Say Intel or Sun develops a 32-core chip. Should that be licensed as one processor?
I can sort of see where Oracle, et al, are coming from. Not that I think Oracle's per-CPU pricing is completely fair, but I can see where they would want to protect their income. Right now they get a big chunk of money for multi-CPU systems, especially large scale server stuff. A 32 or 128-core chip would destroy them if it counted as one CPU.
Now, that's just how I see where Oracle is coming from. I could care less if they die or not, I don't care much for Ellison or the company.
Re:An anonymous, underground internet?
on
The Typo Millionaires
·
· Score: 4, Insightful
This is an interesting idea that I have been wondering about for some time. However, when people start saying things like "don't tell anyone how you got in" that does not make me feel comfortable. It's the whole security by obscurity thing.
I mean, if it were truely a great design then why not switch the entire Internet? A good design can handle the load. A good design is still secure (or anonymous) even if everyone used it.
Are there any other "nets inside the Net" out there?
The C language itself my not be perfect but the syntax really is one of the best. That's why it's used so often and it has nothing to do with the language itself.
Your examples don't make any sense because Ada and Eiffel have a very C-like syntax. As does Pascal, Visual Basic, and a ton of other languages.
C is just a very concise version of the same syntax. This is why it's better than the others. It has power without extra fluff. It's a perfect starting point for making a more powerful language.
Smalltalk does not use a C-like syntax though and that is one reason why no one uses Smalltalk. Its syntax sucks.
The perfect language would have an extended C-like (or C++-like) syntax. The extensions to the syntax would make functional programming easier. They would allow things like heavy use of recursion without performance loss. Being able to choose between mutable and nonmutable variables would be good too (especially if the language made this very efficient).
The problem is distribution. The Internet simply can not (currently) handle it.
Think about it, last year 800 million people watched the Super Bowl. I have no idea how many this time, but more for sure.
And how long was the game? 4 hours give or take an hour for pre/post game crap. That 4 hours at even low quality DVD is pushing minimum 4 GB of data. I won't even mention HDTV and other qualities people might want.
And most broadband ISP's these days have caps. That 4GB of data would put me well over my cap for the day. Not to mention I would be limited to a certain number of shows per month (and that's assuming I don't download the latest Ubuntu ISO and whatnot).
Lets see, 4 GB times 800 million people, is... uh... A number I don't even want to calculate. Even if you only count broadband users (50+ million?) that's impossibly huge.
And even "minor" shows with only a couple million people watching would be hard to impossible to distribute.
LISP is a very powerful and useful but the syntax sucks.
Sure, people will say using the right editor and being used to the syntax help, but seriously, if it were that easy and great then everyone would be using LISP. The plain fact is that relatively few people use LISP these days. I would say this is due to the syntax more than anything else (performance being second although it's pretty good these days).
It's the same problem O'Caml faces. O'Caml is pretty powerful and can be fast but the syntax sucks. Plain SML is much better than the basterdized crap that O'Caml uses. Despite the fact that you can change its syntax, no one ever does because it then becomes non-standard.
Not really. The pre-linking can certainly help regular C applications but its real purpose is to speed up all the dynamic code that Objective-C and poorly written C++ are using. No one had used prelinking much until OS X and KDE came round.
Plain C and C++ with templates are your good friends when properly wielded.
Objective-C sucks. Sorry but run-time binding and dereferencing everything gives you something much slower than it could be. Virtual functions are not your friend. RTTI is not your friend.
That's one of the reasons OS X has to sit there for ages "optimizing" your system after you install anything. It is running the pre-linker on everything. Otherwise your system runs like shit, like the first original OS X release which didn't do any pre-linking. Something they would not have to do if they did not use so much run-time crap. This issue is caused by a combination of Objective-C, poor use of C++, and poor design. You can go too far with OOP. I mean, how much common software do you see written in SmallTalk?
Proper software strikes a balance between disciplines. Imperative, object oriented, functional, they all have their uses. It is a mistake to go all the way in one direction.
