Because an atomic clock only tells you how long a "second" is. Not which UTC second is currently elapsing or when the next one starts.
Take a walk down memory lane!
on
Happy Spamiversary!
·
· Score: 4, Informative
I vividly remember when Canter and Siegel spammed us on USENET. I even bought the "Green Card Lawyers - Spamming the Globe" T-Shirt from Joel Furr.
But I don't think that was actually the first widespread spam. A few months earlier -- in January 1994 -- was the similarly infamous "Global Alert For All: Jesus is Coming Soon" spam... does anyone remember that? It wasn't commercial spam per se, but still spam.
I spent the next few days collecting various funny responses to the spam from dozens of different newsgroups. A few years ago, I put my compilation on the web. Just doing my part to make sure nothing on the Internet ever dies.
First of all, the poster wasn't asking about "IT" or even "programming", they were asking about Computer Science. And Computer Science does kill people. Have you ever read the Risks digest? There are hundreds examples there.
The fact that licensing is not currently required is another story... for some jobs, I think it should be. But regardless of what the licensing situation is, it is unrealistic, and frankly insulting, to expect a shortcut into become serious and good at something.
"Kids out of HS", as you say, pick up IT like loose change. What does that mean, that they can write a VB app and put together a network? Would you trust such a kid out of HS to write software that runs the autopilot on a 747, or the heart monitor in your hospital, or for that matter the firmware of your cell phone?
How would you feel if you got the following question from a patient?
I am a computer scientist, but contemplating a career change perhaps 5 to 10 years down the road. In addition to computers, what I've always loved is anatomy and biology, and I think I have a pretty good appreciation for both. What tips could medical industry insiders offer to one who is willing to pursue an independent educational path towards a career in a Medical field? I'd like to start operating on people right away; Gray's Anatomy seems like a good guide, though I apparently can't get a license by reading it. How can an old newcomer break into the industry?
Why do you say that a new funding mechanism isn't progress?
Surely you see a difference between, say, a 900 foot flight in 1903, vs. technology so economically viable, and efficient, that today you can travel from one end of the country to the other in less than 5 hours for $200. And you even get free pretzels.
Doing something once doesn't make it a solved problem, as you seem to be implying. The real reason I think the Wrights deserve credit for the first "real" flight --- despite the hoardes of Slashdotters who always bring up various examples of someone who flew a feet 6 months earlier -- is that the Wrights made the technology viable. The others all did it once as basically a stunt, and never returned to it. The Wrights' continuous innovation and deep understanding of aerodynamics actually led, directly, to the subsequent ubiquity of airplanes. You can't say that about anyone else.
Something that is "mostly" driven by computer is still "partially" not driven by computer.
Do you really want to be going 650 miles per hour with nothing on board that knows how to deal with anomolies?
Strange things happen in flight that, on occasion, require the pilot to hand-fly the airplane for some reason. Sometimes a minor failure, sometimes a dramatic failure. Usually you don't even hear about it, because the pilot can still land safely. Would you really prefer the outcome of all those incidents to be a crash instead?
I'm sure the computer will do fine in the common case. It's the outlier cases that I worry about. I'd much rather have a human be pilot-in-command, not a computer that just throws up its hands and gives up.
Computerized trains and elevators can just stop if they see that they are operating outside of understood parameters. Planes don't have that option.
My employment contract says that everything I write either at work or at home technically belongs to my employer.
In California, that doesn't matter -- as the original poster pointed out, the California employment code specifically makes such clauses illegal. (That's one of the reasons I enjoy being employed in California.)
If your employment contract said "We own the right to your first-born child," and then tried to sue you to get custody, they wouldn't get very far. Contracts that violate the law are not legal.
As far as I understand it, it similar here. The contract is not enforceable because it is contrary to the law. I mean, the law even says that such employment contracts are unenforceable:
To the extent a provision in an employment agreement purports to require an employee to assign an invention otherwise excluded from being required to be assigned under subdivision (a), the provision is against the public policy of this state and is unenforceable.
When the clock first was announced, I remember reading that the point of it was not to keep accurate time, but to figure out how to create a device that could operate continuously for 10,000 years, surviving the changing social, political, technical, economic, and sociologic landscapes.
