It gets worse. Timing info can be made pretty accurate across one sphere but doesn't work for triangulation because the phone talks to one base station only.
So if you really want to find where it is, three or four base stations have to cooperate and intentionally attenuate themselves so that the phone thinks the other is stronger and initiates a hand-over. (That's how it works with GSM phones, anyway.)
Assuming that the phone even sees three base stations, which is highly unlikely if you're buried under a couple thousand tons of steel and rubble.
Rain? A heavy cloud cover? A bunch of low-tech guerilla fighters suddenly appear from the mist, lob a few grenades or low-tech shells at your base, vanish again, and your super-gun hasn't even booted up its internal computer. Meanwhile, all your high-tech spying equipment is of no use whatsoever because it can't see anything. And no, infrared isn't going to help you here.
Guess what the weather'll be like in Afghanistan in the next few months.
Intelligence is more than technology, and wars aren't decided just by having superior hardware. Didn't somebody learn anything from Vietnam?
The human body does cool by IR radiation, as does any other object. The question is, how fast is that (answer: "In vacuum, less than it generates"). Plus, ordinarily the environment radiates back.
It also cools by simple heat exchange with its environment, which is not a problem in vacuum 'cause there's no environment to exchange heat with there. On Mars, which does have a somewhat hostile atmosphere, this is a problem.
So they'll probably deviate from the standard. So what? They did it with SMB, and Samba still rocks.
However,.NET is no longer solely about bullying consumers to install the latest Explorer instead of Netscape, and neither is it about some Javascript trifles which ideally should be turnoffable anyway. This is B2B stuff, and M$ is trying to convince people to bet their business on it.
Thus, if Company A installs the latest M$ code (you know, the one with the subtle incompatibility), and suddenly they can no longer talk to their business partners, M$ will have major legal problem as soon as somebody figures out exactly what the true reason for the problem is.
What irks me much more than this is, however, that the only reason M$ blows a shitload of money at C# is to piss of Sun. Which is one additional reason to never buy any M$ product.
(.NET might actually be a reasonably good idea, if the primary reason for throwing several more shitloads of money at it was something besides more market share.)
Re:Torched SUV Dealership
on
Eco-Terrorism
·
· Score: 1
Terrorists don't care about their goals. Never did, never will.
Another fine example of this principle are the "pro-life" terrorists who don't hesitate to kill, provided that the victims are old enough.
That being said, SUVs combine the worst aspects of any other type of car. IMHO there's absolutely no reason to buy one. ("My neighbor thinks about bying one in my range of hearing" doesn't count.)
They want to do business in Germany. German law doesn't like racist stuff at all, and in light of both WWII and the right-wing idiots currently playing Hitler I don't really have a problem with this policy.
Censorship is bad, don't take me wrong, but free speech is not exactly the most important basic human right out there. It needs to be balanced with others, like for instance the right to life, dammit, and that right isn't furthered by the idiots who stand at the next street corner and shout "kill the f*cking foreigners".
Sorry, but I'd rather protest restrictions like ebay's blocking of erotica. That stuff at least doesn't promote killing.
That's rather stupid actually, because it is trivial to tell Google not to display the cache.
On the other hand, the search engine people will not be amused when the site owner complains about some Googlebot visiting their blocked-for-robots page.
On the third had, this is the web, it's my computer that's browsing, and my software. If you can't deal with my browsing your text (or maybe I'm more interested in the pretty pictures...) and not even loading the ads which Nielsen says users have blind spots for by now anyway, well, That's Not My Problem.
It works that way in Germany too. However, the money doesn't go to the authors but to the Big Industry Association, who then doles it out to the publishers. Whether any author sees any of that money is a good question which I'd hesitate to bet on a positive answer for.
One very annoying practical effect of this is that scanner software for the German market intentionally slows down the scan so that they don't have to pay the fee to the BIA, which only applies to copiers-and-workalikes if they're faster than some arbitrary X. Or that video cameras intentionally get their input wires lobotomized so that the importer doesn't have to pay for something-that-works-like-a-VCR.
Error (187/4): The character '&' must be written as '&'
which is not exactly/.'s fault. The others, of course, are.