I didn't find that a problem because there are several utilities that will do this for you on Windows.
My problem is that it won't minimize the "system tray" (or equivalent) in either KDE or GNOME. It boggles my mind as to why they can't just add that relatively simple feature. Ugh, come on people! This is 2005, not 1995.
I've been using Evolution for ages because of just that one missing feature. I don't care for Evolution because the S/MIME support limited (no PKCS#11 support) or broken.
I agree with a lot of what you said. I do think part , or most, of the problem is more related to the fact that exercise is hard rather than boredom. It becomes much easier over time and less "boring." No matter the exercise.
I love playing DDR (actually StepMania) for hours at a time though. I have those fancy RedOctane pads and they do work well. However, I ended up building my own hard pads that I could wear shoes on. You really need shoes to protect your shins and ankles (shin splints hurt). I still alternate between the soft and hard pads depending on my mood, but I use the hard pads with shoes most of the time because it lets me play longer without damaging anything (I also run and left weights).
It's not that we wouldn't need science any more, it would just change science radically. The problem space would change. We might solve all the old problems but then there would be a whole new set to work on.
It would be quite interesting for sure. I'm just imagining the computer games we could have.
The "core development team" has not always been the same people.
I have never run any version of MythTV that would not crash for some reason or another. I tend to be a heavy user and push software to its limits though. I just learn what to avoid to prevent crashes, but it has never been 99% bug free for me.
The key word is usually. I was writing professional quality code by college as well. But that was only because by the time I was 18, I already had some 20,000 hours programming experience. Most people do not dedicate their lives to programming like that, especially not at such an early age. Some of us do.
Again: You can protect the stupid people from the world if you want, but you can't protect them from themselves.
Pffft, right. I'm as geeky as they come but I want my system to be secure without me having to think about it. I got code running through my head all day long, the last thing I need to think about is whether or not my system in secure. I do want my system to be secure and protect me though. The OS needs to do that for me because I don't want to care about that stuff.
It seems that Freevo would be far easier to modify and write plugins for than Myth (which mostly written in C).
Probably, Myth is written mostly in C++ though. The code is a horrible hack job though. Written by a college student. Not that all college student projects suck, it's just that usually the inexperienced create poor designs initially.
Even the creators of Freevo claim they did not know Python before they started.
This is not a good thing either. Same problem as above.
Overall, MythTV looks a lot better and has more features because more people have been working on it. The code is often buggy and hard to hack on. However, it does work resonably well and I've been happily using MythTV for over a year with no major issues.
I used Arch Linux for about 6 months not too long ago.
It has some annoying quirks but nothing too bad I guess. The biggest problem is lack of packages and the very slow package manager. Although packages are easy to create, I got tired of having to create packages for every little thing I needed. After looking at the code for the package manager I could tell it was written by someone who did not know what they were doing. Maybe that has improved since 0.6, I don't know.
It is not any faster than any other distro. With the exception of boot time, all distros are really about the same performance-wise. In fact, I found the Arch package manager was incredibly slow (try running Arch on an old machine).
Overall, between the goofy quirks, lack of packages, and lack of coding skill, I went back to my old friend Debian (actually, Ubuntu Linux).
"I've heard of the language since the beginning, used it in the late 90's, and still think it's terrible"
I don't think you understand my experience. I was using Java for development on systems that went into production in the early 90's. I still develop in Java to this day as I have been for the last 10 years. It's not my choice but clients sometimes want Java and sometimes I do work for Sun. Trust me, I know Java.
Ultimately, though, as a "consultant working on all sorts of operating systems and development environments", you should be able to tell the difference between appearance and performance. In this case, I think you're misleading yourself.
I am an extremely intuitive person. Often it is difficult for me to translate my reasons into something a normal person can understand. Usually, like now, I won't even try. Lets just say we disagree and leave it at that. Suffer on my friend.
I didn't want the PSP's LCD screen to become any smaller than this, nor did I want its machine body to become any larger.