Today, after reading the site at length it seems they are using a traditional oscillator steered by the rotation of the earth. That's many orders of magnitude less accurate than one second in 10,000 years, if you're talking about an SI second. At the current rate of earth's rotation, it loses an SI (cesium) second about once every 18 months -- that's why we add leap seconds to UTC.
Perhaps you mean that, 10,000 years from now, the clock will still be aligned within one second to the earth's time (i.e. read 12:00 when the sun is overhead). But this is due to corrections applied to it every day by the sun, which as I mentioned is not, itself is not a particularly stable frequency standard.
I think the best mechanical clocks achieve something on the order of 1 second per month, such as Bill Scolnik's pendulum clock which is in a vacuum tank and mounted on a concrete slab to avoid environmental disturbances.
No, leap years (i.e., a year with an extra day inserted) are due to the fact that the earth doesn't take exactly 365 days to complete a full revolution around the sun.
Leap seconds are due to the fact that the Earth now takes slightly longer to complete a full rotation about its axis than it did when the length of an SI second was fixed.
The two are not interchangeable. Leap years ensure that winter comes in December. Leap seconds ensure that the sun is overhead at noon.
Not a standard. RFC 1644 [rfc-editor.org] is classed experimental; it's not a standards-track protocol.
You're right, of course. My claim of it being a "standard" was 1 part attempt to be pithy, 1 part annoyance at my two I-D's having been stranded in IESG limbo for 2 years.:-)
Perhaps you might want a few facts instead of making a snap judgement based on a 3-sentence description of these people found in a CNN article.
About 2 years ago, the New York Times Magazine had a serious (8000 word) article about this group. My impression from reading it at the time was that cloning might well be possible for a group that has enough money (they do), sufficient technical expertese (they've hired them), enough disregard for the moral implications (their beliefs support the idea), and enough volunteers to be implanted with cloned embryos (again, they do, due to the cult's ethos). Maybe only 1 out of every 100 embryos is viable, but when you actually have hundreds of cult members lining up as volunteers for implantation, how long is it until one of them succeeds?
If you are willing to pay $3 for some enlightenment, you can get the article here.
Re:later on the nature channel
on
Sensors Gone Wild
·
· Score: 4, Interesting
Laugh at sensors hopping...
It's very funny that you link to the DARPA SHM program. The Forbes "Sensors Gone Wild" article that Slashdot linked to today is talking about work done at the Center for Embedded Network Sensing at UCLA (and the closely associated UCLA LECS lab, also run by Deborah Estrin). Now, a few of us lowly graduate students working at the UCLA lab/center also work for Sensoria Corp, which was one of the main contractors for the SHM project. A lot of the research was very complementary. I'll plug my own research here -- the fine grained network time synchronization that we developed at UCLA/LECS is public domain and also made its way into the SHM project. There's other crossover as well (e.g. some of the acoustic ranging work); check out Sensoria's publications page.
I was at the SHM demo on an army base this past March and again this past August, and let me tell you, seeing those things actually hop is quite exciting. Especially when you're the one with your finger on the "ARM ALL" button:-).
Re:Or, if you need something even better than NTP.
on
Do You Have The Time?
·
· Score: 1
There are certain aspects I might try to port into NTP, for certain special cases (e.g. if you are synchronizing machines that are all on the same LAN). But the scheme doesn't actually work over the Internet in general (it requires a broadcast medium).
I'm definitely not out to replace NTP. I love NTP, I just don't think it's the right match for certain problems.
Or, if you need something even better than NTP...
on
Do You Have The Time?
·
· Score: 5, Interesting
Wow, who would have thought that the topic of my PhD thesis would be on the front page?
Right now I'm doing research in very high precision time synchronization for very large numbers of very small things. My lab does work in sensor networks -- get a tiny little computer with a few sensors and a radio, sprinkle thousands of them out over a building or a battlefield or a forest. Have the network tell you where the fire started, where the enemy is lurking, which light bulb needs to be replaced, or a thousand other things.
You need very time sync to do lots of this stuff -- to track motion, for example. Our current testbed times the flight of sound to tell how far apart things are, and for that we need accuracy on the order of 10 microseconds between clocks.
