Technically, that is what they do. Basically, they just say that they do not need classical RDBMS to do their job. I agree with them that RDBMS makes poor implementation of big dictionnaries.:)
For example, if I want to fetch all customers and all their bids in one query I have to use inner join. And that results in LARGE number of rows (Cartesian product of customers and their bids).
I am not sure I get your point. If you do an inner join it means you want all the tuple < player,bid > that makes sense. If there is a lot of them, well, there is a lot of them, there is nothing to do about it. If you complain about each player being repeted on several bid (since they bid more than once). It should not be a problem, as long as you stay in the RBMS, this should not incur any overhead. When you read them, you can just compress them on the fly.
If you really want to remove those "extra" player values, why do you want to have a single query ? You can just make a query for each player.
Also, I'd like to see stuff which is not easily expressed in relational algebra, like running sums or grouping on a computed field.
technically, they cannot be expressed in relationnal algebra, you have to add non algebraic operator to do that. SUM and GROUP BY in SQL are not part of relationnal algrebra. The problem with those operators is that it is difficult to do any optimization on them. Howver, the user may still want to have them. I would also be interested in a language that can express such things efficiently
A couple of years ago (2003), a cryptographic system was said to be secured if it requires more than 2^80 computation. I do not know what is the current standard and I have no clue how to find it.
However 2^110 is still way to large for us at the moment.
To give an estimation. Supose you have one million processors clocked at 10Ghz (which nobody have nowadays), you can do 10^6 * 10^10 = 10^16 ~= 2^48 computations per second. To crack AES and using this machine you'll need 2^110 / 2^48 = 2^62 seconds to do so which is approximatively 150 billions year.
The attack is of theoretical interest mainly but should tell people not to use AES with a cryptographical key smaller than 256.
Is this information really pertinent ? It is the summary of the most requested website through netcraft as far as I understand. if you look at the highest uptime recorded by netcraft at http://uptime.netcraft.com/up/today/top.avg.html then you only see IIS servers. High uptime rhymes with stability. There is certainly an explanation (security update may requires reboot or stuff like that). But it makes me wonder if the stat is relevant. Someone ? Any clue ?
mmm, I am not sure it is only a matter of number of machine you can infected. But a cost/efficience analyze such as in "one hour of work, how many machine could I infect".
My point is that if there was twice much Linux than Windows but requires ten times more effort to infect them, you'll probably stick with infecting Windows since it is more efficient.
Of course they are not different in New York. I just wanted to say that a large cluster of turbines would have a more significant local impact than the one powering a small town.
My question was more about bird migration or any long trip. But there certainly are years with less wind so it won't matter.
I'd just like people to think about it before deploying such large scale new energy. From all the replies I read, it seems some people did, which is nice.
Oh I read a nice study on the impact of high voltage lines on the health of people leaving below.
The study showed a correlation between the presence of these lines and strange health diseases.... even when the lines where powered down... Nocebo effect is the worst thing to fight.
As for birds, modern prop turbines move slowly and kill fewer birds than the same number of high rise buildings with mirrored windows.
I believe GP was not concerned of birds hitting the turbines but the impact of locale wind slowing on birds. In other word, what is the local impact of a cluster of turbine that could power a city as New York and should we care about it ?
Yeah, but we do not have an infinite coal ressource, but solar resource seems infinite (Of course it is not, but we will have worse problem to deal with at this time).
I heard of the ayle project ( http://ayle.fr/en/index.html ) which is not limited to the iphone (I do not even know if it works on it). It is kind of an hack n Slash where you have to physically move to move in the game. It is "advertised" as an hiking game.
I needed to learn fortran in january. I got quickly some slides of a physic teacher online and learned the syntax quite easily from them. I am not sure a book is needed here.
Of course, I have probably missed some subtle points, but I got enough of it to understand/debug/optimize some piece of code.
It worked for years under debian for me.
You just have to configure two serverlayout in/etc/X11/xorg.conf and give the right parameter to X to select the server layout at boot time.
It has to be done manually and it would of course be nice to have a GUI to configure it properly.
As the quality level of the platform continues to hover between poor to mediocre, I think more companies are going to find professional commercial systems more cost effective.
In practice it works, I guess that's the point. About reliability, I still recall the printf bug that crashed the windows 2000 kernel and microsoft stating it was a minor bug.
Overall, my point was, I can get my job done under linux. So for me it is ready for the desktop. If I needed photoshop, it would not be ready for the desktop for me.
