Actually most servers (including Apache and IIS) let you do just that.
Think of almost every Linux distribution - they use FTP.
Debian's apt-get default config file contacts servers via HTTP. The debian admins (coincidentally the same people that wrote the http and ftp clients built in to apt-get) concluded that http was a better default.
Ever tried downloading something, connected at a smooth 100k/s or something and found, a few minutes later, you're pulling 2k/s? That USUALLY doesn't happen with FTP.
Actually it happens exactly as often with HTTP as with FTP, for exactly the same reasons, neither of which have anything to do with the protocols.
But for anything of size, do *yourself* a favor and go with FTP.
I would thoroughly disagree with this opinion. Get a resuming HTTP client.
Or even 24 times a day, if you've got people from 24 different timezones (or to complete the analogy, schools of software design) comparing it to theirs.
Heck, I'd be happier if someone actually DID make an "ultra-quiet" CPU fan... not talking about slowing down the speed of the fan with technology, but actually making it more of a muffled sound without killing it's specs and making it useless.
Noise is a function of air turbulence (and crappy bearings). Fan tip speed directly increases turbulence.
Get a bigger fan, run it more slowly (like your case fan), and speed up the air with a venturi duct before it hits the CPU.
Of course, this is difficult or impractical for 1U cases.
One Duron 700mhz system I have is set up this way - air is admitted from the back of the case through a cardboard duct to a large fan which bolts to a balsawood frame which screws into the heatsink (this fan moves more air than is required by the CPU itself). This forces air out any other exit (mainly through the power supply, which has had its own fan removed). It's about the best I can do without depending on passive cooling or a heat pipe.
The thermal efficient of their engine is about 30 percent, compare that to automobile engine. which is about 25 percent.
A typical automotive gasoline engine has a brake efficiency of 38% these days. State of the art is over 40% and the DOE believes natural gas engines will do 50% by 2010. 25% is a rating I'd expect from something with an FAA certification.
Seeing, for example, that a 2N2222 bipolar NPN transistor is required for an amplifier isn't going to be too useful in the year 2100, I would bet.
Unless semiconductors themselves go out of style (like vacuum tubes), I don't see a reason for not being able to buy a 2N2222-compatible part (or a 555 chip, or anything that widely popular, etc) in the year 2100.
I have a secondhand copy of the fifth edition of the Motorola Semiconductor Data Book, copyright 1970. Most diodes and transistors (basically any 1N or 2N part) that I can buy today are listed there. It lists every physical and electrical characteristic of the 2N2222 I could possibly want to know. The paper is quite white and no pages have fallen out or been ripped. It has received no special care.
So not only would semiconductors have to go out of style but all copies of those data books (and any other work that includes the JEDEC reference data) would need to be destroyed.
Heck, if it's really that important, we can still make a vacuum tube, even if the last one of a particular type is destroyed.
Now, that PROM in your doohickey that lets you unencrypt your thingamabob so it can bootstrap your DRM-compliant whatchamacallit that got a bit too much ultraviolet light over the years... you may be out of luck, especially if there's no chip markings.
Visa owns the trademarks to both E-Visa and to eVisa.
Only in the credit card trade, I would imagine. Visa would have to either prove that their eVisa trademark is 'famous', or prove that evisa.com is in the credit card business to win this suit.
It doesn't take a whole lot of imagination to come up with some very scary scenarios of what could have been put there instead of "Hacked by Chinese!" After all, how many people visiting Windows Update are running versions of IE without known run-arbitrary-code security holes?
And due to the vague commerce clause in the constitution, the courts have no choice but to uphold the constitutionality of it all.
I get the feeling that the Supreme Court is waiting for the right case to come along to put Congress in their place on that one.
"Certainly what is happening [...] under the Commerce Clause is totally
different than what the Framers had in mind."
-- Supreme Court Justice William Rehnquist, Eldred vs. Ashcroft
Can anyone see a problem with 20 foot long extremely well shielded low resistance power supply cables?
Don't worry about the shielding. It's DC. As for the voltage drop, let's figure it out using some basic assumptions: your power supply has 18AWG wires, your motherboard is most sensitive to 5v sag, you have 1 foot cables currently, and that you are currently drawing 100 watts on the 5v line across 6 of these 18awg wires.
