Actually you could implement such a licensing system. Just be a little subtle about it. How about every applicant receives a free super-charged, turbo, thingy-thing, thingy car thingy.
Read your 'right of reply' and still agree with the GP. Posting was far too biased and will be illustrated by a large response in comments.
Having read a little more of Knuth than some random interview, there is a lot more to back up his dislike of unit testing. It's a philosophy thing. You do some pen and paper work about the algorithm your working on, structuring and compartmentalising it into functional sections. Proving that it will work rather than coding some random code up and hoping that your [possible non-exhaustive] testing will reveal the 'bugs' (quoting because the answers may be correct, just using the wrong algorithm).
O(n^2) lines of communication between n individuals Is there a trace of zero? I don't communicate with myself. Although there is one guy down the hall who talks a lot when nobody is around...
I can see why Knuth doesn't like unit testing. Reading his books you see that you really need to understand your algorithm properly and identify all key sections. You need to spend a bit of time by sitting down with pen and paper and draw the flow; derive what happens at limits rather than guess or 'have faith in the gods'. By designing your algorithm properly, the unit test is kinda implied.
In Australia we have a conservative Liberal party, a liberal Labor party (which is the political arm of the union movement), a green party which doesn't support cutting greenhouse gas emissions (i.e. using nuclear power) but supports gay rights and a democrat party who kept the bastards honest by committing political suicide. All easily explained though. We live in the southern hemisphere. This affects all things you know, not just the toilets.
Plug in a new monitor on a Windows box and it just works? It may "just work" but does it work the way you need it to? Are you a Windows newbie? What about when you need to change resolutions? Turn off mirroring? Hardware acceleration? Enable stereo?
If you had *never* done anything like that before on a windows box (including not having any drivers installed) how long would it take to complete the list above? I'll bet that it would take about the same time. The method is just different.
Now that I've done it a few times, it takes less than 5 minutes to enable all necessary hardware features. Just as it does on Windows. And I can save my xorg.conf file on a SAN for future reference. To me, it's about being familiar and willing to learn something new.
I'm a linux newbie and managed to get my 30" and 23" working just fine within 15 minutes... and that included the time to install the nvidia drivers. Try going through ubuntu forums, find a howto guide, print it out and follow step by step. Couldn't be simpler. Enabling stereo was a total time of 1 minute from knowing nothing to getting it working.
Besides, High School teachers have become so retarded over the years it's amazing that graduates know anything. I don't think it is that simple. I believe that the dumbing down of all material has begun from the first year of school. There is a horrible flow on effect from there. It is the curriculum that is at fault. The public schools need to stop focusing the content towards the lower end of the bell curve to at least the middle section.
I put the numerical GPU computing in the same class as the FPGAs. They are for specific purposes, just as algorithm choice. Now you have to make the combinatory choice of algorithm and chip.
I had some researchers here that needed to perform approx. 10 million 2D fft calculations (that number growing larger by the end of year). I believed, for the long term view, to deploy this by batching it for the GPU instead of batching it on our cluster (we don't have access to those really big clusters and utilisation is already 100%). I was able to deploy the solution closer to the source of the data (to lower the internal network traffic) and the 8800GTX was doing them roughly 10x faster than the 5300 series. Ideally, I would have liked to knock up the post-processing with cuda as well, but time didn't permit. Researcher was happy with throughput.
I was just about to take a swipe at the mods for modding you to hell for what is basically an insightful comment. But then I realised that I read at -1.
The reason why I fled the Windows programming environment was because I was getting sick and tired of having to rewrite parts of my code to run on other platforms. Having to change what should be standard stuff was the final straw.
I moved to OSX and what did I find there? The same shit with Carbon and Cocoa. I absolutely refuse to program in them because they are only supported by one platform. I've since moved to Linux. One size fits all... as it should be. Thanks to Trolltech.
For once, I can actually understand one of these larger companies doing this. More of these bigger players need to send a message to stop this childish crap of platform specific programming.
Looks like the entire Engineering profession needs to be outlawed. How quickly can this be pushed through congress? Surely a "think of the children" can be slipped in there.
Fabulous link. I especially liked:
Only 180 of the world's 193 countries are included in the survey, due to an absence of reliable data from the remaining countries.
I dom't thimk that that kimd of joke is fummy at all. Shane om you, you noroms! I get paid to work amd you are prevemting that fron happemimg. Who's gomma reinbuse ny enployer? You?
Suddemly ny natrix coding got all the nore difficult!
Slashdot Mirror
Actually you could implement such a licensing system. Just be a little subtle about it. How about every applicant receives a free super-charged, turbo, thingy-thing, thingy car thingy.
Cool. It does it correctly. Positive zero plus negative zero. Damn you infinity
No, No, no ... you've got it all wrong. Once it's over they evaluate the error of it all.
