Some little problems...
on
Segway vs. Roomba
·
· Score: 5, Interesting
"Yes, some people had problems, but for the most part this fairly unnatural motion on an inverted pendulum went smoothly."
Yeah, problems. Like the demonstrator I saw take a violent Segway header at the Minnesota State Fair last year. While he's giving his pitch, one of the wheels on his Segway hit a tiny patch of water on the slick floor. The wheel spun and smoked like a funny car doing a burn-out, and the guy went down so hard that his head bounced about six inches off of the floor. Next day he was back, with a cast on one arm, and a hockey helmet on his head.
Yeah, the inverted pendulum is a cute trick, but then again, so is riding a unicycle. I wouldn't advise my grandmother to ride either one. On the other hand, I would give her a Roomba. I own one, and apart from driving my pet beagle completely ballistic, it's great.
We could eliminate the entire virus/worm propagation problem by making a few hundred different builds of each release of programs like IE, IIS, and Windows, with each build having an added, random-sized dummy variable in each system call. The existence of a random-sized placebo variable on the stack would make each build's buffer overflows occur at different addresses. Bingo, no more homogenous population of buffer overflows!
Shucks, maybe I should have patented that idea before letting it out...
...how many of the spams got opened in a month, to sell 10,000 packages. Then we could compute the percentage of recipients dumb enough to order the "product." Then we could have a national GDS (Gross Domestic Stupidity) index, and publish it like the GDP, GNP, stock indices, and other important economic indicators. Heck, maybe we could even break it down state-by-state and compare it to things like election results and stuff.
So after my sorta' incendiary post about the PhysOrg's original "world's first 20-layer circuit board" headline being absurd, I went back and surfed the article to see the corrected headline. Got a good laugh out of the content of the Goooooogle ad that popped up just below the headline:
PCB Manufacturing
Order Quality PCBs 2 to 24 Layers. 12-Hour Turns & Same-Day Shipping.
Apparently Epson's breakthrough in layer counts caught on in record time!
...contains an absurdly incorrect statement in its first sentence, to wit:
"Seiko Epson Corporation today announced that it has succeeded in leveraging its proprietary inkjet technology to develop what the company believes is the world's first 20-layer circuit board."
No. Not even close. I have personally worked on circuit boards of as many as 48 layers, as long ago as 1985. The math coprocessor for the
Sperry-1100/90 (code named "Eagle") had a motherboard that was roughly 20" x 36" in size, had forty-eight layers, was about 1/2" thick, had
solid silver bus bars laminated in each side, weighed about forty pounds, and was so hard that if you knocked on it with your knuckle, it would
ring like a bell.
There is no possibility that the people at Epson believe they've built the world's first twenty layer board. Twenty layer boards are a little
uncommon, but far from record-breaking.
Sincerely,
BrakesForElves
Founder and past President
FASTechnologies, Inc. www.fastec.com
1) It's much easier to write a C compiler than a Pascal compiler, therefore the (early) availability of the C language on new platforms became a near certainty.
2) It didn't take project and product managers long to realize that in the era of Moore's Law, platform flexibility had great value. A project stuck on an obsolete platform due to the unavailability of its language on a revolutionary new platform was doomed, perhaps prematurely.
So its portability and ubiquity were C's most significant advantages over Pascal, back when there was a realistic contest.
3) For a time, executables written in C were likely to be considerably faster than those written in Pascal. This was a byproduct of the re-use of the C compiler code itself, versus fresh (read: immature) attempts at Pascal compilers. The C compiler cores got better with each processor port, but the freshly-written Pascal compilers often were not very good.
Today on the x86 platform with Borland's highly-refined 8th-generation compiler core, executables built from well-written Pascal are as fast as those built from coherently-written C, in my experience. It may be possible to write incredibly concise C that'll be a hair faster than the same thing written in Pascal, but arguing that difference is a fool's errand in the days of 4GHz rocket-ship machines executing septillions of NOPs waiting for something to do.
Personally, I choose Delphi these days over C, because I write and support huge projects. It is incredibly easy to pick up Pascal source and quickly figure out exactly what it does. That's the first (and most crucial) step in any software maintenance, and I find that Pascal's support cost savings more than over-balance any possible advantages I've ever realized from using C. When I'm writing something that needs to be extremely fast, I drop into inline assembly, but everything else I code in Delphi these days.
Thanks for pointing that out. No, ignoramus that I am, I was trying to do two things at once while reading the article. I'll read more carefully before I post on/. from now on. Cheers!
Either Mr. Chin is living in a cave, or he wrote this piece some time ago. With Firefox numbers skyrocketing and even CERT suggesting that running IE is inviting virus infections, his statement, "Internet Explorer will continue its chokehold on the World Wide Web" seems quite out of touch with present reality.
...And sadder that even with the instant, world-wide communications we have, people still suggest that comparing the representative government of the US to the repressive, muderous government of Stalin has any validity, at all, ever. In case nobody has noticed, Russia was ruled by Stalin alone, whereas the US is ruled by, er... well, _lawyers_ I guess. (But the line about Stalin still goes).
.
We used to call it "Declaring Octvember", the mythical 45-day month at the end of the fiscal year.
Y'know, the reason for stretching a quarter is that the lazy financial analysts fixate on simple things like a company missing its (or their) estimates, not whether the company's fundamentals and long-term prospects are sound. When lazy analysts drive corporate decision making, prepare for a parade of stupid decisions like declaring Octvember.
You might want to get your friends to write a new completely free operating system from scratch.
That's a better idea than you might think: If we didn't have basically two operating systems running several hundred million computers, we wouldn't see worms like Sasser zap around the world unchecked in twenty minutes. I've been working on a new special-purpose OS (for hotspot management) full-time for the past fifteen months, for just that reason. We can't afford to make it gratis/libris, but it won't catch infections, either.