My research right now centers around a new time sync scheme, called Reference Broadcast Synchronization, which in a recent study I showed is almost an order of magnitude more precise than NTP under the same conditions -- 5 microseconds between a group of nodes with a userspace implementation, and down to 1 microsecond in the in-kernel implementation (which is the resolution of the clock! I'll do better when I have a clock that ticks more than once a microsecond.)
NTP, even under "optimal" conditions -- very high query rate to a stratum 1 GPS-steered clock in our lab--- did no better than 50 microseconds. When we introduced high levels of congestion on the network, NTP degraded by a factor of 30 while RBS was almost unchanged.
Of course, NTP is still a fantastic protocol, and much better than trying to apply RBS to the Internet (which is basically impossible). But for tiny nodes that need very tight time sync, I say, we can do better:-).
Some recent papers you might like are here, including
"Fine Grained Network Time Sync using Reference Broadcasts" -- the original RBS paper
"Wireless Sensor Networks: A New Regime for time synchronization" -- my argument as to why NTP shouldn't be used for sensor networks
"Locating nodes in time and space: A case study" -- description of our testbed that is capable of localizing objects down to 1cm by measuring time of flight of sound, combined with RBS time sync.
It's funny, I'm sitting in the lab right now, tinkering with the testbed when this article should come up.
RMS is not against BitKeeper because it is sold for money. The FSF has repeatedly said that they fully support people selling "free software" to support more free software. They even say right here: we encourage people who redistribute free software to charge as much as they wish or can.
What RMS is against, and what also makes me a little uneasy, is that I have no access to the source code to BitKeeper. The for-no-money version of BitKeeper is crippleware that only works under certain conditions; those conditions can later change. I feel uneasy about putting my entire change history into a format that, in the future, may no longer be accessible by any tool that I have.
Don't you think this is a reasonable thing to be concerned about? This is the issue. Not "Oh no, someone is charging money for free software!".
I think BitKeeper sounds like a very useful tool, and is leaps and bounds better than CVS. But I am afraid to use it for this reason.
NUMA is somewhere in between clustering (e.g. Beowulf) and SMP.
On a normal desktop machine, you typically have one CPU and one set of main memory. The CPU is basically the only user of the memory (other than DMA from peripherals, etc.) so there's no problem.
SMP machines have multiple CPUs, but each process running on each CPU can still see every byte of the same main memory. This can be a bottleneck as you scale up, since larger and larger numbers of processors that can theoretically run in parallel are being serviced by the same, serial memory.
NUMA means that there are multiple sets of main memory -- typically one chunk of main memory for every processor. Despite the fact that memory is physically distributed, it still looks the same as one big set of centralized main memory -- that is, every processor sees the same (large) address space. Every processor can access every byte of memory. Of course, there is a performance penalty for accessing nonlocal memory, but NUMA machines typically have extremely fast interconnects to minimize this cost.
Multi-computers, or clustering, etc. such as Beowulf completely disconnects memory spaces from each other. That is, each processor has its own independent view of its own independent memory. The only way to share data across processors is by explicit message-passing.
I think the advantage of NUMA over beowulf from the point of view of compiling a kernel is just that you can launch 32 parallel copies of gcc, and the the cost of migrating those processes to processors is nearly 0. With beowulf, you'd have to write a special version of 'make' that understood MPI or some other way of manually distributing processes to processors. Even with something like MOSIX, an OS that automatically migrates processes to remote nodes in a multicomputer for you, the cost of process migration is very high compared to the typically short lifetime of a typical instantiation of 'gcc', so it's not a big win. (MOSIX is basically control software on top of a beowulf style cluster, and the kernel mods needed to do transparent process migration)
I hope this clarified the situation rather than further confusing you.:-)
I grappled with exactly this problem for years. I wanted something that would give me superb quality Postscript/PDF, good HTML, and at least passable ASCII text. (In 1994, it was still important to distribute ASCII documentation; not everyone had a web browser.)
I went back and forth with all sorts of things: SGML based solutions, a few more "proprietary" utilities, etc. Finally, the latex-to-other-format conversion tools got to be good enough that I could use LaTeX as my primary format.
