There is no conflict between due process and being active in prevention. You simply raise the penalty if they plead innocent and are later found guilty. Something like an honor court system.
If you knew Pascal, Fortran and Cobol I'm sure you understood DOS no problem. You understood the sequential databases that were popular in the 80s from Cobol. You understood the formulas in Lotus/QuatroPro/Excel from Fortran.
Higher Order Perl is a great way to wet the appetite for FP and learn a few of the major ideas. But it seems like your appetite is already wet you have been just trying books / lessons that are too hard. An easy Haskell book is Thompson's and you can get the first edition cheap. Thompson will work you through the basic syntax enough to do one of the good books.
There are actually pretty good drag and drop website creators. But I don't disagree with your basic point. Alice is by no means a full featured language nor does it pretend to be. It teaches very basic concepts and then becomes totally worthless. But the question is about what is a good language to learn programming not a good language for programming.
Actually quite frequently they can be made parallel. It often turns out that the previous case is needed but not the initial. That is I can generate case 200 as a simple function on the value 199. Then generate 201 using the same function... right through to 399. Then all that has to happen is I get the value for 200 and it binds 399.
Higher Order Perl is a good book and it teaches some nice techniques. But the book is about getting 5% of the FP benefits in Perl. You are missing a ton and working too hard to get the benefits. What is a neat technique in Perl becomes second nature in FP languages and you get many more features of the type that are in that book.
The key to avoiding those sorts of mistakes is to write your code in terms of a single modifier and a fold. Those examples are designed to teach syntax and basic ideas performance is an intermediate topic in FP.
Can anyone name a single instance where a functional paradigm has yielded the best measured performance on a parallel computing problem?
Absolutely. Google's ability to analyze large blocks of the internet, decompose and index is being done using FP. They are way above anyone else in this matter.
Another easy example is YAWS which was written by one guy in about 6 weeks. Outperformed Apache for speed / concurrent connections by a factor of 20.
Monadic methods (for things like mutable arrays) don't allow you to be parallelism safely. You have to pick here, you can have parallelism via. purity or ease of conversion via. mutability.
I hope you are right and this isn't like China's system. Having a well documented total failure will be good for the rest of the free world as well as Australia.
I don't know the numbers for Australia but porn and sex chat represents a huge percentage of America's paid internet usage and has for a long time. What do people do on Second Life?
I don't think the situation is symmetrical at all.
I'd assume if you are going start ISP level net filtering and not just use some netnanny type thing, you block all encrypted communication. You probably also switch from black lists to white lists and some sort of direct connection isn't allowed in TCP. You assume in the system that the people on both sides are engaged in circumvention and make that impossible.
As anyone who has ever had to deal with the police knows the only crimes they really take seriously and try and stop are speeding and murder. The police do very little about property crime.
The obvious solution is in areas where there is lots of copper theft to setup sting operations and good quality surveillance of existing sites. Follow it up with what everyone is recommending... tighter rules on scrap metal foundries and again use stings and audits to verify the rules are being followed.
But none, or little of this is going to happen. The police are far to busy aggressively addressing speeding
The thing that's better about Linux with hardware is if your hardware is having problems the ability to interact at a low level with the hardware is a lifesaver.
What does this even mean? The average user will never figure out how to edit some file in/etc to change module loading options. It is not going to happen. Sorry. Again, head in the clouds, feet in the... well, I don't really want to know.
You may want to work on less insulting language when you don't know what you are talking about./etc doesn't control low level hardware interfaces, that contains some configuration information for higher level interfaces. You handle low level calls directly from the command line, by loading kernel modules with particular arguments, and/or via/dev and/proc.
As for you comments regarding windows you haven't stated any evidence you have merely argued by assertion. As for specific examples of drivers that's not the end of it. Windows has all sorts of code to catch hardware problems for example the older CD-Roms which loaded their virtual SCSI buses improperly. On Linux this required a manual device load while Windows had custom code for each of these buses.
I don't know the specifics of some scanner but my guess is that either OS handles it fine and you just have to force the windows drivers or use a small wrapper. I think your point may have been open source supports their drivers for longer in which case I agree, but that's not the point being addressed.
You have a good point about there already being a system. But that system was expensive to create. The whole theory of wifi is that it is cheap. If it isn't then the basis for the theory is wrong.
What you are describing is more like nationalization of a major utility.
Slashdot Mirror
There is no conflict between due process and being active in prevention. You simply raise the penalty if they plead innocent and are later found guilty. Something like an honor court system.
I agree she got to feel this from the other side.
If you knew Pascal, Fortran and Cobol I'm sure you understood DOS no problem. You understood the sequential databases that were popular in the 80s from Cobol. You understood the formulas in Lotus/QuatroPro/Excel from Fortran.
How exactly did they fail to prepare you?
