Pad2Pad is finally up and running. They not only make boards. They'll put the components on them. For reasonable prices. If it's in Digi-Key, they'll put it on.
This means you can use surface mount devices, ball grid arrays, and other advanced packages. Finally!
They do all the things you can't do at home, like
X-ray inspection of BGA soldering.
It's easy to get PC boards made today via the Internet. Advanced Circuits is fine, and there are other services as well. You design in a downloaded application and send them the file. They make the board, including drilling and plated through holes.
Making your own boards isn't worth it. The iron-on transfer approach has a very high defect rate. Photoetching works better, but you need more equipment and you'll end up spending more than it would cost to have a board made. And do-it-yourself boards won't have plated through holes.
The PackBot has had great success with a similar wheel. In that application, it's used as a sprocket for a track, but it's the same flexible-spoke concept. The PackBot is so rugged that normal usage is to throw it through a window into a building, then drive it around to see what's inside.
Why should kids learn to program? It's a specialist skill with declining value.
Get them a pony. This will teach them to deal with an animal that's cooperative enough that you can do something with it, but independent enough that it's not easy. This prepares them for management.
FastCGI is neat, but a different approach. It's basically another level of client/server processing, in which the CGI programs are servers and the web server is the client. Because the CGI programs don't exit after each transaction, they can be corrupted. That's different than a transaction processor.
Why manhole covers are round
on
Defining Google
·
· Score: 2, Informative
That's the classic Microsoft interview question. And all the published answers are wrong.
Today, covers come in many shapes - round, square, rectangular, oblong with rounded corners. But older covers tend to be round.
This reflects manufacturing technology a century ago. In 1900, you could cast metal, so you could get any rough shape you wanted. Machining was limited to flat planing, grinding, drilling, and turning on a lathe. Milling machines for heavy work didn't exist yet. Welding wasn't working yet, either. Turning on a lathe was the only really high-precision operation available.
So you could make flat things, or round things, or imprecise things, or riveted things. Look at a steam locomotive from about 1900, and that's what you'll see. Almost everything is either flat or a full circle. You won't see arbitrary curves on parts that have to fit. You won't see rectangular inside corners.
Actually, it's not making the cover that was hard in 1900. You could make a square cover. But making the ring into which it fits was tough. The inside of the ring has to be flat, or the cover will rattle. An unmachined casting will be too rough. Some finish machining will be required.
Casting a round ring is straightforward.
You make a wooden master, press it into a box of moulding sand, and pour in molten metal. Straightforward foundry work. Finish-machining the ring on a lathe is easy. The only surfaces that matter are the ones where the lid touches the ring. One clamping of the work to a flat spindle plate, two cuts, one for each surface, using stock lathe cutters that can be resharpened on an ordinary grinding wheel.
This could all be done cheaply in 1900.
Today it's no problem to make a square frame. You'd make a square frame by cutting angle stock into sections and welding the corners. Clean up the welds with a power grinder. Or make a rough casting, then do a quick pass with a CNC grinder to true it up. So today, you see square frames with square covers.
But try to make a heavy square frame with 1900 technology. You can rough cast the frame, but smoothing out the inside edges is a tough job.
You can't use a lathe; the workpiece isn't round. You don't have a milling machine.
You can't get a planer into the corners. It's hand work, with files and grinding stones. That's slow and expensive, unaffordable for a cheap generic product.
And that's the real answer to why manhole covers are round.
Java, remember, was intended as a client-side concept. As a server-side concept, it provides a useable API to the application programmer, but from the OS side, it's a terrible idea.
If you want reliability, what you need is a transaction processing system, where transaction programs are quickly started, do their job for one user, and exit. There have been high-performance systems for transaction processing for decades. That's how mainframes work. CICS, TIP, and of course Tandem worked that way. CGI programs also work that way, but because UNIX/Linux is a lousy transaction processing system, they've fallen
out of favor.
If you wanted a real transaction system for Linux, what you'd need is this:
Some way to make forkable transaction programs with small dynamic memory footprints and small startup costs.
That means a language implementation with shared code, small dynamic data, and short startup time.
Perl, Python, or Java would work with some modifications to the run time environment, but a hard-compiled language would be better.
Copy on write semantics for transaction programs.
Really fast fork. Not fork/exec, fork.
The basic idea is that the web server starts up some CGI program, lets it do all its initialization, and lets it run to the point that it is ready to accept data for a specific transaction. Then it makes a call to wait for transaction data. When a transaction comes in, the server forks off a new copy of the initialized transaction program. That copy deals with one user transaction, and then exits.
