Comment Re:Finland will save money on napkins (Score 1) 523
You use long division for any of those things???
You use long division for any of those things???
Rsync will happily detect and copy changes without propagating whole files, yes. But only on network transfers, and it requires reading the entire file on both ends. When backing up to a local (removable) hard drive, which is what we are discussing here, it is usually faster to copy the file once than to checksum it twice, so that is what rsync does by default.
I have compared the tools, and I do know what they do. I found a hundredfold improvement in the time to backup a set of virtual machines on a linux server after switching from rsync to ZFS.
If the file server uses a file system with checksums, and those checksums are also backed up, then it's a simple matter of reading through the tape and verifying the checksums. You don't need to compare to the original files.
(The probability of a corrupted backup server accidentally creating a correct checksum can be made arbitrarily small. Usually it's something like 2^-256.)
You might want to look at using ZFS instead of rsync. I switched a while back, and it was definitely worth the initial effort of changing the file system on the server.
With rsync you can get inconsistencies because not all files are backed up at the same instant. ZFS snapshots get around this.
If you modify a large file (say a 100 GB virtual machine), rsync will re-backup the entire file. ZFS will keep track of the part that changed and only copy that.
Also if a file on one of your multiple backups is subtly corrupt, you might not notice. Or even if you do compare the copies, you might not know which one is correct. With ZFS, the entire file system is checksummed and a raid or mirror can heal itself.
If you don't live in NYC or near another major metro/subway position, and you're too fucking stupid to call a cab instead of driving home drunk, then drinking becomes very dangerous.
FTFY.
In that case I'm sure you're doing something like this:
\newif\ifdraft
\drafttrue % or \draftfalse
\ifdraft
(should we cite the fine Gabor paper here?)
\fi
The paper in question was most likely not written in LaTeX, or they would have put a percent sign in front of the comment when they first put it in.
After all each time you add a different type of specialty processor into an environment, you introduce another codebase for the application, another toolchain to learn and another set of communication / OS support issues.
That will be an issue only for the OS and library developers. To the applications developer there will be no noticeable difference. It is already the case that you need to use specialized libraries to get maximum performance on common types of tasks.
For example, if you want to use an FFT on a modern "general purpose" processor, you will get much better performance using a standard library function than you would if you wrote your own. There are so may issues with memory access patterns, core and cache utilization, etc. that you will never have time to figure out if you just want to use the FFT (rather than do research on the algorithm itself.)
If a future CPU gets a built in FFT, then the standard library will be updated, and your application will just run faster. No modification necessary.
Keeping the press away is a matter of National Security. That's how it is in every police state.
I used to hang out in a swedish photography/videography forum. Bandwidth is cheap in Sweden, so a lot of these guys were on 100+ Mbit connections and liked to keep a backup in the cloud. Whenever a new "unlimited" storage service came around they'd hop on and upload tens of terabytes of photos/videos. (None of wich could be de-duplicated, since it was all original work.)
Inevitably, the storage service would update its TOS within a year, or go bankrupt.
This is currently modded "funny", but is actually a very good solution to the problem. With water-cooling, all electrical components, except the radiator fan, can be in an air-tight enclosure. Then get an IP rated fan, or a larger, fanless radiator.
On the contrary: This is the ultimate free market. Even the politicians are for sale. All Tesla motors need to do is raise some money, buy half the legislature, and ban the sale of non-electric cars. A kickstarter campaign would probably do it.
It's interesting how it is always "Socialism vs Capitalism", with most people divided into two camps, and very few saying that one or the other migh be better depending on circumstances.
I like capitalism when there's a natural way that businesses can compete. For example, adjacent coffee shops might compete on a number of parameters like price, quality, speed of service, etc. It is possible to establish a new coffee shop in an area that already has one, if you can compete on at least one of theres parameters.
I like socialism when there is no way that cometition will work. For example ISP:s provide basically identical service, so they can only compete on price and marketing. Marketing is bullshit, so rational people will pick the one with the lowest price. This means that with one ISP in place, the other ISP:s have no incentive to build infrastructure in the same area, because competition would drive prices down to where they can't recover the initial investment. The single ISP therefore gets a monopoly. It is preferrable that the monopoly is owned by the state.
The most likely answer to this riddle is that all of the so called researchers are complicit.
Never attribute to malice what can adequately be explained by incompetence.
The researchers did make a measurement that reveals the hidden circuit. They just didn't realize it themselves. (They measured currents in order to estimate heat loss in the cables.) Details here.
The latest report on Rossi's device actually contains clear evidence that the experimental set-up has been tampered with. On page 14 it says:
"Measurements performed during the dummy run with the PCE and ammeter clamps allowed us to measure an average current, for each of the three C_1 cables, of I_1 = 19.7A, and, for each C_2 cable, a current of I_1/2 = I_2 = 9.85 A."
Here, I_1 and I_2 are the line and phase currents of a set of delta-connected resistive load inside the "reactor". The ratio between these currents should therefore be sqrt(3) (approximately 1.73). Since the measured ratio is 2, the curcuit diagram cannot correspond to reality. The reactor probably contains two separate sets of star-connected resistors instead. By feeding current to the second set out of phase with the first, like I suggested in a previous slashdot comment, the current clamps are fooled into giving a too low measurement.
This document (in Swedish) explains it all in detail.
The fact that these measurements were performed and reported also implies that the authors of the report were not part of the fraud. Rossi simply fooled them all.
If all else fails, lower your standards.