My most recent documentation is for FUSD, a Linux framework for user-space devices. The original documentation source is LaTeX. Simply running LaTeX gives you DVI, which you can convert into publication quality Postscript. Using pdflatex (NOT ps2pdf), you can also create very high quality PDF, which includes a real PDF table of contents, cross-references, and URL links. Finally, using latex2html, you can create almost native-quality HTML documentation. And, if you really need ASCII, you can get a reasonable rendering by running lynx (in its ASCII-dump mode) over the HTML.
latex2html comes with special LaTeX macros that let you specify hyperlinks inside your document. They are rendered as real hyperlinks in HTML, and footnotes in the printed versions.
The apartment building had large, J-shaped units with several bedrooms. (Actually, half the units were J and half were reverse J -- hence the U-shaped building, since they met at the middle.) My bedroom and my friend's were both symmetrically at the midpoint of the U, so we were separated by 6 inches of plasterboard as the crow flies. But, since the building was divided in half, giving him a copy of this week's Circuits homework required walking out to the front of the apartment (at the tip of the U), down 3 flights of stairs (no elevator in the building), going across the courtyard, getting buzzed in through security, up 3 flights of stairs -- then going all the way back.
The asymmetry of this situation was what was so comical. So, our plan was to attach a tape that was twice the length of the span between the windows, so that we could attach a bag to it and be able to pass objects back and forth. We did finally do it, too, around the back of the U instead of spanning the tips.
Please, say it isn't so! Say that 9-track tapes will still be available from somewhere! They are so useful for getting in trouble with the police!
I love 9-track tapes: it makes me feel like I'm a member of the old school when using them. Back in college, every other month, I'd stop by my University's mainframe center. They had a stack of retired 9-tracks there that said "Take One", and I'd help myself. I actually backed up some of my old mail onto one of them, using the CS department's old drive. But one day, my life changed. Someone told me that 9-track tapes are made with Kevlar: that's some tough-ass stuff, it is. My job was clear.
At the time, I was living on the 4th (top) floor of a U-shaped dorm, with about 100 feet between the "prongs". I lived at one edge, and as luck would have it, my best friend lived in the other prong. So, of course, we decided that we needed a tape wire running over the street from my window to his.
9-track should be perfect, right? After all, we did have about 6000 feet of it. And it's so strong and light. What could go wrong?
I gave my friend one reel, and I kept the other. 3AM on a Tuesday arrives. Our third accomplice, a friend of ours named Zaki, goes down to the street. My friend and I, in our windows, unreel enough tape that it reaches the ground, where Zaki grabs one end, hauls it over to the other end, and ties them together. My friend starts pulling up, and the tape began to rise ever so majectically over the street. It was a beautiful site.... until....
"What are you doing?" a cop's voice suddenly boomed, approaching Zaki, who was helplessly watching the tape rise above his head.
"Uh... just, uh.. running a little tape wire here, sir," he said, with surprising sincerity. The wire was now about at the level of the 3rd floor as my friend continued reeling it in.
"And what are you going to do when that falls, and hits someone in the head?" the cop inquired. Though not as politely as this text might suggest.
As if on cue, the tape became taut -- my friend had reeled in his side completely. It was at that moment that I realized that Zaki's tape attachment skills should not have been trusted. (After all, I'm the one who owns the Ashley Book of Knots - it should have been my job.) Yes, that's right: the tape came apart. My friend's side was safely in his apartment. But my side? Oh no. It was fluttering down towards the street, right towards the heads of Cop and Zaki, who were intently discussing the merits of the project.
I turned from the window - in a complete panic - and began RUNNING, tape reel in hand, as fast as I could through the apartment! Through the living room, through the corridor, past the kitchen and another corrdidor - with the tape trailing behind me. Finally, I reached my bedroom, and with no where left to run I started spinning in circles as fast as I could, wrapping the tape around me. When I finally fell onto by bed, dizzy, I could only hope that enough tape had been taken in through the window to avoid A Scene.
Luckily, it was. Like I said, that tape is strong.
So let this be a lesson to those of you still in college: use the back windows that face the alley, instead.
Hmmm - which unfortunately comes back to a density of photons, and a length cubed, which unforteunalty comes back to that damn platinum bar in Paris. IIRC - it has a chip in the corner of it too - Ooops. I just dropped your metre - my, how you have just grown!