See http://tech.slashdot.org/comments.pl?sid=535586&threshold=-1&commentsort=0&mode=thread&cid=23208666 to see the poster defend himself. Feel free to rip on him.
Read your 'right of reply' and still agree with the GP. Posting was far too biased and will be illustrated by a large response in comments.
Having read a little more of Knuth than some random interview, there is a lot more to back up his dislike of unit testing. It's a philosophy thing. You do some pen and paper work about the algorithm your working on, structuring and compartmentalising it into functional sections. Proving that it will work rather than coding some random code up and hoping that your [possible non-exhaustive] testing will reveal the 'bugs' (quoting because the answers may be correct, just using the wrong algorithm).
Is there a trace of zero? I don't communicate with myself.
Although there is one guy down the hall who talks a lot when nobody is around
I can see why Knuth doesn't like unit testing. Reading his books you see that you really need to understand your algorithm properly and identify all key sections. You need to spend a bit of time by sitting down with pen and paper and draw the flow; derive what happens at limits rather than guess or 'have faith in the gods'. By designing your algorithm properly, the unit test is kinda implied.
Fun sponge. Way to spoil my dreams.
Woohoo! When I click on the canvas I only get a straight line (vertical, drawing towards the top). What's so special about that?
Yes, I'll go now.
All easily explained though. We live in the southern hemisphere. This affects all things you know, not just the toilets.
Great! I might get the day off work. Perhaps a little island hoping will be in order.
Damn, now why didn't I think of that. Can't wait till I get home.
Plug in a new monitor on a Windows box and it just works? It may "just work" but does it work the way you need it to? Are you a Windows newbie? What about when you need to change resolutions? Turn off mirroring? Hardware acceleration? Enable stereo?
If you had *never* done anything like that before on a windows box (including not having any drivers installed) how long would it take to complete the list above? I'll bet that it would take about the same time. The method is just different.
Now that I've done it a few times, it takes less than 5 minutes to enable all necessary hardware features. Just as it does on Windows. And I can save my xorg.conf file on a SAN for future reference. To me, it's about being familiar and willing to learn something new.
I'm a linux newbie and managed to get my 30" and 23" working just fine within 15 minutes ... and that included the time to install the nvidia drivers. Try going through ubuntu forums, find a howto guide, print it out and follow step by step. Couldn't be simpler. Enabling stereo was a total time of 1 minute from knowing nothing to getting it working.
I don't think it is that simple. I believe that the dumbing down of all material has begun from the first year of school. There is a horrible flow on effect from there. It is the curriculum that is at fault. The public schools need to stop focusing the content towards the lower end of the bell curve to at least the middle section.
Never mind an interest rate attached to the loan as well because when the university shook me by the ankles only a few coins fell out.
I put the numerical GPU computing in the same class as the FPGAs. They are for specific purposes, just as algorithm choice. Now you have to make the combinatory choice of algorithm and chip.
I had some researchers here that needed to perform approx. 10 million 2D fft calculations (that number growing larger by the end of year). I believed, for the long term view, to deploy this by batching it for the GPU instead of batching it on our cluster (we don't have access to those really big clusters and utilisation is already 100%). I was able to deploy the solution closer to the source of the data (to lower the internal network traffic) and the 8800GTX was doing them roughly 10x faster than the 5300 series. Ideally, I would have liked to knock up the post-processing with cuda as well, but time didn't permit. Researcher was happy with throughput.
I've been compiling and running cuda fine on my Fedora 8 x86_64. 8800GTX at 182GFlops for n-body.
Does this mean I don't have to wear deodorant anymore?
Oh no! Does that mean the uid is going to be implemented as a log scale now?
I was just about to take a swipe at the mods for modding you to hell for what is basically an insightful comment. But then I realised that I read at -1.
The reason why I fled the Windows programming environment was because I was getting sick and tired of having to rewrite parts of my code to run on other platforms. Having to change what should be standard stuff was the final straw.
... as it should be. Thanks to Trolltech.
I moved to OSX and what did I find there? The same shit with Carbon and Cocoa. I absolutely refuse to program in them because they are only supported by one platform. I've since moved to Linux. One size fits all
For once, I can actually understand one of these larger companies doing this. More of these bigger players need to send a message to stop this childish crap of platform specific programming.
Looks like the entire Engineering profession needs to be outlawed. How quickly can this be pushed through congress? Surely a "think of the children" can be slipped in there.
I especially liked: Only 180 of the world's 193 countries are included in the survey, due to an absence of reliable data from the remaining countries.
I dom't thimk that that kimd of joke is fummy at all. Shane om you, you noroms! I get paid to work amd you are prevemting that fron happemimg. Who's gomma reinbuse ny enployer? You?
Suddemly ny natrix coding got all the nore difficult!