The "best model" in 1995 mispredicted the temperature in 2000 by 300%.
Let's see... The earth's average temperature surface temperature is about 288K (15C,59F). That would mean that the "best model" either predicted an average surface temperature of 96K (-177C, -287F) or it predicted a temperature of 864K (591C,1095F).
Either that one sucky model, or you're a lying sack of sh*t.
> You are talking about "massive" tables and a "reporting" application. Of course Oracle and DB/2 are the right choices for this. Has anyone ever thought differently?
Yes - everytime a product or technology is overhyped people believe it will do everything. Search for mysql and data warehouse - you'll find plenty of people who think it can handle massive data without really understanding what db2/oracle/informix/etc do that's different.
That is especially true for PHBs. Some hear "free" and assume it will do everything that their $XXk/year database does, without spending tens or hundreds of kilobucks per year. I haven't used MySQL 5.0 yet, but if it's like 4.X, it'll be fine and speedy for databases smaller than memory. It'll also be fine so long as you don't mind some scheduled downtime to compact your frequently updated tables. It'll be fine as long as you excise "count(*)" from your SQL vocabulary.
MySQL (at least version 4) has a long long way to go in query processing and index searching. I hope to be pleasantly surprised by MySQL 5, but I'll bet that were still going to be using 5 year old copies of a dead product (informix) for tables with more than a 10 Mrows or so. (Our biggest tables are in the few gigarow range.)
I spent six years working on my book on evenings and weekends. I paid for the printing out of my pocket. I don't expect to be well-paid for my efforts, but I do expect to be paid.
That's funny, because I usually don't get paid for what I do on weekends and evenings. And yes, I generate copyrighted works during much of that time. The all of them I publish without charging for them. I do it because I enjoy doing it and I like spreading knowledge. So one question is, "Would you have not bothered to write it if you didn't think you would be paid?"
A couple more questions to illustrate some of the potential problems with current copyright law...
The next question is how much compensation is enough? Let's say you put in a couple thousand hours on it. At $20 an hour, that's $40k. Is that enough? Would place it in the public domain after you've made $40k? $100k? $1M? $10M? $100M? Or are there no limits to your entitlement to profit?
Now lets say a publishing giant has noticed your work and wants to buy the copyright from you and you agree on a price. It turns out they sell a competing work on the same subject. Their plan for your work is to take it out of print and keep it there, so people will be forced to buy their inferior rendition that has higher profit margins because it's only got half as many pages. How long should they be able to prevent publication of your work?
Next hypothetical: You decided not to sell to the megapublisher.
Let's say you're going to live another 50 years. In your will, would you make your work public domain? Or are your children entitled to profit from you work? They didn't write it, did they? How many generations deserve to get a ride based upon the couple thousand hours you worked? Is 100 years after you die long enough? Or should your descendents have perpetual rights to income from your work?
But maybe you didn't tell your kids about it and lost the original manuscript before you died. Should someone finding a copy of your book but having no means to find the owner of the copyright be prevented from making a copy for a friend?
What if you had no descendents? Is the copyright still valid?
Note the this is "Hurricanes that hit the U.S." rather than number and severity of Atlantic hurricanes. There a rather significant difference between the two.
You'd (I mean, I) think the graphics card companies would donate some services to help out with the task...
We did get some financial support from a graphics card manufacturer and employed some students on the task for a summer. But corporate priorities change, or maybe they weren't happy with the rate of progress, and the manufacturer didn't make good on promises of subsequent support, which prompted the university to withdraw matching funds, which meant we really couldn't afford to work on it further....
Hence the hope that interested volunteers step up.
I wish BOINC could also be designed to use graphics cards - ala the BrookGPU project - to help with the number crunching duties.
So do I. In fact I keep looking for people to help us develop this.... To no avail.:( Aparently the people who want this most don't have the ability to implement it, and the people who have the ability (assuming they exist) aren't interested.
If anyone wants to help, join the boinc_opt mailing list and send a message.
BTW, David is the titular director of SETI@home, but currently has no managerial duties beyond the BOINC project.
Bzzrt. Wrong answer. Motion is relative, acceleration is not.
Bzzrt. Everyone who doesn't quite understand relativity gets this one wrong. Right answer, wrong reason. The acceleration isn't important. The velocity is.
Think of it this way. Suppose one twin went to a star 100 light years distant at 0.99995c. Assume he has a magic ship that doesn't accelerate. Just boom and it's moving....
To the twin that stayed behind, he see the ship take 100 years to get to the star and 100 years to get back, (i.e. 200 years round trip).
