinsically slow, because access has to be interleaved between all the processors sharing it?
In practice, no. There are clever techniques to get around that, most of which involve caches.
Shared memory between modern multiprocessors is to some extent an illusion, one created by very clever cache hardware design. Each processor has its own cache with local memory, and each cache controller has some information about what the other processors have in their caches. When one processor changes a memory location and another process needs to read it, the cache controllers communicate and update the caches. That slows things down, but it doesn't happen too often. This is called "cache coherency", which you can look up to learn more about the subject. The end result is that it looks to programs like all the processors share the same memory, but most of the time they're really accessing local copies of main memory.
He might not have radar, but he certainly should have a radar reflector.
Radar visibility helps, but the big container ships and oil tankers have such limited maneuverability that they can't be expected to try to avoid agile little sailboats. "A vessel of less than 20 meters in length or a sailing vessel shall not impede the safe passage of a power-driven vessel following a traffic lane." (Rule 10(j)). This is the marine equivalent of "no pedestrians on the freeway".
Here's why: the FBI probably uses this technique, in some cases, to track down child porn.
No, most child pornography is distributed by law enforcement. Mostly by the FBI's Baltimore office.
I've met some of the architects of the Cell processor, and they have a "build it and they will come" attitude. They've designed the computer; it's up to others to make it useful. This is probably not going to fly.
The Cell is a non-shared memory multiprocessor with quite limited memory per processor. There's only 256K per processor, which takes us back to before the 640K IBM PC. There are DMA channels to a bigger memory, but no cacheing. Architecturally, it's very retro; it's very similar to the NCube of the mid-1980s. It's not even superscalar. Cell processors are dumb RISC engines, like the old low-end MIPS machines. They clock fast, but not much gets done per clock.
Yes, you get lots of CPUs, but that may not help. On a server, what are you going to run in a Cell? Not your Java or Perl or Python server app; there's not enough memory. No way will an instance of Apache fit. You could put a copy of the TCP/IP stack in a Cell, but that's not where the CPU time goes in a web server. One IBM document suggests putting "XML acceleration" (i.e. XML parsing) in the server, but that's an answer looking for a problem. It might be useful for streaming video or audio; that's a pipelined process. If you need to compress or decompress or transcode or decrypt, the Cell might be useful. But for most web services, those jobs are done once, not during playout. Even MPEG4 compression might be too much for a Cell; you need at least two frames of storage, and it doesn't have enough memory for that.
Now if they had, say, 16MB per CPU, it might be different.
The track record of non-shared memory supercomputers is terrible. There's a long history of dead ends, from the ILLIAC IV to the BBN Butterfly to the NCube to the Connection Machine. They're easy to design and build, but just not that useful for general purpose computing. Some volumetric simulation problems, like weather prediction, structural analysis, and fluid dynamics can be crammed into those machines, so there are jobs for them, but the applications are limited.
Shared-memory microprocessors look much more promising as general purpose computers. Having eight or sixteen CPUs in a shared-memory multicore configuration is quite useful. That's how SGI servers worked, and they had a good track record. Scaling up today's multicore shared-memory CPUs is repeating that idea, but smaller and cheaper.
At some point, you have to go to non-shared memory, but that doesn't have to happen until you hit maybe 16 CPUs sharing a few gigabytes of memory, which is about when the cache interconnects start to choke and speed of light lag to the far side of the RAM starts to hurt. That might even be pushed harder; there's been talk of 80 CPUs in a shared memory configuration. That's optimistic. But we know 16 will work; SGI had that years ago.
Then you go to a cluster on a chip, which is also well understood.
That's the near future. Not the Cell.
These things are really x86 PCs, with upwards of 64MB of memory. They're quite capable computers. The Neoware C50 is a desktop Linux system with no hard drive, for $259. The Wyse S50 is another comparable Linux box. Wyse even has a dual screen model. The HP model runs Debian. HP is having a sale - buy 3, get one free.
Neoware even has a thin client notebook computer. It's only useful when it has a WLAN connection. This is promoted as a security feature; if it's stolen, there's not much data in it.
This may be the way Linux comes to the enterprise desktop. To many companies, this is a cheaper and easier conversion than moving to Vista.
It's good to see ICANN doing some cleanup. For the past few years, they've been something of a trade group for domain registrars.
A few more TLDs could go. .museum and .aero could be phased out due to lack of interest. The entire list for .museum is a few pages, the domains aren't the top-tier museums, and almost all of them are redirects anyway. .aero has an entry for every airport code (try LAX.AERO), but those were put there by the domain registrar to give the illusion of activity and they're not the primary domain name for those sites. ("LAX.AERO" is really "WWW2.LAWA.ORG").
He went out in a 40-foot C&C 121 yacht. That's a very nice boat, with a epoxy resin laminate hull, carbon fiber reinforcement and masts, Kevlar sails, and a 38HP engine. There hasn't been any weather lately bad enough to give a boat like that any serious trouble. If it ran aground it would probably survive the experience.
But between San Francisco and the Farralon Islands is a major shipping lane. One with fog. Container ships and oil tankers come through there. Sizable fishing boats have been run down and sunk without anyone on a large ship even noticing. There's a USGS Vessel Traffic Service station and established traffic lanes for large ships, but small boats aren't required to check in with traffic control.
oh well, maybe CSS 3 will fix somethings?
Maybe. Layout tables are going into CSS. The syntax looks like this:
body { display: "a@@" "@@@" "@@b" "@@@" "c@@" "@@@" "@@d" }
Sort of like COBOL meets HTML. This does not look promising. Nor does it look well-defined.
Those are the options. And they all suck.