Whoever started the "play games with the parser so that as many HTML errors as possible get interpreted correctly" race ought to be fired. Anybody got a time machine??
Anyway, almost no web pages at all have correct HTML. We should mount a "complain about bad HTML" campaign. I'll start writing a few appropriate letters to google.com.:-/
Re:That's why you need the verify stage
on
From Paper To PDF?
·
· Score: 1
You could also use an OCR engine as one of the inputs.
There are a bunch of mistakes that are fairly common for all OCR enginess -- messing up 'm' as 'rn' being the most common example. There are also a bunch of mistakes fairly common among inexperienced typists.
The good part is that the two sets don't overlap much (if at all), thus you catch most errors.
Does the converter understand variable-bitrate MP3s?
Come to think of it, is Vorbis itself able to encode in variable bitrate? You do need a rather good psychoacoustic model for this to work, but my current experience with LAME has been superb -- you basically get 200+-kbps quality with half the file size.
Yeah, the 16kHz cut is _very_ crisp, especially when you don't turn off the filters.
LAME, for instance, has a whole bunch of filter options, among which -k probably is the most important for such a test (it turns off all filtering).
IMHO, those tests don't consider _anything_ beyond raw bit rate. Has nobody heard about variable bit rates?
Personally, I can't hear above 16kHz any more, so that's not a problem...:-(
In fact, I routinely encode my MP3s with LAME's VBR at what ends up being roughly 140 bps overall. I haven't found any encoding scheme out there that sounds better (for me).
Another factor might be the different psychoacoustic model used by LAME (which is used to figure out which sound components can be safely thrown away). I'm not qualified to say it's better than the others, but it seems to fit what I _hear_ better.
or pass a bound method "m=o.callbackFunction", and later call "m(args)".
That's exactly what a closure is -- you use a function "m" which is dynamically bound to some kind of environment. Whether that environment happens to be an object, or just some variables which have been conveniently defined in the code lines surrounding your function doesn't really matter.
Perl just is damn convenient in that it lets you use both concepts equally easily. With Python you need to go the object-ish way when your program's logic becomes sufficiently complex.
NB, the Perl way to code up the above m is
$m = sub { $obj->method(@_); }
...
$m->(args...)
which arguably doesn't look as nice as the Python way, but (IMHO) that's a feature because I see better what's really happening.
That depends on what you're doing. If you just use Apache::Registry or similar stuff, the modules would be loaded in the child processes, which eats memory like crazy.
The right way to do it is to preload all the modules you need in your PerlStartup code.
I would _much_rather_ spend my money on a semi-open-source database which happens to do most of what I need anyway, than on a huge beast with a kitchen-sink feature list that rivals M$ Word, 100% closed source, whose license even prohibits publishing benchmark results unless you get them approved by Oracle! (Lots of database comparison pages state this as the reason why they don't incude Oracle.)
NB: Keeping a connection pool is a kludge. Why would I want to use a database that's so bloated even the connection setup takes half a second? A good database should let me connect, do a simple quere, and disconnect well over 100 times per second. What is Oracle _doing_ during that time?
The MySQL people have said exactly the same sort of things about the PostgreSQL people. So please stop the name-calling and the quotes around "test", it's not going to get you anywhere.
That being said, the standard MySQL benchmark _still_ is 30 times faster for MySQL 3.23 than on PostgreSQL 7.0 (with fsync turned off, _and_ nonstandard speed-up PostgreSQL features like VACUUM enabled, I might add). The main reason seems to be some sort of failure to use the index in the SELECT and UPDATE test loops on the part of PostgreSQL.
The benchmark, for the curious, works like this:
First it creates a table with an index:
create table bench1 (id int NOT NULL,id2 int NOT NULL,id3 int NOT NULL,dummy1 char(30)); create unique index bench1_index_ on bench1 using btree (id,id2); create index bench1_index_1 on bench1 using btree (id3);
Then it fills the table with 300.000 entries with unique id values.
Then, it issues a query like this:
update bench1 set dummy1='updated' where id=1747
which causes the backend to do one thousandread() calls. For each query.