That's why. It sounds like one of those things that someone just gets "stuck" on. Like a broken record. "I will not compromise on that." To the exclusion of everything else, even usability. Pffft, that's how crappy products get made. We're probably talking about making the PSP 2mm wider or something.
Sometimes we get so far down inside a project that we can't see what is going on. It's good to sometimes to step back and evaluate your position to make sure you don't have blinders on.
Whoop, you're on Linux. SWT does have issues on Linux. However, SWT (the GUI library that Eclipse is written in)... and listen to this closely... SWT is not Java. You're decrying an entire language based on a third-party tool that was neither developed nor supported by the makers of the platform.
I've been a Java developer since Java was just an experimental beta language that no one knew existed. I think I have enough experience and justification to say Java sucks.
I know what SWT is. I would hate to see what Eclipse would be like in Swing or something. Actually, there probably would be no Eclipse in that case because AWT & Swing suck more than Java does.
Who said I run Visual Studio on Linux? I just said that's one of the native IDE's I'm used to. As a consultant I work on all sorts of operating systems and development environments.
And before you start spouting some gibberish please note that I said it was less snappy, not that it wasn't snappy at all. I said that in my original post as well. It's fairly fast on my fast machine but it is slower than the other applications I normally use.
Did you read what I wrote? I just tried it. The latest version.
I think it's in your head because you're not used to using native environments. I'm used to Kdevelop, Anjuta, and Visual Studio. Eclispe is most definitely "funky" feeling compared to those. It's not as snappy.
It all depends on what you get used to, but that doesn't change the fact that it is less snappy than a native application.
I tried the flash demo. It's like a virtual screen that you can zoom in and out. Virtual screens have been available for years. The zooming thing is kind of interesting but not anything new. I assume the interface is based on a vector drawing back-end (display postscript/PDF?).
This has several problems. Is this thing suppose to manage all your documents and applications? Does that mean everything is being displayed and active at the same time? The CPU and memory requirements of this must be off the chart. This thing would totally choke based on the pure amount of data I have on my machine. Can this interface handle a terabyte or more of information?
Spacial interfaces suck anyway. It might seem like it is better for organizing your data because you can group things together and "zoom out" to view everything on a large scale, but in real life you're going to spend too much time zooming in and out trying to find what you are looking for. It is very much like those suck-ass 3D file managers that someone creates every once in a while.
I suppose you could query for items and they could be marked similar to MapQuest, then you could zoom in on it. That sounds like a very tedious to use interface though.
Really, the current UI system that most computers use is not a bad design, it just needs refinement. Modern UI's just need to be better about remembering which data items I've been working with recently and which items go with each other. We are already seeing the beginnings of this with things like "favorites" and "home/desktop" in most file dialogs these days. That just needs to be taken to a higher level and cleaned up.
Sorry if my post is disorganized, I just woke up...
Good sales pitch. It has been a long time since I tried Eclipse so I'm looking at it again now (plus the CDT plugin). First off, I think Java is stupid, but I'll use any IDE if it is good.
My impression is that they have definitely come a long way. The IDE is probably the closest to the nice Visual Studio environment that I've seen. The install was relatively easy... other than the fact I had to download 87 MB for Eclipse only to realize I had to download another 16 MB for a JRE. Ugh, Java... It is a nice IDE though.
However, it still has that funky slow-ass feeling that all applications written in Java have. It's not that it is unusably slow, it's just slower and clunkier feeling than normal native applications. Netbeans feels even worse. I hate that feeling. It feels like I have a runaway process on my machine or something that is sucking CPU away from the IDE (note this is not because of the slowness of my 4G RAM, dual Opteron 250 machine).
I'll probably try using Eclipse for a while because I like the IDE overall, but Java still sucks. Meh, too bad they didn't use a real language to develop Eclipse in. All that wasted money, time, and effort.
Slashdot Mirror
Say Intel or Sun develops a 32-core chip. Should that be licensed as one processor?