Now look at it from the perspective of the twin on the ship. He turns on the engines and suddenly he's moving at 0.99995c. He gets out his handy dandy ruler and measures the distance to his destination. As expected, it's 100*sqrt(1-v^2/c^2) or 1 light year. Also as expected the trip takes 1 year at 0.99995c. When he gets there, he turns the ship around and starts the trip back, which takes another year.
So when he gets back to earth he sees that during his two year trip, his brother has aged 200 years.
Newton's Laws are perfectly fine... force... momentum... energy.
Yes, there are perfectly fine until you realize that force and momentum are artificial concepts that are required in Newton's Laws because the Newton's laws are an incomplete description of the physics involved. Even the concept of kinetic energy is artificial. Kinetic energy is just a term in the taylor series expansion of E=mc^2 around v=0.
I did a search on my name on both Google and Yahoo. Yahoo claims 47,300. Google claims 38,700. Do I care which returns more? Not in the least, since nobody is going to look past the first few pages.
The big difference is that on Google, my home page is #1. On Yahoo, #1 is an innodb bug report I filed in January. A page about memory mapped files several links below my home page is #3, another of assorted links I like is #11, and my home page isn't in the first 100.
Which search engine do you think generates more useful results?
107 Shuttle flights, 17 Apollo with 7 attempts to the moon (6 successes). In flight Apollo deaths Zero, (according to my count) in flight Shuttle Deaths, 13 (also according to my count).
Which is to say that there is essentially no data on how hazardous Apollo missions were, while the data on the safety of shuttle missions is merely sparse.
I think you should also count training/preparation deaths for each, since you can't fly the missions without the prep.
Do a little statistics and you'll see that the shuttle failure rate is (assuming random single point catastrophic failure) is somwhere between 0% and 10% with the measured rate being about 2%.
Do the same for Apollo and you'll get a catastrophic failure rate of between 0% and 22% with the measured rate being 11% (Apollo 1 and Apollo 13). By this measure, there's a good probability that the shuttle is the winner.
Lets try other measures (skiping the statistical treatment).
Vehicles lost per mission: Apollo 0.05, Shuttle 0.02
Deaths per mission: Apollo 0.16, Shuttle 0.12
Deaths per astronaut-day: Apollo 0.009, Shuttle 0.002
There's no real evidence that Apollo was any safer than the Shuttle is. There's also no real reason to believe that the next vehicle will be any safer.
Now that I've moved the (0.5 l/s) water pump and aquarium into the garage, a water cooled power supply is my next home project. I figure I'll build a sealed enclosure for the supply, fill it with transformer oil, and run a copper tube through it for heat transfer. The heatpipe solutions are neat, but I don't need any burns from touching the radiator.
At that point, the only air cooling in my system would be the SDRAM.
Yes, cygwin makes common Unix shells available on windows, but it's just a CLI. It doesn't interact with the rest of windows, the registry, other user space apps, etc.
Bzzztt! (for the registry at least). Cygwin presents the registry as it should be....
A true liberal wouldn't call for government enforced sale of a corporation's assets. An authoritarian would, but not a liberal.
A true liberal would counter that the idea enshrined in our law that a corporation is equivalent to a person threatens our liberty. The idea currently promoted by our government that a corporation cannot be held responsible for violating the law is downright dangerous.
The punishment for violation of antitrust laws can include enforced sale of a corporations assets. You may not agree, but I'm sure you'll jump up and down supporting the idea that people can be deprived of their lives for committing a crime.
"Pay to publish" is the norm in many scientific disciplines. In astronomy for example, the top journals charge authors ~$75 per page. There is actually benefit to this because the authors in most astronomy journals do not give up electronic distribution rights for their publications.
Meanwhile ad or subscription supported journals tend to charge fairly steep fees for electronic access to "their" articles.
Its tough for an author supported journal to say "We're going to charge you in order to publish this, but we're not going to let you distribute it in other ways." Hence, "pay to publish" journals foster free access to journal articles.
If it's standard C, NULL must be zero. There were, in the past, compilers where this was not the case. (Some were compilers for MS-DOS, where offset zero in the data segment might actually be used, and at least one where a zero offset with any segment register value would compare equal to zero. I think older version of VAX C also had a non-zero value for an invalid pointer).
Of course, now, any platform that uses a non-zero in memory value for a null pointer must hide that fact from the programmer. In other words:
/* temporarily ignore the nasal demons that this * code invokes. */
char *a=NULL; /* This assertion must succeed */ assert(a==0);
/* ignore invalid assumption that sizeof(unsigned long)==sizeof(char *) */
/* The following line could do anything, but will do what we want */ unsigned long c=*(unsigned long *)&a;
/* This assertion may fail. If it does, it's probably * because the memory representation of a NULL pointer is not * really zero */ assert(c==0);
I'm not trying to flamebait either. I think OS X is a good desktop operating system. But it's certainly not ready for prime time as a server.
