You are basically saying that people stopped liking Java when Sun went back on their promise to make Java free-as-in-Perl, and that people dislike VB because it's proprietary and from Microsoft.
You are quite right.
But that's not some gratuitous dislike of things commercial, it has concrete technical reasons. As long as Sun was going to go through a public standardization process, its many design flaws were fixable, its bloat could have been limited, and other people could have created better implementatiosn than Sun. As soon as Sun made Java proprietary-with-community-input, all that went out the window.
As for VB, you are confusing the language and its environment. VB-the-language is pretty awful--some of Microsoft's main VB hackers have left the company over that. What has made VB succeed is the programming environment and the copious documentation, training, and marketing. VB-the-language succeeded because a big company kept spending on making it succeed despite the pretty awful nature of the language itself.
So, "free" matters, because something that is "free" usually involves much more community input in its creation, and something that is "free" usually has to make it on its own merits. Java and VB are, and will remain, deeply flawed languages precisely because they are not free: as long as Microsoft and Sun can spend money to paper over the technical problems with those languages, and as long as the community itself doesn't have the final say in what happens to those languages independent of the corporate interests of their owners, those languages will remain sub-par.
(As for VB-like environments for Linux, Python, Glade, and your favorite Python IDE give you equivalent functionality and a better language. What they don't give you is the documentation, standardization, and name recognition of VB.)
Except I forgot to mention that we're getting a new building RSN, and we need to have our servers be both accessible and *fast* from both. [...] AFS's global namespace is not at *all* like/net/host, because even writable volumes can be moved from host to host completely transparently, even when in use!
NFS is just a protocol for clients accessing servers. Many forms of replication and migration can be supported even with just NFSv2, purely on the server side. And NFSv4 has direct support for these features (and I'm not at all convinced that adding all that junk to NFS was a good idea either).
Overall, basically, you are so excited about a bunch of gimmicks that you completely overlook the day-to-day reality of AFS: poor large file support, poor large volume support, exceptionally poor implementation of UNIX file system semantics, and poor handling of error conditions.
(And migrating users duing the day is just silly, even if your network file system nominally supports it.)
WHAT?! I could say the same thing about UNIX's user/group/world semantics, and far more defensibly. ACLs allow all sorts of useful things; I can have a log directory that's append-only except to sysadmins, for instance
Yes, and that is one of the problems with AFS. A network file system for UNIX should foremost try to be as compatible as possible with UNIX, not to invent a completely different set of behaviors.
It's a reasonable choice for mine (a fairly small [approx 40-user] software company with lots of servers and few Windows clients -- and lots of potential for need to scale).
Sounds like your needs are really modest--just about anything would probably work for you.
NFS may be capable of scaling up to my network -- but it sure as hell couldn't scale up to IBM's.
What is "scaling up to IBM's network" supposed to mean? Sure, unlike NFSv2, AFS has reasonable support for security and user ids, but that's been fixed in current versions of NFS, and it doesn't really work all that well in practice in AFS either anyway. In terms of performance, AFS is usually worse. And the much touted global name space for AFS amounts to little more than "/net/host/..." under NFS. What kills AFS is that it has a steep learning curve, is inflexible, and, most of all, doesn't even try to support UNIX semantics.
Only if your mount tables are the same everywhere,
That's what NIS is for. Furthermore, the flexibility of being able to set up machines with different views of the network is crucial in many applications. None of my workstations or servers actually have the same mount tables: they all get some stuff via NIS, and some stuff is modified locally. The restrictions AFS imposes are just unacceptable.
AFS makes administration tremendously easier after one's scaled the initial learning curve.
AFS is an administrative nightmare. Apart from the mess that ACLs cause and the problems of trying to fit real-world file sharing semantics into AFS's straightjacket, just the number of wedged machines due to overfull caches and its complete disregard for UNIX file system semantics cause no end of support hassles. Then, there is the poor support for Windows clients. We started out using AFS because it sounded good on paper, but it was a disaster in terms of support, and we got rid of it again after several years of suffering.