Higher Order Perl is a great way to wet the appetite for FP and learn a few of the major ideas. But it seems like your appetite is already wet you have been just trying books / lessons that are too hard. An easy Haskell book is Thompson's and you can get the first edition cheap. Thompson will work you through the basic syntax enough to do one of the good books.
There are actually pretty good drag and drop website creators. But I don't disagree with your basic point. Alice is by no means a full featured language nor does it pretend to be. It teaches very basic concepts and then becomes totally worthless. But the question is about what is a good language to learn programming not a good language for programming.
I'm looking forward to version 3 as well.
I (and others) wrote a good wikipedia page on this topic
http://en.wikipedia.org/wiki/Educational_programming_language . I'd look at this list
I personally love and can recommend Alice http://www.alice.org/ and had a great deal of success with my daughter with this.
Could you give an example with code?
A great book which covers fold is: http://www.amazon.com/Introduction-Functional-Programming-using-Haskell/dp/0134843460
second everything in this post
Actually quite frequently they can be made parallel. It often turns out that the previous case is needed but not the initial. That is I can generate case 200 as a simple function on the value 199. Then generate 201 using the same function... right through to 399. Then all that has to happen is I get the value for 200 and it binds 399.
Higher Order Perl is a good book and it teaches some nice techniques. But the book is about getting 5% of the FP benefits in Perl. You are missing a ton and working too hard to get the benefits. What is a neat technique in Perl becomes second nature in FP languages and you get many more features of the type that are in that book.
The key to avoiding those sorts of mistakes is to write your code in terms of a single modifier and a fold. Those examples are designed to teach syntax and basic ideas performance is an intermediate topic in FP.
Can anyone name a single instance where a functional paradigm has yielded the best measured performance on a parallel computing problem?
Absolutely. Google's ability to analyze large blocks of the internet, decompose and index is being done using FP. They are way above anyone else in this matter.
Another easy example is YAWS which was written by one guy in about 6 weeks. Outperformed Apache for speed / concurrent connections by a factor of 20.
Haskell. Because it is so different you will be forced to break the habits you need to break.
Monadic methods (for things like mutable arrays) don't allow you to be parallelism safely. You have to pick here, you can have parallelism via. purity or ease of conversion via. mutability.
It isn't recursion vs. iteration but rather pure vs. environmental (i.e. mutable variables) that make parallelism safe.
Not in a functional language, particularly a pure one. This just ends up looking like map/fold or cata and ana morphism calls.
I hope you are right and this isn't like China's system. Having a well documented total failure will be good for the rest of the free world as well as Australia.
You deserve the +3, pro-euthanasia and pro-anorexia this is pure political censorship.
I don't know the numbers for Australia but porn and sex chat represents a huge percentage of America's paid internet usage and has for a long time. What do people do on Second Life?
I don't think the situation is symmetrical at all.
I'd assume if you are going start ISP level net filtering and not just use some netnanny type thing, you block all encrypted communication. You probably also switch from black lists to white lists and some sort of direct connection isn't allowed in TCP. You assume in the system that the people on both sides are engaged in circumvention and make that impossible.
As anyone who has ever had to deal with the police knows the only crimes they really take seriously and try and stop are speeding and murder. The police do very little about property crime.
The obvious solution is in areas where there is lots of copper theft to setup sting operations and good quality surveillance of existing sites. Follow it up with what everyone is recommending... tighter rules on scrap metal foundries and again use stings and audits to verify the rules are being followed.
But none, or little of this is going to happen. The police are far to busy aggressively addressing speeding
The thing that's better about Linux with hardware is if your hardware is having problems the ability to interact at a low level with the hardware is a lifesaver.
What does this even mean? The average user will never figure out how to edit some file in /etc to change module loading options. It is not going to happen. Sorry. Again, head in the clouds, feet in the... well, I don't really want to know.
You may want to work on less insulting language when you don't know what you are talking about. /etc doesn't control low level hardware interfaces, that contains some configuration information for higher level interfaces. You handle low level calls directly from the command line, by loading kernel modules with particular arguments, and/or via /dev and /proc.
As for you comments regarding windows you haven't stated any evidence you have merely argued by assertion. As for specific examples of drivers that's not the end of it. Windows has all sorts of code to catch hardware problems for example the older CD-Roms which loaded their virtual SCSI buses improperly. On Linux this required a manual device load while Windows had custom code for each of these buses.
I don't know the specifics of some scanner but my guess is that either OS handles it fine and you just have to force the windows drivers or use a small wrapper. I think your point may have been open source supports their drivers for longer in which case I agree, but that's not the point being addressed.
You have a good point about there already being a system. But that system was expensive to create. The whole theory of wifi is that it is cheap. If it isn't then the basis for the theory is wrong.
What you are describing is more like nationalization of a major utility.