So you're saying that if a product is meant to be non-lethal, but certain people use it carelessly in a way which causes their death, the maker of that product should not be sued?
That seems to make sense. But then why are we suing tobacco companies?
Slashdot Mirror
Because an atomic clock only tells you how long a "second" is. Not which UTC second is currently elapsing or when the next one starts.
I vividly remember when Canter and Siegel spammed us on USENET. I even bought the "Green Card Lawyers - Spamming the Globe" T-Shirt from Joel Furr.
But I don't think that was actually the first widespread spam. A few months earlier -- in January 1994 -- was the similarly infamous "Global Alert For All: Jesus is Coming Soon" spam... does anyone remember that? It wasn't commercial spam per se, but still spam.
I spent the next few days collecting various funny responses to the spam from dozens of different newsgroups. A few years ago, I put my compilation on the web. Just doing my part to make sure nothing on the Internet ever dies.
First of all, the poster wasn't asking about "IT" or even "programming", they were asking about Computer Science. And Computer Science does kill people. Have you ever read the Risks digest? There are hundreds examples there.
The fact that licensing is not currently required is another story... for some jobs, I think it should be. But regardless of what the licensing situation is, it is unrealistic, and frankly insulting, to expect a shortcut into become serious and good at something.
"Kids out of HS", as you say, pick up IT like loose change. What does that mean, that they can write a VB app and put together a network? Would you trust such a kid out of HS to write software that runs the autopilot on a 747, or the heart monitor in your hospital, or for that matter the firmware of your cell phone?
This question is insulting.
How would you feel if you got the following question from a patient?
I am a computer scientist, but contemplating a career change perhaps 5 to 10 years down the road. In addition to computers, what I've always loved is anatomy and biology, and I think I have a pretty good appreciation for both. What tips could medical industry insiders offer to one who is willing to pursue an independent educational path towards a career in a Medical field? I'd like to start operating on people right away; Gray's Anatomy seems like a good guide, though I apparently can't get a license by reading it. How can an old newcomer break into the industry?
Why do you say that a new funding mechanism isn't progress?
Surely you see a difference between, say, a 900 foot flight in 1903, vs. technology so economically viable, and efficient, that today you can travel from one end of the country to the other in less than 5 hours for $200. And you even get free pretzels.
Doing something once doesn't make it a solved problem, as you seem to be implying. The real reason I think the Wrights deserve credit for the first "real" flight --- despite the hoardes of Slashdotters who always bring up various examples of someone who flew a feet 6 months earlier -- is that the Wrights made the technology viable. The others all did it once as basically a stunt, and never returned to it. The Wrights' continuous innovation and deep understanding of aerodynamics actually led, directly, to the subsequent ubiquity of airplanes. You can't say that about anyone else.
Something that is "mostly" driven by computer is still "partially" not driven by computer.
Do you really want to be going 650 miles per hour with nothing on board that knows how to deal with anomolies?
Strange things happen in flight that, on occasion, require the pilot to hand-fly the airplane for some reason. Sometimes a minor failure, sometimes a dramatic failure. Usually you don't even hear about it, because the pilot can still land safely. Would you really prefer the outcome of all those incidents to be a crash instead?
I'm sure the computer will do fine in the common case. It's the outlier cases that I worry about. I'd much rather have a human be pilot-in-command, not a computer that just throws up its hands and gives up.
Computerized trains and elevators can just stop if they see that they are operating outside of understood parameters. Planes don't have that option.
In California, that doesn't matter -- as the original poster pointed out, the California employment code specifically makes such clauses illegal. (That's one of the reasons I enjoy being employed in California.)
If your employment contract said "We own the right to your first-born child," and then tried to sue you to get custody, they wouldn't get very far. Contracts that violate the law are not legal.
As far as I understand it, it similar here. The contract is not enforceable because it is contrary to the law. I mean, the law even says that such employment contracts are unenforceable:
Read it here.
Where do you get "one second in 10,000 years"?
When the clock first was announced, I remember reading that the point of it was not to keep accurate time, but to figure out how to create a device that could operate continuously for 10,000 years, surviving the changing social, political, technical, economic, and sociologic landscapes.