It performs far, far better than NFS on large networks (and merely somewhat better on smaller ones).
AFS's caching scheme works better than what NFS is doing for small files, but that case is fast and easy anyway. AFS's approach falls apart for just the kind of usage where it would matter most: huge files accessed from many machines.
Both NFS and AFS have very serious problems. But between the two, NFS is far simpler than AFS, is easier to administer in complex real-world environments, respects UNIX file system semantics better, and works better with large files. I can guardedly recommend NFS or SMB ("there is nothing better around, so you might as well use it"), but I can't imagine any environment for which AFS is a reasonable choice anymore. The only thing AFS had ever going for it as far as I'm concerned is that it was fairly secure at a time when NFS had no security whatsoever, but that is not an issue anymore.
Right now, I think the answer is to run NFS: it's by far the easiest to set up and the best in a UNIX environment. AFS, CODA, Intermezzo, and SMB are pretty iffy in comparison.
In the medium term, however, I think WebDAV will become a better option, because it can be served and accessed with standard web servers and clients, in addition to being mappable onto the file system.
The Linux kernel already has WebDAV support (CODA hooks plus some user-mode process), although I'm not sure how well it works.
Of course it is. It gives you a single, unified view of a file system tree that can span many machines.
It doesn't support any: replication, disconnection, sharing, distribution.
Sure it does. Some of that functionality requires more than a vanilla NFS server, but that can be transparent to clients.
It is centralised, requires the same user names|numberpace and security.
Older versions did, current versions don't.
Don't get me wrong, NFS has lots of problems in many environments. But for networking a handful of machines in a home environment, it is nearly perfect. In fact, NFS version 2 is probably the best choice (the old one without all the security stuff). Furthermore, the alternatives (AFS, SMB, CODA, etc.) are harder to administer, perform worse, and have lots of problems with UNIX file system semantics.
UNIX was created by a group of people at Bell Labs. Ultimately, it should be up to them what constitutes UNIX.
Research versions of UNIX were based on bits and pieces from BSD, but they involved removing a lot of functionality, so by looking at the documentation as well as their follow-on, Plan 9, you can get a pretty good idea of what they considered good and bad.
Based on conversations I have had with the Bell Labs folks over Plan 9, I suspect that they probably wouldn't want to take responsibility for the OS X kernel.
The eth1394 driver has been in the Linux kernel for a while. You can find a discussion of using IEEE 1394 for compute clusters here.
And, you can do the same with both USB 1 and USB 2.
Generally, Gigabit Ethernet is more flexible, easier to maintain, and has more third party hardware available for it, but if you have a motherboard with FW and are setting up a special-purpose, low-cost cluster, IEEE1394 or USB2 networking may be a reasonable choice.
The Fujitsu LifeBook P-1000 and P-2000 seem to be good choices and apparently run Linux quite well. They get up to 15h battery life. The keyboard is a bit cramped, though. They start at around $1200.
Some of the Sharp laptops (though not the ultra-small MM10) and the Sony PictureBook seem to be alternatives and have also been reported to run Linux.
The Actius MM10 comes with Windows pre-installed. Sharp's warnings about the machine not running anything other than the pre-installed OS are unusually dire (you may not even be able to install Windows XP Professional on it).
If you have unearthed credible reports of Linux installations on the MM10, please share them with the rest of us. Until then, the MM10 should probably be considered a "Windows-only" machine.
Another interesting question about the machine that has, so far, been unanswered, is whether its use as a dockable disk-drive is OS dependent. That is, is the USB mass storage implemented by hardware that is separate from the main processor, or is it implemented in the BIOS, or is it implemented by some piece of code running under Windows?
The iBook is pretty big compared to many x86-based subnotebooks. It's battery-life is so-so among subnotebooks (a good subnotebook can get 8-12h of running time these days).
As for the software, OS X is not Linux; there is far less UNIX software available for OS X than for Linux and porting software to it can be a pain. PPC Linux works surprisingly well, but it still is second rate compared to the x86 versions of Linux.