So you develop like people used to in the 70's and early 80's? It's not as though you can't use unix. For instance, Sun offers a much better debugger and profiling tools that still work with gcc and those doesn't cost anything for students.
You mean writing code and compiling it ? Code gets written automatically nowadays ? I do as everybody does. I write code. I write test. I write documentation. And sometimes I refactor.
When everything works, scripts are written to benchmark the methods, parameters. Output log are analyzed by others scripts. All those scripts are written with appropriate language : sed, awk, gnuplot, R, matlab, idl, whatever I need. I write them quick and dirty because I need them to work now and for a couple of upcoming months. After that I am going to throw it away since my point would have been made. Probably I will use different tools if I were in a production environment.
All the tools I need are present. Why should I change ?
That's funny, you sound an awful lot like a CS undergrad to me.
And you sounds like someone that did not use a unix on desktop for a long time (which does not make it true).
The HPC market is tied to economic growth.I suspect Oracle and Microsoft are going to wipe out a lot of the linux market with Solaris and Windows respectively, now that they're focused on HPC.
We'll see. HPC market is there for as long as there is computer. HPC has driven hardware manufacturing all this, we'll see what comes next. I'll be happy to get something better than linux on cluster of PCs.
I am not bound to linux. I like it because I can change everything I need to study how a machine works. I like free software because it is what we should do morally, but it does not make it a good economical choice automatically. Linux (or GNU/Linux) is just a choice that satisfies both my professional/personnal requirement as well as my ideological wishes.
My point is just that "Linux is not ready for the desktop" is false. I know a lot of people that use it daily and they are not necessarily into IT or CS. I exemplify it on my case. I could tell you about this girl that study history of art whose only software need is a webbrowser, a document writing software and a presentation software. Ubuntu/open office/firefox gets the job done for her. I could also talk you about this guy that is webdesigner/master/what ever you call it. He needs adobe software. Linux can not do anything for him...
That is very strange how experiences differ from one guy to an other one. You can find hundreds of people saying linux never worked for me I spent weeks on it. And hundreds of people saying I gave a ubuntu DVD to my knitting-all-the-day grand mother and she never called me because she had a problem.
The interesting point is that it is also true if you swap ubuntu and windows...
First of all, I am not looking for employment, I am employed. Thank you very much. If you really are concerned for my employment, high performance computing systems does not run on windows nowadays, but usually in a linux environment. Test done on Ms windows cluster version (or whatever it is called) shows that it is pretty shitty.
The linux kernel can not be compiled with visual studio. And the scripts I use dailly are far more easy to write with gnu tools: sh/sed/awk/perl/ruby/tr/gnuplot/make/gcc than with windows tools. But it is possible to do the same thing in windows of course (C is turning complete anyway).
What I do is a "professional" use of linux since I work with it and people that work with me (in the same university or not) work with it also. Of course, it is still possible to define professional otherwise...
If you limit your scope of education to the linux platform, you never will be, either.
My education is not limited to linux. I tried to do low level things in windows and I just know that windows is crappy for that. It lacks automatic configuration tool at fine grain. It lacks good disk I/O performances (even when finely tuned). A couple of years ago infiband performances was a joke (don't know how it works today). I do not even talk about strange architecture such as cell processors, cluster of ARMs or the bluegene architecture. No windows run on that AFAIK.
Mac OS X does not represent modern kernel design by any stretch of the imagination.
Could you define "modern kernel design" for me ?
Windows, on the other hand, has a professionally designed hybrid kernel.
List of contributors to the linux kernel : Companies like IBM, Intel, SGI, MIPS, Freescale, HP, etc. are all working to ensure that Linux runs well on their hardware. (source : http://www.linuxfoundation.org/publications/linuxkerneldevelopment.php ). They clearly are teens in a basement, not professionals.
You're studying NUMA? So you're a student fiddling around with the linux kernel? Or- don't tell me some unfortunate company is paying you to waste your time with that instead of licensing a professional operating system...
No, I am a scholar. And I am studying numa effect because machines are getting numa whether you like it or not. you can not expect to scale processors without distributing the memory among processors. BTW, numa also models cluster computing fairly accurately.
You actually don't need to recompile windows to get to these features because it's got a more modern modular design.
can you easily change the scheduler to do gang scheduling ? can you change memory protection if you need it ? or schedule IO access manually ? Perhaps you can in windows, I do not care in fact. I need a system where I can change things and potentially everything. I know that kernel developers are available in the linux community to answer my questions and a bunch of documentation is available. In the worst case I can just read the source. The linux kernel is open. That's what I need.
if you used Windows on the host desktop then you'd have access to Outlook and enterprise-level networking and productivity applications on your workstation.