No wonder it's slow. An EXPLAIN query states that it's using the index, though. I have no clue what happens here. I've sent this to the pgsql-general mailing list and have just reposted it to -hackers.
Oh yes, the benchmark also revealed that CREATE TABLE in PostgreSQL 7.0 leaks about 2k of memory.
Re:Record Companies Settle Antitrust Suit
on
An MP3 Update
·
· Score: 1
Ouch! I guess I'll have to take a hard look at my cookies file, to see who else stores my password in the cookie.
Hey,/., there is no need at all to store my password in the cookie. A random number, stored in my user record, will work just as well, and (even better)/. can change it periodically -- thus, any replay attacks stop working after a day or so.
IMHO, software slowness isn't caused by not using assembly language -- most programs just don't
have
any small tight inner loops. It's caused by bad design, horribly inefficient data structures, feeping creaturism.
In the Linux kernel, for instance, people routinely align their data structures by cache line. That's the level of assembly programming I feel comfortable with. But I wouldn't trust a kernel with one of the core loops (the scan for free memory, for instance) with anything. Too many mistakes have shown up in far simpler assembly-language statements.
Besides, the real tight inner loops (stuff like memcpy() and friends) usually are already inlined behind your back, by the compiler and libc, in any decent C development environment.
But when the people who created the stuff aren't around any more, I frankly fail to have any moral problems with copying the stuff. Nobody's giving me any free rides in this world, and I fail why I should be giving one to any large record company.
Relying on those before us? You betcha!
on
Laptop Exams?
·
· Score: 2
Our whole 40000-whatever year history is about relying on what others have done before us. The whole of human culture and technology happened because of it. We lose something when old skills and tools are supplanted by new ones, sure, but we also gain a lot from it.
The use of books came with the loss of oral history. The use of pocket calculators came with the loss of the slide rule.
As long as you know the basics, you're not screwed at all. The reason we do long multiplication (or learn poems) in school, for instance, isn't that we need the skill, these days, but to understand the principles so that (a) you get the understanding and (b) you can fall back on it if the calculator's batteries give out.
You can do without all the fancy Internet searches or computers or calculators or books. It would take a lot longer, though. In fact it would take so much longer that you wouldn't be able to come up with something new in the field in the first place.
Keys tied to computers? Indivviduals? Try both!
on
SSH v. SRP
·
· Score: 1
the key is (effectively) tied to one or more computers rather being tied to the individual,
You can set up ssh to have per-host keys and shared user ID space. Sure -- just like rlogin with encryption.
But you can use per-user keys just as easily. It's all a matter of choice , which is a Good Thing.
Since SRP isn't about encryption anyway (their website says so), but "only" authentication, then why would I want to use it? If somebody can watch me type and they see my screen output, they don't need a login on my system anyway.
using Linux instead of Windows, not because it's the OS I want to use, but because I can get drivers for proprietary hardware for it when I can't get enough information to write that driver for my preferred OS.
Hold on here. It doesn't matter if the hardware is proprietary. Either it's documented or it isn't. If it is, you can get the docs too. If it's not, somebody reverse-engineered something, and you can do it too.
How is the Linux monopoly going to be a change from the Windows monopoly?
The words "Open Source" do mean something, you know.
The world's most portable OS: http://www.netbsd.org.
Or BSDI, or FreeBSD, or OpenBSD, or...
Linux doesn't have that kind of nonsense, and IMHO that's a major factor why everybody's talking about Linux and the BSDs are... let's say, somewhat less popular.
Slashdot Mirror
It gets worse. Timing info can be made pretty accurate across one sphere but doesn't work for triangulation because the phone talks to one base station only.
So if you really want to find where it is, three or four base stations have to cooperate and intentionally attenuate themselves so that the phone thinks the other is stronger and initiates a hand-over. (That's how it works with GSM phones, anyway.)
Assuming that the phone even sees three base stations, which is highly unlikely if you're buried under a couple thousand tons of steel and rubble.
Guess what the weather'll be like in Afghanistan in the next few months.