I can sort of see where Oracle, et al, are coming from. Not that I think Oracle's per-CPU pricing is completely fair, but I can see where they would want to protect their income. Right now they get a big chunk of money for multi-CPU systems, especially large scale server stuff. A 32 or 128-core chip would destroy them if it counted as one CPU.
Now, that's just how I see where Oracle is coming from. I could care less if they die or not, I don't care much for Ellison or the company.
This is an interesting idea that I have been wondering about for some time. However, when people start saying things like "don't tell anyone how you got in" that does not make me feel comfortable. It's the whole security by obscurity thing.
I mean, if it were truely a great design then why not switch the entire Internet? A good design can handle the load. A good design is still secure (or anonymous) even if everyone used it.
Are there any other "nets inside the Net" out there?
The iSight has serious problems in low-light conditions. And by "low light" I mean typical indoor lighting.
I mean, the quality looks good... when you can make anything out. The iSight really needs direct sunlight or otherwise a very well lit environment.
The C language itself my not be perfect but the syntax really is one of the best. That's why it's used so often and it has nothing to do with the language itself.
Your examples don't make any sense because Ada and Eiffel have a very C-like syntax. As does Pascal, Visual Basic, and a ton of other languages.
C is just a very concise version of the same syntax. This is why it's better than the others. It has power without extra fluff. It's a perfect starting point for making a more powerful language.
Smalltalk does not use a C-like syntax though and that is one reason why no one uses Smalltalk. Its syntax sucks.
The perfect language would have an extended C-like (or C++-like) syntax. The extensions to the syntax would make functional programming easier. They would allow things like heavy use of recursion without performance loss. Being able to choose between mutable and nonmutable variables would be good too (especially if the language made this very efficient).
The problem is distribution. The Internet simply can not (currently) handle it.
Think about it, last year 800 million people watched the Super Bowl. I have no idea how many this time, but more for sure.
And how long was the game? 4 hours give or take an hour for pre/post game crap. That 4 hours at even low quality DVD is pushing minimum 4 GB of data. I won't even mention HDTV and other qualities people might want.
And most broadband ISP's these days have caps. That 4GB of data would put me well over my cap for the day. Not to mention I would be limited to a certain number of shows per month (and that's assuming I don't download the latest Ubuntu ISO and whatnot).
Lets see, 4 GB times 800 million people, is... uh... A number I don't even want to calculate. Even if you only count broadband users (50+ million?) that's impossibly huge.
And even "minor" shows with only a couple million people watching would be hard to impossible to distribute.
It's too much data.
LISP is a very powerful and useful but the syntax sucks.
Sure, people will say using the right editor and being used to the syntax help, but seriously, if it were that easy and great then everyone would be using LISP. The plain fact is that relatively few people use LISP these days. I would say this is due to the syntax more than anything else (performance being second although it's pretty good these days).
It's the same problem O'Caml faces. O'Caml is pretty powerful and can be fast but the syntax sucks. Plain SML is much better than the basterdized crap that O'Caml uses. Despite the fact that you can change its syntax, no one ever does because it then becomes non-standard.
Not really. The pre-linking can certainly help regular C applications but its real purpose is to speed up all the dynamic code that Objective-C and poorly written C++ are using. No one had used prelinking much until OS X and KDE came round.
Plain C and C++ with templates are your good friends when properly wielded.
Objective-C sucks. Sorry but run-time binding and dereferencing everything gives you something much slower than it could be. Virtual functions are not your friend. RTTI is not your friend.
That's one of the reasons OS X has to sit there for ages "optimizing" your system after you install anything. It is running the pre-linker on everything. Otherwise your system runs like shit, like the first original OS X release which didn't do any pre-linking. Something they would not have to do if they did not use so much run-time crap. This issue is caused by a combination of Objective-C, poor use of C++, and poor design. You can go too far with OOP. I mean, how much common software do you see written in SmallTalk?
Proper software strikes a balance between disciplines. Imperative, object oriented, functional, they all have their uses. It is a mistake to go all the way in one direction.
