if someone has to handle those documents then how do you know they aren't a spy.
if someone has the money and determination to plant a spy for many YEARS whilst they work thier way into sensitive positions what exactly can you do about it?
but i think sin and cos aren't done in hardware on even modern processors. The difference is that the floating point operations required to do them to the accuracy required are much faster nowadays.
if you've ever progrmmed a for-loop, then you have basically integrated!;) thats numerical integration which is as you say piss easy i suspect the GP was reffering to exact integration.
the trouble with exact integration is its not always possible and when it is possible there are many different methods and it can be hard to tell which one fits your problem.
unlike differentiation where you can take any combination of functions (which you know how to differentiate individually) and press through it appllying rules to eventually get an answer.
if he is just having trouble remembering stuff in maths it probablly is that he is disinterested (not that its nessacerally easy to solve that). if he has problems accross the board with memorising stuff then thats a different matter.
i personally find it greatly helps to see a derivation even if its not a mathematically rigourous one. once you have some understanding of whats going on remembering can get much easier.
On an 80386 (with no 80387), one can still write software which essentially "solves" the problems that can only be solved by trig and calculus. This proves (as mathematicians know I'm sure) that trig and calculus are reducible to complex algebraic constructs using tests/branches, iterators, etc.
no it proves that they can be APPROXIMATED by such constructs and those approximations aren't exacly all that cheap in terms of cpu time either especially if you don't have hardware floating point math.
and errors from approximations can easilly add up especailly where approximations of integration (which is basically turning a rate of change into a value e.g. going from acceleration to velocity or velocity to position) are involved.
The idea is to simplyfy a problem as much as you can using maths and then code an implementaion that provides a sufficiantly good approximation to that simplified problem.
Yes you are right... it's more than what people will bear... but life's a bitch. I sure as hell don't run processes that talk on the Net with administrative credentials. but do you run them with the credentials to access your important data or do you go to the pains to partition your computer usage into different user accounts for different tasks?
the correct answer at an abstract mathematical level assuming all numbers given are exact values would be 10000000.00000001.
Sometimes you can take precision is implied but you have to be very carefull with doing so especially with integers after all is 100 a number given to 1 SF or a number given to 3 SF where two of the digits happened to be zero? you can't tell!
When using a calculator or computer you have to understand that the response won't be an exact answer for most functions and you need to know if the loss of precision is significant or not.
to take one possible example suppose for a simulation you decided to represent the position of your boat and something it is towing as relative to its home base using a pair of numbers.
Mathematically the reference point is arbitary and so this seems fine. However you then decide to represent those numbers as whatever floating point type is conviniant and suddenly the accuracy decreases as you move away from the reference until it eventually gets bad enough that your simulation breaks in some way.
Nievely written algorithms can lose a huge ammount of precision way beyond the obvious loss from the data type limits when converted from mathematical perfection to computer data types.
Lets take the problem of offering access to irc from your website to those who don't have a special client installed and look at the options. The reasoning here should apply to anything where realtime updating is desired not just irc char.
1: java applet This is by far the most common method and works pretty well. However unfortunately windows does not ship with a jvm as standard anymore.
2: activex Works on any windows/ie system, but doesn't really work anywhere else. However it has to be signed which puts people off. Also locks out most other operating systems/browsers.
3:.net Technically very similar to java although more windows biased, needs the.net framework installed which is not on all windows systems at this stage. Also locks out most other operating systems/browsers.
4: Refreshing works but there is some delay and the flicker can become highly annoying. The higher you make the refresh rate the worse the flicker and the higher the server load.
5: streaming into a frame Works with any browser that supports frames and incremental rendering but is pretty ugly and inflexible. Also breaks with some proxies though that can usually be worked arround by using https. The only implementation i know of (older versions of cgiirc) also requires a huge ammount of server side rescources.
6: streaming javascript. This can give far nicer results than streaming into a frame but needs javascript enabled in the browser and browser detection is probablly needed to make everything behave right. As with the one above the only implementation i know of (newer versions of cgiirc) requires a huge ammount of server side rescources.