Ohm's law says that a load drawing 100 watts at 5 volts is.05 ohms. 18awg wire is.00751 ohms per foot, divided by six wires is.00125 ohms per foot. Round trip for your existing power supply is.0025 ohms, or 4.7% of your total load resistance. Therefore your voltage drop is.238 volts.
Extended to 40 feet of cable round trip, we've got.05 ohms of resistance. Which is the same as your load. Which means half the voltage gets used up in the motherboard, half gets used up in the wire. Unacceptable.
Same numbers with a single run of 4-gauge wire:.000292 ohms per foot, 40 feet,.012 ohms,.012+.050=.062, 5*(.050/.062)=4.03v. Still unacceptable.
You'll want to run your own numbers to make sure, but it looks like you'll need stupid amounts of copper for this.
It's the fault of the labels for signing every jackass garage band it 'discovers' to multi-album contracts.
Other reasons:
CD price fixing - Why take a risk on bands you've never heard of when the album is $15 instead of $5?
Radio - Clearchannel et al. make more money with smaller playlists.
I'm sure there's lots of other things to take onto that list, but my point is the RIAA has a lot more control over that statistic than they will admit. Therefore one must conclude that they really like the status quo.
It's mostly bullshit, but they probably estimate it by estimating the wear on the moving parts and extending the data out until the parts are so out of spec to cause failure.
No, that's the Rated Life, which is a completely different spec that rarely makes an appearance in advertising. Which is a shame, because MTBF statistics are useless without knowing the rated life statistic.
A MTBF of one million hours means that if you have (for example) 1000 drives, and replace ALL of them at the end of their rated life, you will experience a failure every 1000 hours (every 6 weeks).
Oh, and anyone who maintains a farm of machines with 1000 of these drives will find out just how fake those specs are.
I know exactly what I mean by "heterodyne detection."
The reason other people don't is because the word heterodyne is more typically used to discuss combining two signals for some intended purpose - for example, the combining of a carrier signal and an audio frequency signal. So, it's correct to say that any AM signal is a "heterodyne happening." It's also correct to say that two overlapping AM signals are a "heterodyne happening." You can avoid vagueness by referring to the latter as "interference".
It's not human nature to attempt to disect an already familiar word's base meaning when the context seems wrong.
Re:One of my favourite quotes...
on
Want Freedom?
·
· Score: 5, Informative
Great quote. I just wish we knew who said it. It'safake.
Re:Over selling not inherently evil...
on
How to Test Your T1?
·
· Score: 3, Interesting
any time a network segment reaches 80% utilization at any point in the day
Numbers are so much fun. I'm sure you already know this, but it's worth pointing out:
Network links (and CPU's, for that matter) at any given instant in time are either at 0% utilization or 100% utilization. Anything in the middle is an average, so you need to ask what it's an average of. For stock MRTG or Cricket setups, it's a five minute average - a spike to 100% means the line was pegged solid for at least five minutes. If the 80% threshold you mentioned is more like an hour average, the line may well have been fully saturated for 48 minutes of that hour.
Oh, and it's useful to ask questions about this stuff before signing any service contract that includes burst traffic fees.
I'm probably going to regret asking this, but how can a microwave oven be used as a communication tool?
"Some disassembly required"
The magnetron is an extremely high-power wideband 2.4GHz RF transmitter, whose waveguide can be routed to something like a parabolic dish instead of the customary faraday cage with food inside. This is how radar works.
Magnetrons and Klystrons scale to hundreds of kilowatts. For some interesting reading, do a google search on it.
Yes skew if a very big deal for differential signals of any speed. Sure maybe adding an inch on a USB cable won't hurt, but more than 3 inches will.
The wavelength of a 24mhz signal (f.e. USB) is 12.5 meters. Let's round 3 inches up to.1 meters - it's.8% of our wavelength, or 2.88 degrees. 1 - cos(2.88) =.0012, or.12% of RMS. Have you been living in wonder at how people make a 250mhz signal go across these lines? (gigabit ethernet)
These are all issues if you do not know that they can cause problems.
Ahh, so that's why having a technician look at a computer problem instantly makes it unreproducible.