Intelligence is more than technology, and wars aren't decided just by having superior hardware. Didn't somebody learn anything from Vietnam?
It also cools by simple heat exchange with its environment, which is not a problem in vacuum 'cause there's no environment to exchange heat with there. On Mars, which does have a somewhat hostile atmosphere, this is a problem.
However, .NET is no longer solely about bullying consumers to install the latest Explorer instead of Netscape, and neither is it about some Javascript trifles which ideally should be turnoffable anyway. This is B2B stuff, and M$ is trying to convince people to bet their business on it.
Thus, if Company A installs the latest M$ code (you know, the one with the subtle incompatibility), and suddenly they can no longer talk to their business partners, M$ will have major legal problem as soon as somebody figures out exactly what the true reason for the problem is.
What irks me much more than this is, however, that the only reason M$ blows a shitload of money at C# is to piss of Sun. Which is one additional reason to never buy any M$ product.
(.NET might actually be a reasonably good idea, if the primary reason for throwing several more shitloads of money at it was something besides more market share.)
Another fine example of this principle are the "pro-life" terrorists who don't hesitate to kill, provided that the victims are old enough.
That being said, SUVs combine the worst aspects of any other type of car. IMHO there's absolutely no reason to buy one. ("My neighbor thinks about bying one in my range of hearing" doesn't count.)
Censorship is bad, don't take me wrong, but free speech is not exactly the most important basic human right out there. It needs to be balanced with others, like for instance the right to life, dammit, and that right isn't furthered by the idiots who stand at the next street corner and shout "kill the f*cking foreigners".
Sorry, but I'd rather protest restrictions like ebay's blocking of erotica. That stuff at least doesn't promote killing.
On the other hand, the search engine people will not be amused when the site owner complains about some Googlebot visiting their blocked-for-robots page.
On the third had, this is the web, it's my computer that's browsing, and my software. If you can't deal with my browsing your text (or maybe I'm more interested in the pretty pictures...) and not even loading the ads which Nielsen says users have blind spots for by now anyway, well, That's Not My Problem.
One very annoying practical effect of this is that scanner software for the German market intentionally slows down the scan so that they don't have to pay the fee to the BIA, which only applies to copiers-and-workalikes if they're faster than some arbitrary X. Or that video cameras intentionally get their input wires lobotomized so that the importer doesn't have to pay for something-that-works-like-a-VCR.
Error (187/4): The character '&' must be written as '&'
which is not exactly /.'s fault. The others, of course, are.
Whoever started the "play games with the parser so that as many HTML errors as possible get interpreted correctly" race ought to be fired. Anybody got a time machine?? Anyway, almost no web pages at all have correct HTML. We should mount a "complain about bad HTML" campaign. I'll start writing a few appropriate letters to google.com. :-/
There are a bunch of mistakes that are fairly common for all OCR enginess -- messing up 'm' as 'rn' being the most common example. There are also a bunch of mistakes fairly common among inexperienced typists.
The good part is that the two sets don't overlap much (if at all), thus you catch most errors.
Does the converter understand variable-bitrate MP3s?
Come to think of it, is Vorbis itself able to encode in variable bitrate? You do need a rather good psychoacoustic model for this to work, but my current experience with LAME has been superb -- you basically get 200+-kbps quality with half the file size.
LAME, for instance, has a whole bunch of filter options, among which -k probably is the most important for such a test (it turns off all filtering).
IMHO, those tests don't consider _anything_ beyond raw bit rate. Has nobody heard about variable bit rates?
Personally, I can't hear above 16kHz any more, so that's not a problem... :-(
In fact, I routinely encode my MP3s with LAME's VBR at what ends up being roughly 140 bps overall. I haven't found any encoding scheme out there that sounds better (for me).
Another factor might be the different psychoacoustic model used by LAME (which is used to figure out which sound components can be safely thrown away). I'm not qualified to say it's better than the others, but it seems to fit what I _hear_ better.
That's exactly what a closure is -- you use a function "m" which is dynamically bound to some kind of environment. Whether that environment happens to be an object, or just some variables which have been conveniently defined in the code lines surrounding your function doesn't really matter.