I've had nothing but trouble integrating MacOS X as a server into an existing linux and solaris network. There is no commonality with administration tools and technique. NIS support is an afterthought. Mount maps need to be massaged. The NFS server rejects connections randomly. The admin interface changes from OS rev to OS rev, sometimes without backward compatibility. (Gee, I upgraded, why doesn't NFS work anymore?) And to top it all, you're still limited to 32 bit addressing in your processes. Oh, and there's no RAID5 software that really works in a high availability application.
Then there are occasion random crashes under load and "can't create process-resource not available" problems...
If you are used to administering UNIX systems, do yourself a favor, get yourself a cheap multiprocessor Opteron, install your favorite Linux or *BSD distro and get some work done. If you must have a Mac, go ahead and buy one, but be sure you wipe the drive and install Linux or *BSD on it.
Slashdot Mirror
Let's see... The earth's average temperature surface temperature is about 288K (15C,59F). That would mean that the "best model" either predicted an average surface temperature of 96K (-177C, -287F) or it predicted a temperature of 864K (591C,1095F).
Either that one sucky model, or you're a lying sack of sh*t.
If your tables are small, just dump them to disk and reload after the update.
Call me ten years ago when this was a medium sized database. Now if it's less than 1TB, it's a small database.
Yes - everytime a product or technology is overhyped people believe it will do everything. Search for mysql and data warehouse - you'll find plenty of people who think it can handle massive data without really understanding what db2/oracle/informix/etc do that's different.
That is especially true for PHBs. Some hear "free" and assume it will do everything that their $XXk/year database does, without spending tens or hundreds of kilobucks per year. I haven't used MySQL 5.0 yet, but if it's like 4.X, it'll be fine and speedy for databases smaller than memory. It'll also be fine so long as you don't mind some scheduled downtime to compact your frequently updated tables. It'll be fine as long as you excise "count(*)" from your SQL vocabulary.
MySQL (at least version 4) has a long long way to go in query processing and index searching. I hope to be pleasantly surprised by MySQL 5, but I'll bet that were still going to be using 5 year old copies of a dead product (informix) for tables with more than a 10 Mrows or so. (Our biggest tables are in the few gigarow range.)
That's funny, because I usually don't get paid for what I do on weekends and evenings. And yes, I generate copyrighted works during much of that time. The all of them I publish without charging for them. I do it because I enjoy doing it and I like spreading knowledge. So one question is, "Would you have not bothered to write it if you didn't think you would be paid?"
A couple more questions to illustrate some of the potential problems with current copyright law...
The next question is how much compensation is enough? Let's say you put in a couple thousand hours on it. At $20 an hour, that's $40k. Is that enough? Would place it in the public domain after you've made $40k? $100k? $1M? $10M? $100M? Or are there no limits to your entitlement to profit?
Now lets say a publishing giant has noticed your work and wants to buy the copyright from you and you agree on a price. It turns out they sell a competing work on the same subject. Their plan for your work is to take it out of print and keep it there, so people will be forced to buy their inferior rendition that has higher profit margins because it's only got half as many pages. How long should they be able to prevent publication of your work?
Next hypothetical: You decided not to sell to the megapublisher. Let's say you're going to live another 50 years. In your will, would you make your work public domain? Or are your children entitled to profit from you work? They didn't write it, did they? How many generations deserve to get a ride based upon the couple thousand hours you worked? Is 100 years after you die long enough? Or should your descendents have perpetual rights to income from your work?
But maybe you didn't tell your kids about it and lost the original manuscript before you died. Should someone finding a copy of your book but having no means to find the owner of the copyright be prevented from making a copy for a friend?
What if you had no descendents? Is the copyright still valid?
Note the this is "Hurricanes that hit the U.S." rather than number and severity of Atlantic hurricanes. There a rather significant difference between the two.
We did get some financial support from a graphics card manufacturer and employed some students on the task for a summer. But corporate priorities change, or maybe they weren't happy with the rate of progress, and the manufacturer didn't make good on promises of subsequent support, which prompted the university to withdraw matching funds, which meant we really couldn't afford to work on it further....
Hence the hope that interested volunteers step up.
So do I. In fact I keep looking for people to help us develop this.... To no avail. :( Aparently the people who want this most don't have the ability to implement it, and the people who have the ability (assuming they exist) aren't interested.
If anyone wants to help, join the boinc_opt mailing list and send a message.
BTW, David is the titular director of SETI@home, but currently has no managerial duties beyond the BOINC project.
Bzzrt. Everyone who doesn't quite understand relativity gets this one wrong. Right answer, wrong reason. The acceleration isn't important. The velocity is.