Today, after reading the site at length it seems they are using a traditional oscillator steered by the rotation of the earth. That's many orders of magnitude less accurate than one second in 10,000 years, if you're talking about an SI second. At the current rate of earth's rotation, it loses an SI (cesium) second about once every 18 months -- that's why we add leap seconds to UTC.
Perhaps you mean that, 10,000 years from now, the clock will still be aligned within one second to the earth's time (i.e. read 12:00 when the sun is overhead). But this is due to corrections applied to it every day by the sun, which as I mentioned is not, itself is not a particularly stable frequency standard.
I think the best mechanical clocks achieve something on the order of 1 second per month, such as Bill Scolnik's pendulum clock which is in a vacuum tank and mounted on a concrete slab to avoid environmental disturbances.
No, leap years (i.e., a year with an extra day inserted) are due to the fact that the earth doesn't take exactly 365 days to complete a full revolution around the sun.
Leap seconds are due to the fact that the Earth now takes slightly longer to complete a full rotation about its axis than it did when the length of an SI second was fixed.
The two are not interchangeable. Leap years ensure that winter comes in December. Leap seconds ensure that the sun is overhead at noon.
Not a standard. RFC 1644 [rfc-editor.org] is classed experimental; it's not a standards-track protocol.
:-)
You're right, of course. My claim of it being a "standard" was 1 part attempt to be pithy, 1 part annoyance at my two I-D's having been stranded in IESG limbo for 2 years.
Which is a standard What is everyone complaining about?
Perhaps you might want a few facts instead of making a snap judgement based on a 3-sentence description of these people found in a CNN article.
About 2 years ago, the New York Times Magazine had a serious (8000 word) article about this group. My impression from reading it at the time was that cloning might well be possible for a group that has enough money (they do), sufficient technical expertese (they've hired them), enough disregard for the moral implications (their beliefs support the idea), and enough volunteers to be implanted with cloned embryos (again, they do, due to the cult's ethos). Maybe only 1 out of every 100 embryos is viable, but when you actually have hundreds of cult members lining up as volunteers for implantation, how long is it until one of them succeeds?
If you are willing to pay $3 for some enlightenment, you can get the article here.
Laugh at sensors hopping...
:-).
It's very funny that you link to the DARPA SHM program. The Forbes "Sensors Gone Wild" article that Slashdot linked to today is talking about work done at the Center for Embedded Network Sensing at UCLA (and the closely associated UCLA LECS lab, also run by Deborah Estrin). Now, a few of us lowly graduate students working at the UCLA lab/center also work for Sensoria Corp, which was one of the main contractors for the SHM project. A lot of the research was very complementary. I'll plug my own research here -- the fine grained network time synchronization that we developed at UCLA/LECS is public domain and also made its way into the SHM project. There's other crossover as well (e.g. some of the acoustic ranging work); check out Sensoria's publications page.
I was at the SHM demo on an army base this past March and again this past August, and let me tell you, seeing those things actually hop is quite exciting. Especially when you're the one with your finger on the "ARM ALL" button
There are certain aspects I might try to port into NTP, for certain special cases (e.g. if you are synchronizing machines that are all on the same LAN). But the scheme doesn't actually work over the Internet in general (it requires a broadcast medium).
I'm definitely not out to replace NTP. I love NTP, I just don't think it's the right match for certain problems.
Right now I'm doing research in very high precision time synchronization for very large numbers of very small things. My lab does work in sensor networks -- get a tiny little computer with a few sensors and a radio, sprinkle thousands of them out over a building or a battlefield or a forest. Have the network tell you where the fire started, where the enemy is lurking, which light bulb needs to be replaced, or a thousand other things.
You need very time sync to do lots of this stuff -- to track motion, for example. Our current testbed times the flight of sound to tell how far apart things are, and for that we need accuracy on the order of 10 microseconds between clocks.
My research right now centers around a new time sync scheme, called Reference Broadcast Synchronization, which in a recent study I showed is almost an order of magnitude more precise than NTP under the same conditions -- 5 microseconds between a group of nodes with a userspace implementation, and down to 1 microsecond in the in-kernel implementation (which is the resolution of the clock! I'll do better when I have a clock that ticks more than once a microsecond.)