Awesome, thanks for those links. I didn't know they existed.
I didn't find that a problem because there are several utilities that will do this for you on Windows.
My problem is that it won't minimize the "system tray" (or equivalent) in either KDE or GNOME. It boggles my mind as to why they can't just add that relatively simple feature. Ugh, come on people! This is 2005, not 1995.
I've been using Evolution for ages because of just that one missing feature. I don't care for Evolution because the S/MIME support limited (no PKCS#11 support) or broken.
I agree with a lot of what you said. I do think part , or most, of the problem is more related to the fact that exercise is hard rather than boredom. It becomes much easier over time and less "boring." No matter the exercise.
I love playing DDR (actually StepMania) for hours at a time though. I have those fancy RedOctane pads and they do work well. However, I ended up building my own hard pads that I could wear shoes on. You really need shoes to protect your shins and ankles (shin splints hurt). I still alternate between the soft and hard pads depending on my mood, but I use the hard pads with shoes most of the time because it lets me play longer without damaging anything (I also run and left weights).
It's not that we wouldn't need science any more, it would just change science radically. The problem space would change. We might solve all the old problems but then there would be a whole new set to work on.
It would be quite interesting for sure. I'm just imagining the computer games we could have.
I see it in kids today all the time.
This is most certainly due to living in the post-Napster, post-9/11, political & legal environment.
The "core development team" has not always been the same people.
I have never run any version of MythTV that would not crash for some reason or another. I tend to be a heavy user and push software to its limits though. I just learn what to avoid to prevent crashes, but it has never been 99% bug free for me.
The key word is usually. I was writing professional quality code by college as well. But that was only because by the time I was 18, I already had some 20,000 hours programming experience. Most people do not dedicate their lives to programming like that, especially not at such an early age. Some of us do.
Again: You can protect the stupid people from the world if you want, but you can't protect them from themselves.
Pffft, right. I'm as geeky as they come but I want my system to be secure without me having to think about it. I got code running through my head all day long, the last thing I need to think about is whether or not my system in secure. I do want my system to be secure and protect me though. The OS needs to do that for me because I don't want to care about that stuff.
It seems that Freevo would be far easier to modify and write plugins for than Myth (which mostly written in C).
Probably, Myth is written mostly in C++ though. The code is a horrible hack job though. Written by a college student. Not that all college student projects suck, it's just that usually the inexperienced create poor designs initially.
Even the creators of Freevo claim they did not know Python before they started.
This is not a good thing either. Same problem as above.
Overall, MythTV looks a lot better and has more features because more people have been working on it. The code is often buggy and hard to hack on. However, it does work resonably well and I've been happily using MythTV for over a year with no major issues.
I used Arch Linux for about 6 months not too long ago.
It has some annoying quirks but nothing too bad I guess. The biggest problem is lack of packages and the very slow package manager. Although packages are easy to create, I got tired of having to create packages for every little thing I needed. After looking at the code for the package manager I could tell it was written by someone who did not know what they were doing. Maybe that has improved since 0.6, I don't know.
It is not any faster than any other distro. With the exception of boot time, all distros are really about the same performance-wise. In fact, I found the Arch package manager was incredibly slow (try running Arch on an old machine).
Overall, between the goofy quirks, lack of packages, and lack of coding skill, I went back to my old friend Debian (actually, Ubuntu Linux).
"I've heard of the language since the beginning, used it in the late 90's, and still think it's terrible"
I don't think you understand my experience. I was using Java for development on systems that went into production in the early 90's. I still develop in Java to this day as I have been for the last 10 years. It's not my choice but clients sometimes want Java and sometimes I do work for Sun. Trust me, I know Java.
Ultimately, though, as a "consultant working on all sorts of operating systems and development environments", you should be able to tell the difference between appearance and performance. In this case, I think you're misleading yourself.
I am an extremely intuitive person. Often it is difficult for me to translate my reasons into something a normal person can understand. Usually, like now, I won't even try. Lets just say we disagree and leave it at that. Suffer on my friend.