I do not need that kind of feature. I need a mailing system, a web browser and my compilation/debugging/analyzing tools that are fairly easy to write in a unix environment.
Why, your company might even be able to enforce group policies!
Which are useless with computer scientist.
But somehow it just doesn't seem like any company would be wasting their time with this unless they were completely incapable of cost-benefit analysis.
tells that to microsoft, IBM, and Bull that works on numa platforms...
So what is this linux desktop really offering you?
All the tools I need, that are perfectly adapted to my problems and that I can change/shape to something more suitable if I need.
The more educational institutions rely on linux to teach operating system concepts, the more we'll be preparing students to design the future operating systems of the 1970's.
Strangely I thought that microkernels came out of universities and were used by mac os and windows nowadays... I must be wrong.
Long story short, Linux works on the desktop - it's far more stable than the Redmond equivalent
parent +1. I worked in a company that did not need windows application. All the machines were running debian and it worked very well. It is an anecdote, but it worked there.
So let's talk about linux on a desktop for professional use. I am studying many/multi core architectures with numa memory. The goal is to achieve the best parallelism possible on small data size. It requires a very fine tuning of the machine. No graphics (too much process running), custom kernel to change scheduling policies, custom kernel to access hardware counters (basically, PAPI support). This cannot be done on windows, so the testing machine runs linux. Obviously the desktop that controls it runs linux also. Otherwise it would be pain in the ass.
You can say that it a niche. Well Gaming is a niche too.
Slashdot Mirror
go back to flat files aka DAT files.
Technically, that is what they do. Basically, they just say that they do not need classical RDBMS to do their job. I agree with them that RDBMS makes poor implementation of big dictionnaries. :)
For example, if I want to fetch all customers and all their bids in one query I have to use inner join. And that results in LARGE number of rows (Cartesian product of customers and their bids).
I am not sure I get your point. If you do an inner join it means you want all the tuple < player,bid > that makes sense. If there is a lot of them, well, there is a lot of them, there is nothing to do about it. If you complain about each player being repeted on several bid (since they bid more than once). It should not be a problem, as long as you stay in the RBMS, this should not incur any overhead. When you read them, you can just compress them on the fly.
If you really want to remove those "extra" player values, why do you want to have a single query ? You can just make a query for each player.
Also, I'd like to see stuff which is not easily expressed in relational algebra, like running sums or grouping on a computed field.
technically, they cannot be expressed in relationnal algebra, you have to add non algebraic operator to do that. SUM and GROUP BY in SQL are not part of relationnal algrebra. The problem with those operators is that it is difficult to do any optimization on them. Howver, the user may still want to have them. I would also be interested in a language that can express such things efficiently
A couple of years ago (2003), a cryptographic system was said to be secured if it requires more than 2^80 computation. I do not know what is the current standard and I have no clue how to find it.
However 2^110 is still way to large for us at the moment.
To give an estimation. Supose you have one million processors clocked at 10Ghz (which nobody have nowadays), you can do 10^6 * 10^10 = 10^16 ~= 2^48 computations per second. To crack AES and using this machine you'll need 2^110 / 2^48 = 2^62 seconds to do so which is approximatively 150 billions year.
The attack is of theoretical interest mainly but should tell people not to use AES with a cryptographical key smaller than 256.
I read the web from the ethernet stream, you noobs.
I played SC mostly on lan and almost never on Bnet. I just found it more fun that way.
Unless they use a bios based virus such as the one presented here : http://it.slashdot.org/article.pl?sid=09/03/23/1248214 . I know this one is a prototype, but I still find it relevant.
Is this information really pertinent ? It is the summary of the most requested website through netcraft as far as I understand. if you look at the highest uptime recorded by netcraft at http://uptime.netcraft.com/up/today/top.avg.html then you only see IIS servers. High uptime rhymes with stability. There is certainly an explanation (security update may requires reboot or stuff like that). But it makes me wonder if the stat is relevant. Someone ? Any clue ?
mmm, I am not sure it is only a matter of number of machine you can infected. But a cost/efficience analyze such as in "one hour of work, how many machine could I infect". My point is that if there was twice much Linux than Windows but requires ten times more effort to infect them, you'll probably stick with infecting Windows since it is more efficient.