This gets around the big problem with CGI programs - the initialization cost dominates the transaction. CGI is a good idea from the reliability and security standpoint. If the startup overhead problem is fixed, it's a good solution for high volume applications. You don't want to be doing class loading or source compilation for every transaction. On the other hand, you don't want to share the run time environment between transactions.
As indoor oceans go, this CargoLifter one is unimpressive. It's just a big pool. Phoenix SeaGaia OceanDome is much better. They have surf, powered by really big pumps. And the roof opens.
It is worth noting, that for a desktop OS to survive in the long term, it needs games!
I was just telling this to the QNX people, who finally implemented OpenGL, but made the SDK for it an extra-cost item. (Even though, underneath, it's Mesa.)
This scenario is playing out in TV land right now. Four times as many users have video on demand from a cable company as own PVRs.
The home "network computer" was a good idea, but Microsoft went to great lengths to kill it, with proprietary browser extensions and such. There should have been an i-Opener in every hotel room in the developed world. A stateless, diskless client is just what you need in that application.
This sounds like some MPAA exec's fantasy of how the Internet works. Small armies of "curries" manually FTPing files from one server to another? Get real.
I think somebody made all this up to impress Wired, and they fell for it.
Unfortunately, Bramah's "unpickable" lock was simply a round pin-tumbler lock. Like the ones Kryptonite used. Better design, though. The clever feature of the Bramah lock is that there's only one return spring for all the pins. So picking is really slow. Every time you get the setting wrong, you have to release all the pins and start over.
They can't even get the food supply to the ISS under control, let alone build a new launch vehicle. But they put resources into PR stunts like this. There's way too much "outreach" (read PR) at NASA. They need to close about half their centers, cut 80% of the PR operation, and focus on their core activity - putting stuff in space.
If Disney or Speilberg or Burger King wanted to pay for this, that's fine. But it's a misuse of Government funds for NASA to be doing it. I want to find out what line item in NASA's budget request covers this.
Rolex, the company, now has a serious problem. They're not in the watch business. They're in the status symbol business. (Which, amusingly, their CEO admits.)
The only thing they have going for them is their "cool factor". Without that, they're a company that sells overpriced watches.
All those Rolex spams have made their brand a joke. They're very close to losing their "cool brand" identity.
Only a badly-run department would have a problem with this. Dispatch is supposed to know where the cars are. It's easier on everybody if they have an automatically updated map. Otherwise there's constant radio chatter as Dispatch tries to manually keep track of where everybody is.
Well, actually there was desktop publishing well before the Mac. The original desktop publishing was the Alto/Bravo/Dover combo in the late 1970s. Interleaf on Sun 1 workstations appeared around 1982 and lasted until about 1998.
And, of course, there was the Xerox Star.
Interleaf was way, way ahead of its time. Visualize Word 97 plus FrameMaker, but fifteen years earlier. But the company had a terrible business model. They were selling a combined hardware/software package consisting of four Sun workstations, a file server, and a laser printer.
Back then, people didn't think that software-only businesses could become big.
As I recall (and this may be apocryphal - somebody correct me) some workstations overcame this in a second way - they ran two 68K's in parallel, one a clock cyle or two ahead of the other and, when the early one faulted, they asserted an interrupt (which saved state properly) on the second processor.
Yes. Early model Apollo workstations did that.
There was a period of screaming frustration among 68K-based workstation vendors in the early 1980s because Motorola was years late with the 68010 and
a matching MMU. Motorola didn't really get the whole chipset working until the 68020 CPU, the matching 68451 MMU and 68881 FPU. That's the Mac II configuration, although most Mac II machines went out the door with a cheap "jumper" chip in place of the MMU. The MacOS couldn't use an MMU anyway.
Apollo basically did most of what Sun did, but first.
It was all proprietary technology - Domain OS, Apollo Token Ring, Domain Network File System.
They had graphics before Sun. They were bigger than Sun for a while, but Sun's ability to mooch off Government-funded BSD work was too big an edge for Apollo to overcome.
The Lisa was Apple's useful machine. A multitasking OS, virtual memory, a hard drive, and networking made it a usable machine. But at $10K, it was far too expensive. And Apple's abysmal hard drive, the LisaFile, hurt it badly. But the real problem was that Motorola was years late with the MMU for the 68000. The Lisa had an MMU built out of register-level parts, which ran the parts count and the system cost way up. And there was a bug in the 68000 which made page fault processing unsafe. Instruction backout/resumption didn't work. So the compiler had to generate only idempotent memory-referencing instructions, ones that if done twice had the same effect as doing it once.