Buy an iBook if you like OS X and want to run OS X applications. For anybody looking for a Linux laptop, there are better choices out there.
There really is no such thing as a "powerline Ethernet". Ethernet specifies the electrical characteristics of the network cables, and your home wiring doesn't conform.
In any case, I kind of don't see the point. Why not just a a wireless repeater? With 802.11a/g that's a lot faster than this, and it means you don't have two separate networking technologies to hassle with. And usually, you want to cover an area completely anyway, so all wireless access points need to overlap, which mean that they can act as repeaters.
Apple has been imitating and bundling other people's technologies throughout their corporate history. Their entire product line is based on ideas and technologies developed elsewhere.
Apple didn't invent color management or color science either. They simply happened to have a large graphics arts user community, so they were the first to incorporate this stuff into their systems.
Microsoft, Apple, and Linux are each years behind the state of the art in many areas. Windows XP is some VMS work-alike on steroids, and OS X is a warmed-over version of NeXTStep. And Linux still gives you that warm-and-fuzzy UNIX feeling from, oh, 15 years ago. That's the way it is with commercial or real-world systems. Just because Apple happens to incorporate some feature into their system first doesn't give them claim to it in perpetuity.
Wrong. The Apple Lisa had a pre-emptive multi-tasking OS with protected memory, but the hardware cost too much (the Motorola 68k in particular had a paging bug at the time that required them to use their own MMU).
Yes, and the Lisa itself was a clone of a similarity priced Xerox machine, which also failed. It's not like Apple invented the technology behind either the Lisa or the Mac.
The Mac won in the marketplace over the Lisa, therefore it can be argued that co-operative multi-tasking and a simple memory model are better.
And Windows won in the marketplace over Apple. It can therefore, by your reasoning, be argued that Windows is better. Right?
'A scenario could be imagined where an attacker would do this to delay the arrival of an important letter,
Letters that are that important should be sent by registered mail.
to wreak havoc on the postal system for political reasons,
Provided the US government isn't subsidizing junk mail (if they are, they should stop), every piece of junk mail that is sent makes the USPS a small profit. Well, then let them "wreak" away.
or even worse, to serve as a diversion for a terrorist act, such as the mailing of a contaminated letter.'
I somehow have a hard time seeing how this is a serious risk, over and above the general risk of "contaminated letters".
Remember that security consultants and "experts", like politicians, have a tendency to create unnecessary fear in order to hype up their own importance.
Linux has half a percent of the desktop market. Apple, with MacOS, has something like 4-5%, I think?
Repeating made up numbers doesn't make them true. The fact is that nobody knows how many people are using Linux. It actually seems likely to me that there are more Linux desktop users out there than there are OS X users (many of them dual-booting Windows).
In any case, no matter what the market share is, Apple needs to realize that Linux users can cause them real problems: as this shows, the Linux community is capable of developing codecs themselves if Apple doesn't supply them. In the long run, that erodes Apple's control over a market segment that they desperately want to control. Linux MPEG-4 encoders are already among the best and fastest in the business, for example.
Of course, that's what I would like to see happen anyway: so, keep up the good work--keep NOT shipping proprietary codecs for Linux, we don't want them anyway.
"Exponential growth" just means that something grows as k^t for some constant k, and just about everything does. Your checking account grows exponentially, even though you probably only get 1% interest: after t years, you have 1.01^t dollars, or about 2^(t/70) dollars, if you like.
Most mature industries grow exponentially because they grow with the economy as a whole. Any economist that would propose that the economy not grow exponentially would be lynched by politicians--our whole economic system assumes that we can sustain at least modest exponential growth indefinitely (whether that is reasonable is a different question).
The IT industry will continue to grow exponentially, just like almost any other sector . What it won't do is have growth rates that exceed that of the economy as a whole. That is probably what the Economist means.