I didn't want the PSP's LCD screen to become any smaller than this, nor did I want its machine body to become any larger.
That's why. It sounds like one of those things that someone just gets "stuck" on. Like a broken record. "I will not compromise on that." To the exclusion of everything else, even usability. Pffft, that's how crappy products get made. We're probably talking about making the PSP 2mm wider or something.
Sometimes we get so far down inside a project that we can't see what is going on. It's good to sometimes to step back and evaluate your position to make sure you don't have blinders on.
Whoop, you're on Linux. SWT does have issues on Linux. However, SWT (the GUI library that Eclipse is written in) ... and listen to this closely... SWT is not Java. You're decrying an entire language based on a third-party tool that was neither developed nor supported by the makers of the platform.
I've been a Java developer since Java was just an experimental beta language that no one knew existed. I think I have enough experience and justification to say Java sucks.
I know what SWT is. I would hate to see what Eclipse would be like in Swing or something. Actually, there probably would be no Eclipse in that case because AWT & Swing suck more than Java does.
Who said I run Visual Studio on Linux? I just said that's one of the native IDE's I'm used to. As a consultant I work on all sorts of operating systems and development environments.
And before you start spouting some gibberish please note that I said it was less snappy, not that it wasn't snappy at all. I said that in my original post as well. It's fairly fast on my fast machine but it is slower than the other applications I normally use.
Did you read what I wrote? I just tried it. The latest version.
I think it's in your head because you're not used to using native environments. I'm used to Kdevelop, Anjuta, and Visual Studio. Eclispe is most definitely "funky" feeling compared to those. It's not as snappy.
It all depends on what you get used to, but that doesn't change the fact that it is less snappy than a native application.
I tried the flash demo. It's like a virtual screen that you can zoom in and out. Virtual screens have been available for years. The zooming thing is kind of interesting but not anything new. I assume the interface is based on a vector drawing back-end (display postscript/PDF?).
This has several problems. Is this thing suppose to manage all your documents and applications? Does that mean everything is being displayed and active at the same time? The CPU and memory requirements of this must be off the chart. This thing would totally choke based on the pure amount of data I have on my machine. Can this interface handle a terabyte or more of information?
Spacial interfaces suck anyway. It might seem like it is better for organizing your data because you can group things together and "zoom out" to view everything on a large scale, but in real life you're going to spend too much time zooming in and out trying to find what you are looking for. It is very much like those suck-ass 3D file managers that someone creates every once in a while.
I suppose you could query for items and they could be marked similar to MapQuest, then you could zoom in on it. That sounds like a very tedious to use interface though.
Really, the current UI system that most computers use is not a bad design, it just needs refinement. Modern UI's just need to be better about remembering which data items I've been working with recently and which items go with each other. We are already seeing the beginnings of this with things like "favorites" and "home/desktop" in most file dialogs these days. That just needs to be taken to a higher level and cleaned up.
Sorry if my post is disorganized, I just woke up...
Good sales pitch. It has been a long time since I tried Eclipse so I'm looking at it again now (plus the CDT plugin). First off, I think Java is stupid, but I'll use any IDE if it is good.
My impression is that they have definitely come a long way. The IDE is probably the closest to the nice Visual Studio environment that I've seen. The install was relatively easy... other than the fact I had to download 87 MB for Eclipse only to realize I had to download another 16 MB for a JRE. Ugh, Java... It is a nice IDE though.
However, it still has that funky slow-ass feeling that all applications written in Java have. It's not that it is unusably slow, it's just slower and clunkier feeling than normal native applications. Netbeans feels even worse. I hate that feeling. It feels like I have a runaway process on my machine or something that is sucking CPU away from the IDE (note this is not because of the slowness of my 4G RAM, dual Opteron 250 machine).
I'll probably try using Eclipse for a while because I like the IDE overall, but Java still sucks. Meh, too bad they didn't use a real language to develop Eclipse in. All that wasted money, time, and effort.