NTP, even under "optimal" conditions -- very high query rate to a stratum 1 GPS-steered clock in our lab--- did no better than 50 microseconds. When we introduced high levels of congestion on the network, NTP degraded by a factor of 30 while RBS was almost unchanged.
Of course, NTP is still a fantastic protocol, and much better than trying to apply RBS to the Internet (which is basically impossible). But for tiny nodes that need very tight time sync, I say, we can do better
Some recent papers you might like are here, including
- "Fine Grained Network Time Sync using Reference Broadcasts" -- the original RBS paper
- "Wireless Sensor Networks: A New Regime for time synchronization" -- my argument as to why NTP shouldn't be used for sensor networks
- "Locating nodes in time and space: A case study" -- description of our testbed that is capable of localizing objects down to 1cm by measuring time of flight of sound, combined with RBS time sync.
It's funny, I'm sitting in the lab right now, tinkering with the testbed when this article should come up.RMS is not against BitKeeper because it is sold for money. The FSF has repeatedly said that they fully support people selling "free software" to support more free software. They even say right here: we encourage people who redistribute free software to charge as much as they wish or can.
What RMS is against, and what also makes me a little uneasy, is that I have no access to the source code to BitKeeper. The for-no-money version of BitKeeper is crippleware that only works under certain conditions; those conditions can later change. I feel uneasy about putting my entire change history into a format that, in the future, may no longer be accessible by any tool that I have.
Don't you think this is a reasonable thing to be concerned about? This is the issue. Not "Oh no, someone is charging money for free software!".
I think BitKeeper sounds like a very useful tool, and is leaps and bounds better than CVS. But I am afraid to use it for this reason.
[I] dash to the doorframe where your supposed to be in an earthquake
Actually, that is a myth (#5). Hide under a table next time.
NUMA is somewhere in between clustering (e.g. Beowulf) and SMP.
:-)
On a normal desktop machine, you typically have one CPU and one set of main memory. The CPU is basically the only user of the memory (other than DMA from peripherals, etc.) so there's no problem.
SMP machines have multiple CPUs, but each process running on each CPU can still see every byte of the same main memory. This can be a bottleneck as you scale up, since larger and larger numbers of processors that can theoretically run in parallel are being serviced by the same, serial memory.
NUMA means that there are multiple sets of main memory -- typically one chunk of main memory for every processor. Despite the fact that memory is physically distributed, it still looks the same as one big set of centralized main memory -- that is, every processor sees the same (large) address space. Every processor can access every byte of memory. Of course, there is a performance penalty for accessing nonlocal memory, but NUMA machines typically have extremely fast interconnects to minimize this cost.
Multi-computers, or clustering, etc. such as Beowulf completely disconnects memory spaces from each other. That is, each processor has its own independent view of its own independent memory. The only way to share data across processors is by explicit message-passing.
I think the advantage of NUMA over beowulf from the point of view of compiling a kernel is just that you can launch 32 parallel copies of gcc, and the the cost of migrating those processes to processors is nearly 0. With beowulf, you'd have to write a special version of 'make' that understood MPI or some other way of manually distributing processes to processors. Even with something like MOSIX, an OS that automatically migrates processes to remote nodes in a multicomputer for you, the cost of process migration is very high compared to the typically short lifetime of a typical instantiation of 'gcc', so it's not a big win. (MOSIX is basically control software on top of a beowulf style cluster, and the kernel mods needed to do transparent process migration)
I hope this clarified the situation rather than further confusing you.
Excellent! I'd love to be able to generate PDF and PS from the exact same DVI source file; this looks like just the thing.
I grappled with exactly this problem for years. I wanted something that would give me superb quality Postscript/PDF, good HTML, and at least passable ASCII text. (In 1994, it was still important to distribute ASCII documentation; not everyone had a web browser.)
I went back and forth with all sorts of things: SGML based solutions, a few more "proprietary" utilities, etc. Finally, the latex-to-other-format conversion tools got to be good enough that I could use LaTeX as my primary format.