The original Mac was a dismal flop. Ever use a 128K Mac? No hard drive. One floppy (dumb). No MMU. No multitasking. You spend all your time changing floppies and looking at the watch icon. It sold badly and for a while, Apple looked doomed.
Apples's big success wasn't the Mac at all. It was the LaserWriter. The LaserWriter saved the Mac line. Once the LaserWriter was out, there was a reason to use the Mac. Before the LaserWriter, the Mac was an expensive toy.
Tax information isn't that hard to get. Remember that all the European OECD countries provide medical care, so you have to add insurance premiums into US tax rates to compare.
US $14.20 (That's the OECD's number; the AFL-CIO says $12.03.)
UK $13.33
Australia $13.14
Japan $12.83
For the US, that number has been flat for well over a decade. For the European countries, it's climbing. That data is four years old.
That study is worth a look. The US has the lowest level of legal employment protection, and the lowest level of collective bargaining for wages, of any of the OECD countries.
More recent data is available, but not for free. The OECD Database requires a paid subscription.
It's a public policy decision that this is allowed to happen. It's not happening in Europe, due to "protectionist" policies. We are paying the price of a government controlled by big business.
An hour's worth of work in France or Germany now buys more than an hour's worth of work in the US. The US is ahead on per capita income only because of longer working hours. And it's not ahead by much. US per capita income was 2x of that in France in 1980. Now it's about 1.2x, and when the dollar drops a little more...
The head of Germany's Fraunhofer Institute was over at Stanford a few weeks ago, chewing out Americans for letting the Government lose all the manufacturing jobs. Germany didn't let that happen.
Congress can turn this around any time it wants to.
Never forget that on election day.
RegisterFly, a service which "cloaks" domain registrations by using RegisterFly's contact information in place of the actual registrant, may be committing felonies by so doing.
From the CAN-SPAM act:
Sec. 1037. Fraud and related activity in connection with electronic mail
`(a) IN GENERAL- Whoever, in or affecting interstate or foreign commerce, knowingly--....
(4) registers, using information that materially falsifies the identity of the actual registrant, for five or more electronic mail accounts or online user accounts or two or more domain names, and intentionally initiates the transmission of multiple commercial electronic mail messages from any combination of such accounts or domain names,....
"or conspires to do so, shall be punished as provided in subsection (b)."
(2) a fine under this title, imprisonment for not more than 3 years, or both, if--...
(B) the offense is an offense under subsection (a)(4) and involved 20 or more falsified electronic mail or online user account registrations, or 10 or more falsified domain name registrations;
The CAN-SPAM act is soft on spamming, but tough on
spam-related fraud. That can be useful.
Note the "or conspires to do so" clause. Knowingly assisting in a criminal offense satisfies the legal definition of conspiracy. "Cloaking services" are in deep trouble if they knowingly provide that service for a spammer.
Unlike ISP's, there's no "safe harbor" for them.
As for the "knowingly" part, whenever you find a spam associated with a "cloaked" domain, send a note to the cloaking service, and post that you've done so to some public spam forum that's indexed by search engines. That will put them on record as knowingly cooperating in a criminal conspiracy. The next person who gets a spam from the same party will have that information as legal ammunition.
Slashdot Mirror
This means you can use surface mount devices, ball grid arrays, and other advanced packages. Finally! They do all the things you can't do at home, like X-ray inspection of BGA soldering.
Making your own boards isn't worth it. The iron-on transfer approach has a very high defect rate. Photoetching works better, but you need more equipment and you'll end up spending more than it would cost to have a board made. And do-it-yourself boards won't have plated through holes.
The PackBot has had great success with a similar wheel. In that application, it's used as a sprocket for a track, but it's the same flexible-spoke concept. The PackBot is so rugged that normal usage is to throw it through a window into a building, then drive it around to see what's inside.
Get them a pony. This will teach them to deal with an animal that's cooperative enough that you can do something with it, but independent enough that it's not easy. This prepares them for management.
FastCGI is neat, but a different approach. It's basically another level of client/server processing, in which the CGI programs are servers and the web server is the client. Because the CGI programs don't exit after each transaction, they can be corrupted. That's different than a transaction processor.
Today, covers come in many shapes - round, square, rectangular, oblong with rounded corners. But older covers tend to be round.
This reflects manufacturing technology a century ago. In 1900, you could cast metal, so you could get any rough shape you wanted. Machining was limited to flat planing, grinding, drilling, and turning on a lathe. Milling machines for heavy work didn't exist yet. Welding wasn't working yet, either. Turning on a lathe was the only really high-precision operation available.