This is an opportunity for somebody. Probably somebody in China.
I find it amusing that the CSS zealots out there will rant about making sites that are pure CSS
I know. There's such hype about CSS layout. The truth is, the CSS machinery for layout is dumber than the table machinery. With CSS, the basic mechanisms for layout are "float", "clear", explicit sizing, and absolute positioning. Because the first two are weaker than tables, absolute positioning is overused. Then you get things like text appearing on top of other text, something that never happens with tables.
Dreamweaver had table-based layout working great years ago. Most of the complaints about tables come from people writing the tags by hand, but with a proper layout tool, it works very well.
CSS layout should have been constraint-based. You should be able to express concepts like "align bottom edge of this box with bottom edge of previous box" and "align left edge of this box with right edge of previous box". Then you could handle situations like the dreaded "3-column problem" cleanly. The right way to do it would be to allow arbitrary linear constraints, such as "box must have same width as previous box" and "box must have 0.5 width of enclosing box", dump them all into a linear constraint engine, crank the simplex algorithm, and get out a layout that satisfies all the constraints.
If you allow constraints with respect to non-adjacent boxes, you can do grid-based layout, too, and do all the things tables do. You'd want to do layout like this in a GUI tool, not by writing constraints by hand, of course. Drag box corners to other corners and edges and have them snap into lock.
This isn't a new idea; it's called "parametric CAD" in engineering design. But it's something the designers of CSS probably didn't know about.
In Hong Kong, there are virtual girlfriends for cell phones.
This thing is really insidious. The girls are high-maintenance. You have to text them frequently and buy them stuff.
It can act as usb mass storage and thus be carrying infected files.
Did they export the whole drive, or what? One would expect they'd export an empty file system, onto which one could load music or other content, and provide some means to reset it to empty if it became corrupted.
What is this thing running? Some consumer-grade version of Windows, apparently. Clearly, not only is the virus "on the hard drive", but it actually gets executed.
You'd think they'd at least run XP Embedded, or Windows CE, with only the necessary components present. I don't expect a read-only image of QNX in ROM in a non-critical consumer infotainment device, but this is lame.
There's considerable unhappiness in San Francisco about Lucasfilm's operation. It's in the Presidio, which used to be a military base and is now a national park. It's the only national park which has to make a profit, due to a Bush Administration deal. Letterman Army Hospital was torn down to make room for the Lucasfilm facility. The San Francisco Bay Guardian complains about this constantly, as they try to keep the Presidio from turning into an industrial park. The Lucasfilm move to the Presidio was something of a dot-com boom excess, when people thought SF was the place to be.
Pixar, in Emeryville, Tippett, in Berkeley, and Dreamworks, in Redwood City, are the innovative animation companies in the Bay Area. And of course, there's EA, SCEA, and some other game companies. Lucasfilm doesn't seem to get much attention.
There are data centers in San Francisco proper with far more storage, too. The Internet Archive has several petabytes of storage. There's a large colocation facility at the 6th St. offramp from I-280.
Every year, I see these "Linux is ready for the desktop" articles. But it never happens. Back in 2004, WalMart offered a $499 Linux laptop. They don't do that any more. Lenovo, HP, and Dell have fooled around with Linux laptops, but try to order one on line. Search for "linux laptop" on Dell, and you get back "Dell recommends Windows Vista(TM) Business." There are some off-brand Linux laptops available, but they're overpriced.
Linux on the desktop looked closer three years ago than it does now.
Much of this new stuff sounds like features of QNX. QNX has a "sporadic scheduler", for when you need things like 10ms of CPU every 100ms. QNX has had I/O cancellation for years. In QNX, you can set a timeout on any system call that blocks. If you set a 35ms timeout on a write, after 36 milliseconds, you'll have control back. Very useful in real-time systems where you're doing something less important, like logging, that should never take very long but, in some trouble condition, might. QNX has had prioritized I/O for years, too.
It all works, too. I've done compiles on QNX while running a real time program on the same machine, without the real time program missing a deadline.
Of course, in Vista, it's all more complicated.
Everyone assumed that Apple's $499/$599 prices for the iPhone was subsidized by Cingular.
Who thought that? At that price, it had better be profitable.
Cell phone internals are becoming very cheap. Check out the Texas Instruments "LoCosto" two-chip cell phone. Manufacturing costs are approaching $20. This isn't being reflected in the prices seen at US carrier's retail outlets, though. The handset price is inflated there, then "discounted" to compel users to sign up for "plans".
That's coming from Froogle. Companies that sign up with Froogle or Google Checkout have some additional info about them. Try, say, Super Warehouse, which Google describes as "Online retailer of color laser printers, laptops, hard drives, LCD monitors, and digital cameras". That text isn't from the "www.superwarehouse.com" web page, which starts out "Printers - Scanners - Toner - Monitors - Projectors & More at Super Warehouse".
It worked for the DARPA Grand Challenge, but not in the way most people think.
The prize was the carrot. But there was a stick, too. The Grand Challenge was a real threat to robotics funding at CMU and Stanford, which had been getting DARPA money for decades but were progressing very slowly. Originally, neither university's robotics group intended to enter. But there were apparently hints that if the non-university entries did significantly better than the people DARPA had been funding, the funding for the big robotics labs would be in jeopardy.
At Stanford, the management of the AI Lab had to be replaced to get real results. That was done, and things picked up quite a bit. There was a significant breakthrough in robotic vision for the Stanford vehicle. That was the real payoff in all this.
The CMU vehicles were what someone above called a "point solution" - the searchlight-sized gyro-stabilized gimbal with a line scanning LIDAR was a technological dead end.