NONE of theese options clearly beats the others in every respect.
the issue is that people run their desktop under an administrative account... which means when rogue code enters your system it has free will to do anything.
Running under a non-admin account may save some time reinstalling but unless you are prepared to split yourself into multiple users for different tasks (which is more of a pain than i suspect most users will bear) thats about all it will do.
and remember on a linux system if someone comprimises your user account its fairly easy to set you up with a local binary dir and put a fake su binary in there (which records the password and then passes it to the real su). Again there are things you can do about this but they are almost certianly more pain than most will bear.
what university course was this? CS? some kind of engineering?
i'm studying electronic systems engineering and the blunt fact is most of the students have either never programmed in anything before or have only a very primitive knowlage of VB and nothing else. The last thing they wan't to do is have to teach the complexities of input in java before they teach basic programming.
btw at my university they used a non-standard input unit to simplify input but it wasn't gui. They also used an IDE known as bluej which lets you create and manipulate objects directly letting you grasp the basic concepts of java without having to build full working apps right from the start.
i should clarify that by ebook reader technology i mean technology that lets you treat an ebook like a paper book, carry it arround read it anywhere curl up in bed with it withhout worrying too much about damaging it etc.
reading on a PC or laptop is just about acceptable but has major power and portablity issues and the current generation of pdas look pretty horrible for trying to actually read a book.
ofc if ebook reader technology became commonly availible i think its likely the publishing industry will have problems with people "sharing" copies of ebooks just as they do with music.
afaict the main thing that controls book piracy atm is its a pain to actually make a copy of a book especially if you wan't it bound properly so people only bother copying things like textbooks which are far more expensive than most mass market books for the same size of book.
i've seen "redundant array of inexpensive disks" (which i belive was the original) and "redundant array of independent disks" but never the one you mention. Care to cite a source?
surely if a governement ended up under almost total trade embargo it would have no good reason not to pass a law rendering foriegn copyrights totally unenforceable.
once that was done being unable to purchase software licenses would no longer by a problem. and if they wanted windows source i suspect they could get thier hands on it somehow loads of universities etc have people with access to it, it can't be that hard to find one thats bribeable.
How many times has IBM done this before? Are their tech support staff trained to keep a straight face while asking for the caller's address? Do most other vendors do this too? Or was this done specially because the UC laptop was high-profile?
why would they need the address? surely they could just look at thier phone logs to get something the telco could identify the call with and give that to the police.
well gas ovens are pretty simple really and at least here in britan they add a stinky substance to the gas so you can smell leaks. also obvious danger of bodiliy harm makes people take more care.
steam irons are again pretty simple and again have an obvious danger of bodily harm so again people take more care.
cars have a mandatory training and licensing programme in all civilised countries i know of.
the problem with computers is people view them like a vcr or a phone, something where they can't really do any harm through ignorance. Sadly in the days of the e-mail instant messaging online shopping etc this simply isn't the case.
You do realise that the large UK chain GAME chain used a uk.com address and had it printed on their bags at one stage?
you do realise that deciding to call your company "game" is going to involve some compromises when chosing a domain ending (tld or short second level domain). A slightly dubious ending that has some association with the uk was probablly preferable to some obscure CCTLD or mangling thier name.
they seem to have got game.net now but i bet they didn't get it easilly or cheaply.
if you used / as the only seperator you would need to find some other way of differentiating between the parts of the dns hostname and the parts of the file path on the destination server.
Slashdot Mirror
hmm sounds interesting but just how modern do the browsers have to be for this to work?
i thought it just needed a sendmail combatible interface for sending mail (/usr/bin/sendmail) which most mtas do seems to provide.
if someone has to handle those documents then how do you know they aren't a spy.
if someone has the money and determination to plant a spy for many YEARS whilst they work thier way into sensitive positions what exactly can you do about it?
360 is a nice easilly divisible number thats the only real reason. The french made an attempt to decimalise angles but it never really took off.