There is still no comments on the article and it is already tagged as troll! :)
hey, that's three words! :)
Of course they are not different in New York. I just wanted to say that a large cluster of turbines would have a more significant local impact than the one powering a small town.
My question was more about bird migration or any long trip. But there certainly are years with less wind so it won't matter.
I'd just like people to think about it before deploying such large scale new energy. From all the replies I read, it seems some people did, which is nice.
Oh I read a nice study on the impact of high voltage lines on the health of people leaving below. The study showed a correlation between the presence of these lines and strange health diseases.... even when the lines where powered down... Nocebo effect is the worst thing to fight.
As for birds, modern prop turbines move slowly and kill fewer birds than the same number of high rise buildings with mirrored windows.
I believe GP was not concerned of birds hitting the turbines but the impact of locale wind slowing on birds. In other word, what is the local impact of a cluster of turbine that could power a city as New York and should we care about it ?
Yeah, but we do not have an infinite coal ressource, but solar resource seems infinite (Of course it is not, but we will have worse problem to deal with at this time).
Are there around some tests about other open source software that could help us understand the problem ? We can find some on open office : http://www.oooninja.com/2009/03/multiplatform-benchmark-30.html Or Tomcat : http://mediakey.dk/~cc/tomcat-performance-linux-faster-than-windows/ But that does not seem to gie a clear understanding of what's happening.
I heard of the ayle project ( http://ayle.fr/en/index.html ) which is not limited to the iphone (I do not even know if it works on it). It is kind of an hack n Slash where you have to physically move to move in the game. It is "advertised" as an hiking game.
I needed to learn fortran in january. I got quickly some slides of a physic teacher online and learned the syntax quite easily from them. I am not sure a book is needed here. Of course, I have probably missed some subtle points, but I got enough of it to understand/debug/optimize some piece of code.
It worked for years under debian for me. You just have to configure two serverlayout in /etc/X11/xorg.conf and give the right parameter to X to select the server layout at boot time.
It has to be done manually and it would of course be nice to have a GUI to configure it properly.
As the quality level of the platform continues to hover between poor to mediocre, I think more companies are going to find professional commercial systems more cost effective.
In practice it works, I guess that's the point. About reliability, I still recall the printf bug that crashed the windows 2000 kernel and microsoft stating it was a minor bug.
Overall, my point was, I can get my job done under linux. So for me it is ready for the desktop. If I needed photoshop, it would not be ready for the desktop for me.
So you develop like people used to in the 70's and early 80's? It's not as though you can't use unix. For instance, Sun offers a much better debugger and profiling tools that still work with gcc and those doesn't cost anything for students.
You mean writing code and compiling it ? Code gets written automatically nowadays ? I do as everybody does. I write code. I write test. I write documentation. And sometimes I refactor.
When everything works, scripts are written to benchmark the methods, parameters. Output log are analyzed by others scripts. All those scripts are written with appropriate language : sed, awk, gnuplot, R, matlab, idl, whatever I need. I write them quick and dirty because I need them to work now and for a couple of upcoming months. After that I am going to throw it away since my point would have been made. Probably I will use different tools if I were in a production environment.
All the tools I need are present. Why should I change ?
That's funny, you sound an awful lot like a CS undergrad to me.
And you sounds like someone that did not use a unix on desktop for a long time (which does not make it true).
The HPC market is tied to economic growth.I suspect Oracle and Microsoft are going to wipe out a lot of the linux market with Solaris and Windows respectively, now that they're focused on HPC.
We'll see. HPC market is there for as long as there is computer. HPC has driven hardware manufacturing all this, we'll see what comes next. I'll be happy to get something better than linux on cluster of PCs.
I am not bound to linux. I like it because I can change everything I need to study how a machine works. I like free software because it is what we should do morally, but it does not make it a good economical choice automatically. Linux (or GNU/Linux) is just a choice that satisfies both my professional/personnal requirement as well as my ideological wishes.
My point is just that "Linux is not ready for the desktop" is false. I know a lot of people that use it daily and they are not necessarily into IT or CS. I exemplify it on my case. I could tell you about this girl that study history of art whose only software need is a webbrowser, a document writing software and a presentation software. Ubuntu/open office/firefox gets the job done for her. I could also talk you about this guy that is webdesigner/master/what ever you call it. He needs adobe software. Linux can not do anything for him...