My most recent documentation is for FUSD, a Linux framework for user-space devices. The original documentation source is LaTeX. Simply running LaTeX gives you DVI, which you can convert into publication quality Postscript. Using pdflatex (NOT ps2pdf), you can also create very high quality PDF, which includes a real PDF table of contents, cross-references, and URL links. Finally, using latex2html, you can create almost native-quality HTML documentation. And, if you really need ASCII, you can get a reasonable rendering by running lynx (in its ASCII-dump mode) over the HTML.
latex2html comes with special LaTeX macros that let you specify hyperlinks inside your document. They are rendered as real hyperlinks in HTML, and footnotes in the printed versions.
The apartment building had large, J-shaped units with several bedrooms. (Actually, half the units were J and half were reverse J -- hence the U-shaped building, since they met at the middle.) My bedroom and my friend's were both symmetrically at the midpoint of the U, so we were separated by 6 inches of plasterboard as the crow flies. But, since the building was divided in half, giving him a copy of this week's Circuits homework required walking out to the front of the apartment (at the tip of the U), down 3 flights of stairs (no elevator in the building), going across the courtyard, getting buzzed in through security, up 3 flights of stairs -- then going all the way back.
The asymmetry of this situation was what was so comical. So, our plan was to attach a tape that was twice the length of the span between the windows, so that we could attach a bag to it and be able to pass objects back and forth. We did finally do it, too, around the back of the U instead of spanning the tips.
I love 9-track tapes: it makes me feel like I'm a member of the old school when using them. Back in college, every other month, I'd stop by my University's mainframe center. They had a stack of retired 9-tracks there that said "Take One", and I'd help myself. I actually backed up some of my old mail onto one of them, using the CS department's old drive. But one day, my life changed. Someone told me that 9-track tapes are made with Kevlar: that's some tough-ass stuff, it is. My job was clear.
At the time, I was living on the 4th (top) floor of a U-shaped dorm, with about 100 feet between the "prongs". I lived at one edge, and as luck would have it, my best friend lived in the other prong. So, of course, we decided that we needed a tape wire running over the street from my window to his. 9-track should be perfect, right? After all, we did have about 6000 feet of it. And it's so strong and light. What could go wrong?
I gave my friend one reel, and I kept the other. 3AM on a Tuesday arrives. Our third accomplice, a friend of ours named Zaki, goes down to the street. My friend and I, in our windows, unreel enough tape that it reaches the ground, where Zaki grabs one end, hauls it over to the other end, and ties them together. My friend starts pulling up, and the tape began to rise ever so majectically over the street. It was a beautiful site.... until....
"What are you doing?" a cop's voice suddenly boomed, approaching Zaki, who was helplessly watching the tape rise above his head.
"Uh... just, uh.. running a little tape wire here, sir," he said, with surprising sincerity. The wire was now about at the level of the 3rd floor as my friend continued reeling it in.
"And what are you going to do when that falls, and hits someone in the head?" the cop inquired. Though not as politely as this text might suggest.
As if on cue, the tape became taut -- my friend had reeled in his side completely. It was at that moment that I realized that Zaki's tape attachment skills should not have been trusted. (After all, I'm the one who owns the Ashley Book of Knots - it should have been my job.) Yes, that's right: the tape came apart. My friend's side was safely in his apartment. But my side? Oh no. It was fluttering down towards the street, right towards the heads of Cop and Zaki, who were intently discussing the merits of the project.
I turned from the window - in a complete panic - and began RUNNING, tape reel in hand, as fast as I could through the apartment! Through the living room, through the corridor, past the kitchen and another corrdidor - with the tape trailing behind me. Finally, I reached my bedroom, and with no where left to run I started spinning in circles as fast as I could, wrapping the tape around me. When I finally fell onto by bed, dizzy, I could only hope that enough tape had been taken in through the window to avoid A Scene.
Luckily, it was. Like I said, that tape is strong.
So let this be a lesson to those of you still in college: use the back windows that face the alley, instead.
It's possible to implement user-Space devices in Linux, too.
The platinum bar no longer defines the meter. A meter is defined as the distance travelled by light through a vacuum in a particular amount of time. Time, of course, being the most accuruate realization of a unit that mankind has yet achieved.
That seems to make sense. But then why are we suing tobacco companies?