So you could make flat things, or round things, or imprecise things, or riveted things. Look at a steam locomotive from about 1900, and that's what you'll see. Almost everything is either flat or a full circle. You won't see arbitrary curves on parts that have to fit. You won't see rectangular inside corners.
Actually, it's not making the cover that was hard in 1900. You could make a square cover. But making the ring into which it fits was tough. The inside of the ring has to be flat, or the cover will rattle. An unmachined casting will be too rough. Some finish machining will be required.
Casting a round ring is straightforward. You make a wooden master, press it into a box of moulding sand, and pour in molten metal. Straightforward foundry work. Finish-machining the ring on a lathe is easy. The only surfaces that matter are the ones where the lid touches the ring. One clamping of the work to a flat spindle plate, two cuts, one for each surface, using stock lathe cutters that can be resharpened on an ordinary grinding wheel. This could all be done cheaply in 1900.
Today it's no problem to make a square frame. You'd make a square frame by cutting angle stock into sections and welding the corners. Clean up the welds with a power grinder. Or make a rough casting, then do a quick pass with a CNC grinder to true it up. So today, you see square frames with square covers.
But try to make a heavy square frame with 1900 technology. You can rough cast the frame, but smoothing out the inside edges is a tough job. You can't use a lathe; the workpiece isn't round. You don't have a milling machine. You can't get a planer into the corners. It's hand work, with files and grinding stones. That's slow and expensive, unaffordable for a cheap generic product.
And that's the real answer to why manhole covers are round.
If you want reliability, what you need is a transaction processing system, where transaction programs are quickly started, do their job for one user, and exit. There have been high-performance systems for transaction processing for decades. That's how mainframes work. CICS, TIP, and of course Tandem worked that way. CGI programs also work that way, but because UNIX/Linux is a lousy transaction processing system, they've fallen out of favor.
If you wanted a real transaction system for Linux, what you'd need is this:
-
Some way to make forkable transaction programs with small dynamic memory footprints and small startup costs.
That means a language implementation with shared code, small dynamic data, and short startup time.
Perl, Python, or Java would work with some modifications to the run time environment, but a hard-compiled language would be better.
-
Copy on write semantics for transaction programs.
-
Really fast fork. Not fork/exec, fork.
The basic idea is that the web server starts up some CGI program, lets it do all its initialization, and lets it run to the point that it is ready to accept data for a specific transaction. Then it makes a call to wait for transaction data. When a transaction comes in, the server forks off a new copy of the initialized transaction program. That copy deals with one user transaction, and then exits.This gets around the big problem with CGI programs - the initialization cost dominates the transaction. CGI is a good idea from the reliability and security standpoint. If the startup overhead problem is fixed, it's a good solution for high volume applications. You don't want to be doing class loading or source compilation for every transaction. On the other hand, you don't want to share the run time environment between transactions.
I have a track record at predicting failures. I did Downside's Deathwatch.
As indoor oceans go, this CargoLifter one is unimpressive. It's just a big pool. Phoenix SeaGaia OceanDome is much better. They have surf, powered by really big pumps. And the roof opens.
I was just telling this to the QNX people, who finally implemented OpenGL, but made the SDK for it an extra-cost item. (Even though, underneath, it's Mesa.)
The home "network computer" was a good idea, but Microsoft went to great lengths to kill it, with proprietary browser extensions and such. There should have been an i-Opener in every hotel room in the developed world. A stateless, diskless client is just what you need in that application.
I think somebody made all this up to impress Wired, and they fell for it.
Picking a Bramah lock is quite possible, but requires some specialized tools.
When you allow a story about some bill on Slashdot, cite the bill, or provide a link. Stories like this are useless.
"All these people keep bemoaning the fact that they can't communicate. If they can't communicate, the least they can do is SHUT UP." -- Tom Lerher
They can't even get the food supply to the ISS under control, let alone build a new launch vehicle. But they put resources into PR stunts like this. There's way too much "outreach" (read PR) at NASA. They need to close about half their centers, cut 80% of the PR operation, and focus on their core activity - putting stuff in space.
If Disney or Speilberg or Burger King wanted to pay for this, that's fine. But it's a misuse of Government funds for NASA to be doing it. I want to find out what line item in NASA's budget request covers this.
All those Rolex spams have made their brand a joke. They're very close to losing their "cool brand" identity.