Think of it this way. Suppose one twin went to a star 100 light years distant at 0.99995c. Assume he has a magic ship that doesn't accelerate. Just boom and it's moving....
To the twin that stayed behind, he see the ship take 100 years to get to the star and 100 years to get back, (i.e. 200 years round trip).
Now look at it from the perspective of the twin on the ship. He turns on the engines and suddenly he's moving at 0.99995c. He gets out his handy dandy ruler and measures the distance to his destination. As expected, it's 100*sqrt(1-v^2/c^2) or 1 light year. Also as expected the trip takes 1 year at 0.99995c. When he gets there, he turns the ship around and starts the trip back, which takes another year.
So when he gets back to earth he sees that during his two year trip, his brother has aged 200 years.
Yes, there are perfectly fine until you realize that force and momentum are artificial concepts that are required in Newton's Laws because the Newton's laws are an incomplete description of the physics involved. Even the concept of kinetic energy is artificial. Kinetic energy is just a term in the taylor series expansion of E=mc^2 around v=0.
Better yet, they could fund the war for a week.
The big difference is that on Google, my home page is #1. On Yahoo, #1 is an innodb bug report I filed in January. A page about memory mapped files several links below my home page is #3, another of assorted links I like is #11, and my home page isn't in the first 100.
Which search engine do you think generates more useful results?
Which is to say that there is essentially no data on how hazardous Apollo missions were, while the data on the safety of shuttle missions is merely sparse.
I think you should also count training/preparation deaths for each, since you can't fly the missions without the prep.
Do a little statistics and you'll see that the shuttle failure rate is (assuming random single point catastrophic failure) is somwhere between 0% and 10% with the measured rate being about 2%.
Do the same for Apollo and you'll get a catastrophic failure rate of between 0% and 22% with the measured rate being 11% (Apollo 1 and Apollo 13). By this measure, there's a good probability that the shuttle is the winner.
Lets try other measures (skiping the statistical treatment).
There's no real evidence that Apollo was any safer than the Shuttle is. There's also no real reason to believe that the next vehicle will be any safer.
Now that I've moved the (0.5 l/s) water pump and aquarium into the garage, a water cooled power supply is my next home project. I figure I'll build a sealed enclosure for the supply, fill it with transformer oil, and run a copper tube through it for heat transfer. The heatpipe solutions are neat, but I don't need any burns from touching the radiator.
At that point, the only air cooling in my system would be the SDRAM.
We're looking for someone to help us implement this very thing.
Anyone up for the challenge?
If you didn't post as an Anonymous Coward, maybe, just maybe, a qualified person would contact you.
Bzzztt! (for the registry at least). Cygwin presents the registry as it should be....
A true liberal would counter that the idea enshrined in our law that a corporation is equivalent to a person threatens our liberty. The idea currently promoted by our government that a corporation cannot be held responsible for violating the law is downright dangerous.
The punishment for violation of antitrust laws can include enforced sale of a corporations assets. You may not agree, but I'm sure you'll jump up and down supporting the idea that people can be deprived of their lives for committing a crime.
Hence nearly every astronomical paper of the last century is available online free of charge, either at the NASA astrononomical data system or through the reprint/preprint archive.
Meanwhile ad or subscription supported journals tend to charge fairly steep fees for electronic access to "their" articles.
Its tough for an author supported journal to say "We're going to charge you in order to publish this, but we're not going to let you distribute it in other ways." Hence, "pay to publish" journals foster free access to journal articles.
Thank you for your honesty.
It's bad when anything is considered to be above critisism.
Of course, now, any platform that uses a non-zero in memory value for a null pointer must hide that fact from the programmer. In other words:
Watch out. If you actually point out any problems with OS X server, you'll get modded as a troll by the Mac lobby.
Back in the old days when I was a wee lad, we used to have something that did this. I seem to recall it was called an HTTP proxy.
I've had nothing but trouble integrating MacOS X as a server into an existing linux and solaris network. There is no commonality with administration tools and technique. NIS support is an afterthought. Mount maps need to be massaged. The NFS server rejects connections randomly. The admin interface changes from OS rev to OS rev, sometimes without backward compatibility. (Gee, I upgraded, why doesn't NFS work anymore?) And to top it all, you're still limited to 32 bit addressing in your processes. Oh, and there's no RAID5 software that really works in a high availability application.
Then there are occasion random crashes under load and "can't create process-resource not available" problems...
If you are used to administering UNIX systems, do yourself a favor, get yourself a cheap multiprocessor Opteron, install your favorite Linux or *BSD distro and get some work done. If you must have a Mac, go ahead and buy one, but be sure you wipe the drive and install Linux or *BSD on it.