Ethernet is also a differential signal - signal skew is simply not that big of a deal, especially at USB speeds. Light travels a long ways in one clock cycle.
Impedance also isn't a big deal. According to this, it's supposed to be 90 ohms, and the spec allows for 15% variation. Cat5 is about 100 ohms.
What is important is timing. According to this, the low-speed timing limitation is 18ns, or 300000000*.000000018 = 5.4 meters. Significantly under 30 feet.
This reminds me of third grade when I'd program the Commodore 64 in the back of the classroom to generate a constant 18khz tone designed to covertly give everyone in the back row a headache. At the time, I was convinced they all deserved much worse.:)
Slashdot Mirror
False premise, false implication, false conclusion.
Actually most servers (including Apache and IIS) let you do just that.
Think of almost every Linux distribution - they use FTP.
Debian's apt-get default config file contacts servers via HTTP. The debian admins (coincidentally the same people that wrote the http and ftp clients built in to apt-get) concluded that http was a better default.
Ever tried downloading something, connected at a smooth 100k/s or something and found, a few minutes later, you're pulling 2k/s? That USUALLY doesn't happen with FTP.
Actually it happens exactly as often with HTTP as with FTP, for exactly the same reasons, neither of which have anything to do with the protocols.
But for anything of size, do *yourself* a favor and go with FTP.
I would thoroughly disagree with this opinion. Get a resuming HTTP client.
Queueing up files is up to the client and has nothing to do with the protocol or server.
Or even 24 times a day, if you've got people from 24 different timezones (or to complete the analogy, schools of software design) comparing it to theirs.
Noise is a function of air turbulence (and crappy bearings). Fan tip speed directly increases turbulence.
Get a bigger fan, run it more slowly (like your case fan), and speed up the air with a venturi duct before it hits the CPU.
Of course, this is difficult or impractical for 1U cases.
One Duron 700mhz system I have is set up this way - air is admitted from the back of the case through a cardboard duct to a large fan which bolts to a balsawood frame which screws into the heatsink (this fan moves more air than is required by the CPU itself). This forces air out any other exit (mainly through the power supply, which has had its own fan removed). It's about the best I can do without depending on passive cooling or a heat pipe.
A typical automotive gasoline engine has a brake efficiency of 38% these days. State of the art is over 40% and the DOE believes natural gas engines will do 50% by 2010. 25% is a rating I'd expect from something with an FAA certification.
Unless semiconductors themselves go out of style (like vacuum tubes), I don't see a reason for not being able to buy a 2N2222-compatible part (or a 555 chip, or anything that widely popular, etc) in the year 2100.
I have a secondhand copy of the fifth edition of the Motorola Semiconductor Data Book, copyright 1970. Most diodes and transistors (basically any 1N or 2N part) that I can buy today are listed there. It lists every physical and electrical characteristic of the 2N2222 I could possibly want to know. The paper is quite white and no pages have fallen out or been ripped. It has received no special care.
So not only would semiconductors have to go out of style but all copies of those data books (and any other work that includes the JEDEC reference data) would need to be destroyed.
Heck, if it's really that important, we can still make a vacuum tube, even if the last one of a particular type is destroyed.
Now, that PROM in your doohickey that lets you unencrypt your thingamabob so it can bootstrap your DRM-compliant whatchamacallit that got a bit too much ultraviolet light over the years... you may be out of luck, especially if there's no chip markings.
Only in the credit card trade, I would imagine. Visa would have to either prove that their eVisa trademark is 'famous', or prove that evisa.com is in the credit card business to win this suit.
Things traveling at over 3% of the speed of light are categorically not in orbit around the sun.
No, but they have been cracked before:. html
http://www.attrition.org/security/commentary/ms16
It doesn't take a whole lot of imagination to come up with some very scary scenarios of what could have been put there instead of "Hacked by Chinese!" After all, how many people visiting Windows Update are running versions of IE without known run-arbitrary-code security holes?
I get the feeling that the Supreme Court is waiting for the right case to come along to put Congress in their place on that one.
Don't worry about the shielding. It's DC. As for the voltage drop, let's figure it out using some basic assumptions: your power supply has 18AWG wires, your motherboard is most sensitive to 5v sag, you have 1 foot cables currently, and that you are currently drawing 100 watts on the 5v line across 6 of these 18awg wires.