Only a badly-run department would have a problem with this. Dispatch is supposed to know where the cars are. It's easier on everybody if they have an automatically updated map. Otherwise there's constant radio chatter as Dispatch tries to manually keep track of where everybody is.
Interleaf was way, way ahead of its time. Visualize Word 97 plus FrameMaker, but fifteen years earlier. But the company had a terrible business model. They were selling a combined hardware/software package consisting of four Sun workstations, a file server, and a laser printer. Back then, people didn't think that software-only businesses could become big.
Yes. Early model Apollo workstations did that.
There was a period of screaming frustration among 68K-based workstation vendors in the early 1980s because Motorola was years late with the 68010 and a matching MMU. Motorola didn't really get the whole chipset working until the 68020 CPU, the matching 68451 MMU and 68881 FPU. That's the Mac II configuration, although most Mac II machines went out the door with a cheap "jumper" chip in place of the MMU. The MacOS couldn't use an MMU anyway.
Apollo basically did most of what Sun did, but first. It was all proprietary technology - Domain OS, Apollo Token Ring, Domain Network File System. They had graphics before Sun. They were bigger than Sun for a while, but Sun's ability to mooch off Government-funded BSD work was too big an edge for Apollo to overcome.
Reality is more like this:
The Lisa was Apple's useful machine. A multitasking OS, virtual memory, a hard drive, and networking made it a usable machine. But at $10K, it was far too expensive. And Apple's abysmal hard drive, the LisaFile, hurt it badly. But the real problem was that Motorola was years late with the MMU for the 68000. The Lisa had an MMU built out of register-level parts, which ran the parts count and the system cost way up. And there was a bug in the 68000 which made page fault processing unsafe. Instruction backout/resumption didn't work. So the compiler had to generate only idempotent memory-referencing instructions, ones that if done twice had the same effect as doing it once.
The original Mac was a dismal flop. Ever use a 128K Mac? No hard drive. One floppy (dumb). No MMU. No multitasking. You spend all your time changing floppies and looking at the watch icon. It sold badly and for a while, Apple looked doomed.
Apples's big success wasn't the Mac at all. It was the LaserWriter. The LaserWriter saved the Mac line. Once the LaserWriter was out, there was a reason to use the Mac. Before the LaserWriter, the Mac was an expensive toy.
Tax information isn't that hard to get. Remember that all the European OECD countries provide medical care, so you have to add insurance premiums into US tax rates to compare.
For the US, that number has been flat for well over a decade. For the European countries, it's climbing. That data is four years old.
That study is worth a look. The US has the lowest level of legal employment protection, and the lowest level of collective bargaining for wages, of any of the OECD countries.
More recent data is available, but not for free. The OECD Database requires a paid subscription.
An hour's worth of work in France or Germany now buys more than an hour's worth of work in the US. The US is ahead on per capita income only because of longer working hours. And it's not ahead by much. US per capita income was 2x of that in France in 1980. Now it's about 1.2x, and when the dollar drops a little more...
The head of Germany's Fraunhofer Institute was over at Stanford a few weeks ago, chewing out Americans for letting the Government lose all the manufacturing jobs. Germany didn't let that happen.
Congress can turn this around any time it wants to. Never forget that on election day.
From the CAN-SPAM act:
The CAN-SPAM act is soft on spamming, but tough on spam-related fraud. That can be useful.Sec. 1037. Fraud and related activity in connection with electronic mail
`(a) IN GENERAL- Whoever, in or affecting interstate or foreign commerce, knowingly-- ....
(4) registers, using information that materially falsifies the identity of the actual registrant, for five or more electronic mail accounts or online user accounts or two or more domain names, and intentionally initiates the transmission of multiple commercial electronic mail messages from any combination of such accounts or domain names, ....
"or conspires to do so, shall be punished as provided in subsection (b)."
(2) a fine under this title, imprisonment for not more than 3 years, or both, if-- ...
(B) the offense is an offense under subsection (a)(4) and involved 20 or more falsified electronic mail or online user account registrations, or 10 or more falsified domain name registrations;
Note the "or conspires to do so" clause. Knowingly assisting in a criminal offense satisfies the legal definition of conspiracy. "Cloaking services" are in deep trouble if they knowingly provide that service for a spammer. Unlike ISP's, there's no "safe harbor" for them.
As for the "knowingly" part, whenever you find a spam associated with a "cloaked" domain, send a note to the cloaking service, and post that you've done so to some public spam forum that's indexed by search engines. That will put them on record as knowingly cooperating in a criminal conspiracy. The next person who gets a spam from the same party will have that information as legal ammunition.
When you've got that info, report it as Internet fraud..