Most serious stuff uses radians instead because they simplify a lot of formulae (including critically the differentiation and integration of trig).
but i think sin and cos aren't done in hardware on even modern processors. The difference is that the floating point operations required to do them to the accuracy required are much faster nowadays.
if you've ever progrmmed a for-loop, then you have basically integrated! ;)
thats numerical integration which is as you say piss easy i suspect the GP was reffering to exact integration.
the trouble with exact integration is its not always possible and when it is possible there are many different methods and it can be hard to tell which one fits your problem.
unlike differentiation where you can take any combination of functions (which you know how to differentiate individually) and press through it appllying rules to eventually get an answer.
it depends
if he is just having trouble remembering stuff in maths it probablly is that he is disinterested (not that its nessacerally easy to solve that). if he has problems accross the board with memorising stuff then thats a different matter.
i personally find it greatly helps to see a derivation even if its not a mathematically rigourous one. once you have some understanding of whats going on remembering can get much easier.
On an 80386 (with no 80387), one can still write software which essentially "solves" the problems that can only be solved by trig and calculus. This proves (as mathematicians know I'm sure) that trig and calculus are reducible to complex algebraic constructs using tests/branches, iterators, etc.
no it proves that they can be APPROXIMATED by such constructs and those approximations aren't exacly all that cheap in terms of cpu time either especially if you don't have hardware floating point math.
and errors from approximations can easilly add up especailly where approximations of integration (which is basically turning a rate of change into a value e.g. going from acceleration to velocity or velocity to position) are involved.
The idea is to simplyfy a problem as much as you can using maths and then code an implementaion that provides a sufficiantly good approximation to that simplified problem.
Yes you are right... it's more than what people will bear... but life's a bitch. I sure as hell don't run processes that talk on the Net with administrative credentials.
but do you run them with the credentials to access your important data or do you go to the pains to partition your computer usage into different user accounts for different tasks?
the correct answer at an abstract mathematical level assuming all numbers given are exact values would be 10000000.00000001.
Sometimes you can take precision is implied but you have to be very carefull with doing so especially with integers after all is 100 a number given to 1 SF or a number given to 3 SF where two of the digits happened to be zero? you can't tell!
When using a calculator or computer you have to understand that the response won't be an exact answer for most functions and you need to know if the loss of precision is significant or not.
to take one possible example suppose for a simulation you decided to represent the position of your boat and something it is towing as relative to its home base using a pair of numbers.
Mathematically the reference point is arbitary and so this seems fine. However you then decide to represent those numbers as whatever floating point type is conviniant and suddenly the accuracy decreases as you move away from the reference until it eventually gets bad enough that your simulation breaks in some way.
Nievely written algorithms can lose a huge ammount of precision way beyond the obvious loss from the data type limits when converted from mathematical perfection to computer data types.
Lets take the problem of offering access to irc from your website to those who don't have a special client installed and look at the options. The reasoning here should apply to anything where realtime updating is desired not just irc char.
.net .net framework installed which is not on all windows systems at this stage. Also locks out most other operating systems/browsers.
1: java applet
This is by far the most common method and works pretty well. However unfortunately windows does not ship with a jvm as standard anymore.
2: activex
Works on any windows/ie system, but doesn't really work anywhere else. However it has to be signed which puts people off. Also locks out most other operating systems/browsers.
3:
Technically very similar to java although more windows biased, needs the
4: Refreshing
works but there is some delay and the flicker can become highly annoying. The higher you make the refresh rate the worse the flicker and the higher the server load.
5: streaming into a frame
Works with any browser that supports frames and incremental rendering but is pretty ugly and inflexible. Also breaks with some proxies though that can usually be worked arround by using https. The only implementation i know of (older versions of cgiirc) also requires a huge ammount of server side rescources.
6: streaming javascript.
This can give far nicer results than streaming into a frame but needs javascript enabled in the browser and browser detection is probablly needed to make everything behave right. As with the one above the only implementation i know of (newer versions of cgiirc) requires a huge ammount of server side rescources.
NONE of theese options clearly beats the others in every respect.
the issue is that people run their desktop under an administrative account... which means when rogue code enters your system it has free will to do anything.