Perl just is damn convenient in that it lets you use both concepts equally easily. With Python you need to go the object-ish way when your program's logic becomes sufficiently complex.
NB, the Perl way to code up the above m is
$m = sub { $obj->method(@_); }
...
$m->(args...)
which arguably doesn't look as nice as the Python way, but (IMHO) that's a feature because I see better what's really happening.
How about interconnectability between SIP and H.323?
...except for these unimportant differences:
[ 250 lines of "diff -u" output omitted ]
Don't asusme that the kernels in "testing" are related to anything. The real kernel is in subdirectory v2.4.
The right way to do it is to preload all the modules you need in your PerlStartup code.
NB: Keeping a connection pool is a kludge. Why would I want to use a database that's so bloated even the connection setup takes half a second? A good database should let me connect, do a simple quere, and disconnect well over 100 times per second. What is Oracle _doing_ during that time?
That being said, the standard MySQL benchmark _still_ is 30 times faster for MySQL 3.23 than on PostgreSQL 7.0 (with fsync turned off, _and_ nonstandard speed-up PostgreSQL features like VACUUM enabled, I might add). The main reason seems to be some sort of failure to use the index in the SELECT and UPDATE test loops on the part of PostgreSQL.
The benchmark, for the curious, works like this:
First it creates a table with an index:
Then it fills the table with 300.000 entries with unique id values.Then, it issues a query like this:
which causes the backend to do one thousand read() calls. For each query.No wonder it's slow. An EXPLAIN query states that it's using the index, though. I have no clue what happens here. I've sent this to the pgsql-general mailing list and have just reposted it to -hackers.
Oh yes, the benchmark also revealed that CREATE TABLE in PostgreSQL 7.0 leaks about 2k of memory.
Of course, it's here.
Hey, /., there is no need at all to store my password in the cookie. A random number, stored in my user record, will work just as well, and (even better) /. can change it periodically -- thus, any replay attacks stop working after a day or so.
Please fix that. Now. Thank you.
- have
any small tight inner loops. It's caused by bad design, horribly inefficient data structures, feeping creaturism.In the Linux kernel, for instance, people routinely align their data structures by cache line. That's the level of assembly programming I feel comfortable with. But I wouldn't trust a kernel with one of the core loops (the scan for free memory, for instance) with anything. Too many mistakes have shown up in far simpler assembly-language statements.
Besides, the real tight inner loops (stuff like memcpy() and friends) usually are already inlined behind your back, by the compiler and libc, in any decent C development environment.
But when the people who created the stuff aren't around any more, I frankly fail to have any moral problems with copying the stuff. Nobody's giving me any free rides in this world, and I fail why I should be giving one to any large record company.
The use of books came with the loss of oral history. The use of pocket calculators came with the loss of the slide rule.
As long as you know the basics, you're not screwed at all. The reason we do long multiplication (or learn poems) in school, for instance, isn't that we need the skill, these days, but to understand the principles so that (a) you get the understanding and (b) you can fall back on it if the calculator's batteries give out.
You can do without all the fancy Internet searches or computers or calculators or books. It would take a lot longer, though. In fact it would take so much longer that you wouldn't be able to come up with something new in the field in the first place.
You can set up ssh to have per-host keys and shared user ID space. Sure -- just like rlogin with encryption.
But you can use per-user keys just as easily. It's all a matter of choice , which is a Good Thing.
Since SRP isn't about encryption anyway (their website says so), but "only" authentication, then why would I want to use it? If somebody can watch me type and they see my screen output, they don't need a login on my system anyway.
Hold on here. It doesn't matter if the hardware is proprietary. Either it's documented or it isn't. If it is, you can get the docs too. If it's not, somebody reverse-engineered something, and you can do it too.
How is the Linux monopoly going to be a change from the Windows monopoly?
The words "Open Source" do mean something, you know.
The world's most portable OS: http://www.netbsd.org.
Or BSDI, or FreeBSD, or OpenBSD, or ...
Linux doesn't have that kind of nonsense, and IMHO that's a major factor why everybody's talking about Linux and the BSDs are ... let's say, somewhat less popular.