That is very strange how experiences differ from one guy to an other one. You can find hundreds of people saying linux never worked for me I spent weeks on it. And hundreds of people saying I gave a ubuntu DVD to my knitting-all-the-day grand mother and she never called me because she had a problem.
The interesting point is that it is also true if you swap ubuntu and windows...
First of all, I am not looking for employment, I am employed. Thank you very much. If you really are concerned for my employment, high performance computing systems does not run on windows nowadays, but usually in a linux environment. Test done on Ms windows cluster version (or whatever it is called) shows that it is pretty shitty.
The linux kernel can not be compiled with visual studio. And the scripts I use dailly are far more easy to write with gnu tools: sh/sed/awk/perl/ruby/tr/gnuplot/make/gcc than with windows tools. But it is possible to do the same thing in windows of course (C is turning complete anyway).
What I do is a "professional" use of linux since I work with it and people that work with me (in the same university or not) work with it also. Of course, it is still possible to define professional otherwise...
If you limit your scope of education to the linux platform, you never will be, either.
My education is not limited to linux. I tried to do low level things in windows and I just know that windows is crappy for that. It lacks automatic configuration tool at fine grain. It lacks good disk I/O performances (even when finely tuned). A couple of years ago infiband performances was a joke (don't know how it works today). I do not even talk about strange architecture such as cell processors, cluster of ARMs or the bluegene architecture. No windows run on that AFAIK.
Mac OS X does not represent modern kernel design by any stretch of the imagination.
Could you define "modern kernel design" for me ?
Windows, on the other hand, has a professionally designed hybrid kernel.
List of contributors to the linux kernel : Companies like IBM, Intel, SGI, MIPS, Freescale, HP, etc. are all working to ensure that Linux runs well on their hardware. (source : http://www.linuxfoundation.org/publications/linuxkerneldevelopment.php ). They clearly are teens in a basement, not professionals.
You're studying NUMA? So you're a student fiddling around with the linux kernel? Or- don't tell me some unfortunate company is paying you to waste your time with that instead of licensing a professional operating system...
No, I am a scholar. And I am studying numa effect because machines are getting numa whether you like it or not. you can not expect to scale processors without distributing the memory among processors. BTW, numa also models cluster computing fairly accurately.
You actually don't need to recompile windows to get to these features because it's got a more modern modular design.
can you easily change the scheduler to do gang scheduling ? can you change memory protection if you need it ? or schedule IO access manually ? Perhaps you can in windows, I do not care in fact. I need a system where I can change things and potentially everything. I know that kernel developers are available in the linux community to answer my questions and a bunch of documentation is available. In the worst case I can just read the source. The linux kernel is open. That's what I need.
if you used Windows on the host desktop then you'd have access to Outlook and enterprise-level networking and productivity applications on your workstation.
I do not need that kind of feature. I need a mailing system, a web browser and my compilation/debugging/analyzing tools that are fairly easy to write in a unix environment.
Why, your company might even be able to enforce group policies!
Which are useless with computer scientist.
But somehow it just doesn't seem like any company would be wasting their time with this unless they were completely incapable of cost-benefit analysis.
tells that to microsoft, IBM, and Bull that works on numa platforms...
So what is this linux desktop really offering you?
All the tools I need, that are perfectly adapted to my problems and that I can change/shape to something more suitable if I need.
The more educational institutions rely on linux to teach operating system concepts, the more we'll be preparing students to design the future operating systems of the 1970's.
Strangely I thought that microkernels came out of universities and were used by mac os and windows nowadays... I must be wrong.
Long story short, Linux works on the desktop - it's far more stable than the Redmond equivalent
parent +1. I worked in a company that did not need windows application. All the machines were running debian and it worked very well. It is an anecdote, but it worked there.
So let's talk about linux on a desktop for professional use. I am studying many/multi core architectures with numa memory. The goal is to achieve the best parallelism possible on small data size. It requires a very fine tuning of the machine. No graphics (too much process running), custom kernel to change scheduling policies, custom kernel to access hardware counters (basically, PAPI support). This cannot be done on windows, so the testing machine runs linux. Obviously the desktop that controls it runs linux also. Otherwise it would be pain in the ass.
You can say that it a niche. Well Gaming is a niche too.
And don't come with that "Free" is not "free" kwap. In PRACTICE free and Free have turned out to be the same thing.
No it's not the same thing. I use Free software and not free software. I want to keep the control on it.
PS: I agree things should work "out of the box" even if it does not in windows...