Running under a non-admin account may save some time reinstalling but unless you are prepared to split yourself into multiple users for different tasks (which is more of a pain than i suspect most users will bear) thats about all it will do.
and remember on a linux system if someone comprimises your user account its fairly easy to set you up with a local binary dir and put a fake su binary in there (which records the password and then passes it to the real su). Again there are things you can do about this but they are almost certianly more pain than most will bear.
what university course was this? CS? some kind of engineering?
i'm studying electronic systems engineering and the blunt fact is most of the students have either never programmed in anything before or have only a very primitive knowlage of VB and nothing else. The last thing they wan't to do is have to teach the complexities of input in java before they teach basic programming.
btw at my university they used a non-standard input unit to simplify input but it wasn't gui. They also used an IDE known as bluej which lets you create and manipulate objects directly letting you grasp the basic concepts of java without having to build full working apps right from the start.
i should clarify that by ebook reader technology i mean technology that lets you treat an ebook like a paper book, carry it arround read it anywhere curl up in bed with it withhout worrying too much about damaging it etc.
reading on a PC or laptop is just about acceptable but has major power and portablity issues and the current generation of pdas look pretty horrible for trying to actually read a book.
ofc if ebook reader technology became commonly availible i think its likely the publishing industry will have problems with people "sharing" copies of ebooks just as they do with music.
afaict the main thing that controls book piracy atm is its a pain to actually make a copy of a book especially if you wan't it bound properly so people only bother copying things like textbooks which are far more expensive than most mass market books for the same size of book.
i've seen "redundant array of inexpensive disks" (which i belive was the original) and "redundant array of independent disks" but never the one you mention. Care to cite a source?
surely if a governement ended up under almost total trade embargo it would have no good reason not to pass a law rendering foriegn copyrights totally unenforceable.
once that was done being unable to purchase software licenses would no longer by a problem. and if they wanted windows source i suspect they could get thier hands on it somehow loads of universities etc have people with access to it, it can't be that hard to find one thats bribeable.
How many times has IBM done this before? Are their tech support staff trained to keep a straight face while asking for the caller's address? Do most other vendors do this too? Or was this done specially because the UC laptop was high-profile?
why would they need the address? surely they could just look at thier phone logs to get something the telco could identify the call with and give that to the police.
httpd.conf is a pseudo XML. I don't see how it could possibly be a more pleasant than XML
well it combines the power of a nestable structure with the conciseness of simple name=value pairs for the values themselves.
which is easier to read and edit to you?
<optionname value=shit/> (XML)
optionname=shit (apache)
well gas ovens are pretty simple really and at least here in britan they add a stinky substance to the gas so you can smell leaks. also obvious danger of bodiliy harm makes people take more care.
steam irons are again pretty simple and again have an obvious danger of bodily harm so again people take more care.
cars have a mandatory training and licensing programme in all civilised countries i know of.
the problem with computers is people view them like a vcr or a phone, something where they can't really do any harm through ignorance. Sadly in the days of the e-mail instant messaging online shopping etc this simply isn't the case.
httpd.conf is a pseudo XML. I don't see how it could possibly be a more pleasant than XML
well it combines the power of a nestable structure with the conciseness of simple name=value pairs for the values themselves.
which is easier to read and edit to you?
(XML)
optionname=shit (apache)
but relying on us army stats for ww2 sounds like it would paint a rather misleading picture given that they were absent for so much of the war.
You do realise that the large UK chain GAME chain used a uk.com address and had it printed on their bags at one stage?
you do realise that deciding to call your company "game" is going to involve some compromises when chosing a domain ending (tld or short second level domain). A slightly dubious ending that has some association with the uk was probablly preferable to some obscure CCTLD or mangling thier name.
they seem to have got game.net now but i bet they didn't get it easilly or cheaply.
if you used / as the only seperator you would need to find some other way of differentiating between the parts of the dns hostname and the parts of the file path on the destination server.
i'm pretty sure its lpg (propane iirc ) not natural gas (methane). Presumablly for safety reasons.