Slashdot Mirror
"Yes, some people had problems, but for the most part this fairly unnatural motion on an inverted pendulum went smoothly."
Yeah, problems. Like the demonstrator I saw take a violent Segway header at the Minnesota State Fair last year. While he's giving his pitch, one of the wheels on his Segway hit a tiny patch of water on the slick floor. The wheel spun and smoked like a funny car doing a burn-out, and the guy went down so hard that his head bounced about six inches off of the floor. Next day he was back, with a cast on one arm, and a hockey helmet on his head.
Yeah, the inverted pendulum is a cute trick, but then again, so is riding a unicycle. I wouldn't advise my grandmother to ride either one. On the other hand, I would give her a Roomba. I own one, and apart from driving my pet beagle completely ballistic, it's great.
We could eliminate the entire virus/worm propagation problem by making a few hundred different builds of each release of programs like IE, IIS, and Windows, with each build having an added, random-sized dummy variable in each system call. The existence of a random-sized placebo variable on the stack would make each build's buffer overflows occur at different addresses. Bingo, no more homogenous population of buffer overflows!
Shucks, maybe I should have patented that idea before letting it out...
(Then again, maybe we don't really want to know).
So after my sorta' incendiary post about the PhysOrg's original "world's first 20-layer circuit board" headline being absurd, I went back and surfed the article to see the corrected headline. Got a good laugh out of the content of the Goooooogle ad that popped up just below the headline:
PCB Manufacturing
Order Quality PCBs 2 to 24 Layers. 12-Hour Turns & Same-Day Shipping.
Apparently Epson's breakthrough in layer counts caught on in record time!
Here's my letter to PhysOrg.com:
...contains an absurdly incorrect statement in its first sentence, to wit:
Hello,
The article...
http://www.physorg.com/news1789.html
"Seiko Epson Corporation today announced that it has succeeded in leveraging its proprietary inkjet technology to develop what the company believes is the world's first 20-layer circuit board."
No. Not even close. I have personally worked on circuit boards of as many as 48 layers, as long ago as 1985. The math coprocessor for the Sperry-1100/90 (code named "Eagle") had a motherboard that was roughly 20" x 36" in size, had forty-eight layers, was about 1/2" thick, had solid silver bus bars laminated in each side, weighed about forty pounds, and was so hard that if you knocked on it with your knuckle, it would ring like a bell.
There is no possibility that the people at Epson believe they've built the world's first twenty layer board. Twenty layer boards are a little uncommon, but far from record-breaking.
Sincerely,
BrakesForElves
Founder and past President
FASTechnologies, Inc. www.fastec.com
IMO: C surpassed Pascal because:
1) It's much easier to write a C compiler than a Pascal compiler, therefore the (early) availability of the C language on new platforms became a near certainty.
2) It didn't take project and product managers long to realize that in the era of Moore's Law, platform flexibility had great value. A project stuck on an obsolete platform due to the unavailability of its language on a revolutionary new platform was doomed, perhaps prematurely.
So its portability and ubiquity were C's most significant advantages over Pascal, back when there was a realistic contest.
3) For a time, executables written in C were likely to be considerably faster than those written in Pascal. This was a byproduct of the re-use of the C compiler code itself, versus fresh (read: immature) attempts at Pascal compilers. The C compiler cores got better with each processor port, but the freshly-written Pascal compilers often were not very good.
Today on the x86 platform with Borland's highly-refined 8th-generation compiler core, executables built from well-written Pascal are as fast as those built from coherently-written C, in my experience. It may be possible to write incredibly concise C that'll be a hair faster than the same thing written in Pascal, but arguing that difference is a fool's errand in the days of 4GHz rocket-ship machines executing septillions of NOPs waiting for something to do.
Personally, I choose Delphi these days over C, because I write and support huge projects. It is incredibly easy to pick up Pascal source and quickly figure out exactly what it does. That's the first (and most crucial) step in any software maintenance, and I find that Pascal's support cost savings more than over-balance any possible advantages I've ever realized from using C. When I'm writing something that needs to be extremely fast, I drop into inline assembly, but everything else I code in Delphi these days.
Thanks for pointing that out. No, ignoramus that I am, I was trying to do two things at once while reading the article. I'll read more carefully before I post on /. from now on. Cheers!
Either Mr. Chin is living in a cave, or he wrote this piece some time ago. With Firefox numbers skyrocketing and even CERT suggesting that running IE is inviting virus infections, his statement, "Internet Explorer will continue its chokehold on the World Wide Web" seems quite out of touch with present reality.
...And sadder that even with the instant, world-wide communications we have, people still suggest that comparing the representative government of the US to the repressive, muderous government of Stalin has any validity, at all, ever. In case nobody has noticed, Russia was ruled by Stalin alone, whereas the US is ruled by, er... well, _lawyers_ I guess. (But the line about Stalin still goes). .
We used to call it "Declaring Octvember", the mythical 45-day month at the end of the fiscal year. Y'know, the reason for stretching a quarter is that the lazy financial analysts fixate on simple things like a company missing its (or their) estimates, not whether the company's fundamentals and long-term prospects are sound. When lazy analysts drive corporate decision making, prepare for a parade of stupid decisions like declaring Octvember.
You might want to get your friends to write a new completely free operating system from scratch.
That's a better idea than you might think: If we didn't have basically two operating systems running several hundred million computers, we wouldn't see worms like Sasser zap around the world unchecked in twenty minutes. I've been working on a new special-purpose OS (for hotspot management) full-time for the past fifteen months, for just that reason. We can't afford to make it gratis/libris, but it won't catch infections, either.