Ohm's law says that a load drawing 100 watts at 5 volts is .05 ohms. 18awg wire is .00751 ohms per foot, divided by six wires is .00125 ohms per foot. Round trip for your existing power supply is .0025 ohms, or 4.7% of your total load resistance. Therefore your voltage drop is .238 volts.
Extended to 40 feet of cable round trip, we've got .05 ohms of resistance. Which is the same as your load. Which means half the voltage gets used up in the motherboard, half gets used up in the wire. Unacceptable.
Same numbers with a single run of 4-gauge wire: .000292 ohms per foot, 40 feet, .012 ohms, .012+.050=.062, 5*(.050/.062)=4.03v. Still unacceptable.
You'll want to run your own numbers to make sure, but it looks like you'll need stupid amounts of copper for this.
Only if you accept them. Are you familiar with the legal term "quid pro quo"? It's one of the fundamentals of contract law.
It's never the license that denies you rights. Licenses can only license you rights.
The only truly public license is no copyright.
False. It's actually dependant on bandwidth and signal-to-noise ratio. See Shannon's Law.
Other reasons:
I'm sure there's lots of other things to take onto that list, but my point is the RIAA has a lot more control over that statistic than they will admit. Therefore one must conclude that they really like the status quo.
No, that's the Rated Life, which is a completely different spec that rarely makes an appearance in advertising. Which is a shame, because MTBF statistics are useless without knowing the rated life statistic.
A MTBF of one million hours means that if you have (for example) 1000 drives, and replace ALL of them at the end of their rated life, you will experience a failure every 1000 hours (every 6 weeks).
Oh, and anyone who maintains a farm of machines with 1000 of these drives will find out just how fake those specs are.
Yeah, that abstract is scary - it's one of the longest run-on sentences I've seen in a long time.
The reason other people don't is because the word heterodyne is more typically used to discuss combining two signals for some intended purpose - for example, the combining of a carrier signal and an audio frequency signal. So, it's correct to say that any AM signal is a "heterodyne happening." It's also correct to say that two overlapping AM signals are a "heterodyne happening." You can avoid vagueness by referring to the latter as "interference".
It's not human nature to attempt to disect an already familiar word's base meaning when the context seems wrong.
Great quote. I just wish we knew who said it. It's a fake.
Numbers are so much fun. I'm sure you already know this, but it's worth pointing out:
Network links (and CPU's, for that matter) at any given instant in time are either at 0% utilization or 100% utilization. Anything in the middle is an average, so you need to ask what it's an average of. For stock MRTG or Cricket setups, it's a five minute average - a spike to 100% means the line was pegged solid for at least five minutes. If the 80% threshold you mentioned is more like an hour average, the line may well have been fully saturated for 48 minutes of that hour.
Oh, and it's useful to ask questions about this stuff before signing any service contract that includes burst traffic fees.
"Some disassembly required"
The magnetron is an extremely high-power wideband 2.4GHz RF transmitter, whose waveguide can be routed to something like a parabolic dish instead of the customary faraday cage with food inside. This is how radar works.
Magnetrons and Klystrons scale to hundreds of kilowatts. For some interesting reading, do a google search on it.
The wavelength of a 24mhz signal (f.e. USB) is 12.5 meters. Let's round 3 inches up to .1 meters - it's .8% of our wavelength, or 2.88 degrees. 1 - cos(2.88) = .0012, or .12% of RMS. Have you been living in wonder at how people make a 250mhz signal go across these lines? (gigabit ethernet)
These are all issues if you do not know that they can cause problems.
Ahh, so that's why having a technician look at a computer problem instantly makes it unreproducible.
Impedance also isn't a big deal. According to this, it's supposed to be 90 ohms, and the spec allows for 15% variation. Cat5 is about 100 ohms.
What is important is timing. According to this, the low-speed timing limitation is 18ns, or 300000000*.000000018 = 5.4 meters. Significantly under 30 feet.
Size of solder joints... sheesh.
This reminds me of third grade when I'd program the Commodore 64 in the back of the classroom to generate a constant 18khz tone designed to covertly give everyone in the back row a headache. At the time, I was convinced they all deserved much worse. :)