The Economist just lost a lot of credibility in my eyes: misusing the term "exponential" in this way is something that just shouldn't get past the editors of any publication that claims any competence in economics.
Yes, Microsoft gives you consistently gray-and-blue dialog boxes. Lots of them. You can click around in them for hours trying to get anything done. But it's consistent.
and while the people exist who can administer those, the configuration for say, Apache, is wildly different to just about anything else, and anything else from each other. Just an observation.
And you think that the configuration of different Windows utilities is consistent just because you see them all through dialog boxes? Underneath that "consistent" veneer of Windows is the same variety of configuration semantics as exists for Linux. But, hey, it appears consistent.
The 2nd point is support. It's impeccable, and having guaranteed 24-hour help for those times when things foo bar up so badly we can't repair things is essential to running a service for our clients.
Support costs money. If you pay enough for Windows support, you get "impeccable" Windows support. If you pay enough for Linux support, you get "impeccable" Linux support. However, if you are any reasonably sized business, you are big enough to have enough Linux expertise in-house so that you don't need any outside support.
I won't comment on their business practices, but suffice it to say that's their choice.
I won't comment on your business practices. Suffice it to say that there are a lot of companies like yours that have way too much money and don't know what they are doing.
You get lots of nano-particles into your lungs every day, much of it from Diesel engines and coal power plants. Before worrying about NRAM, we should worry about those first.
on the other hand you can't upgrade the memory speed beyond what your entire CPU supports
Wow, I have to upgrade the entire CPU. Like, I get a new motherboard with 128Gbytes of nanotube memory, but I can't put bits and pieces of my old Opteron on it. Perfectly shocking. What was AMD thinking.
Palm is trying to pull a Microsoft: establish a poor, cumbersome API as an industry standard and reap profits from a strong market position. And it's working, sort of: while the operating system is pretty awful internally, PalmOS is where all the applications are, it's what the handheld developers know, and where most of the nifty hardware is being developed.
Palm is just like Microsoft, only that their software is even worse. As far as I'm concerned, Palm isn't cool, and my m505, which I bought a few years ago, is probably the last Palm handheld I will ever own.
You didn't understand the question. Yes, Sun has added generics to the language and compiler. But without changes to the JVM, Java generics are not type-safe across compilation boundaries.
Scientific and numerical computing is not general purpose.
No, and server-side computing, desktop application development, graphics, or any other specific application isn't "general purpose computing" either. What makes a language "general purpose" is that it has acceptable support for all those application areas, even if it doesn't excel at any of them.
In your market you need absolute controll over your app and java certainly does not allow this.
It is folly to think of numerical computing as being a separate "market". Do you want E-mail filtering? Do you want text-based retrieval? Do you want to browse images? Do you want to mine databases for patterns? Do you want to analyze business data? Do you want to edit video? Do you want nicely laid out text? Do you want great looking outline fonts? Do you want good process scheduling? Do you want good intrusion detection? Do you want smart opponents in games? Good, modern implementations of all of those have strong numerical components, in addition to requiring good UIs and other facilities. Java makes it a lot of unnecessary work right now to do a decent job at implementing their numerical components.
If you read the review you will see support for generics.
You are saying that Sun is extending the JVM to support generics in 1.5? Ummm--where exactly does it say that?
Value classes are still not there and I agree with the notion that operator overloading is evil and wish it was never invented. Its only purpose is to produce very bad code and create debugging nightmares. You can cast values to different data types when necessary. It beats using overloaded operators depending on value type because it can be confusing to other people looking at your code through a debugger. This is not the early 80's anymore where ram is a premium. Operator overloading was added when c++ programes were measured in kb's.
The applications we write barely fit into several Gigabytes of memory and they involve complicated mathematical formulas.
Basically, your attitude is typical, and it's ignorant: "I don't need it, so it must be useless". Fine, but as a result, Java is simply unsuitable for an increasing range of applications. And the fact that Java's C++ interface is so inefficient and cumbersome means that Java doesn't have an easy workaround either.
C# looks like a pretty general purpose, modern programming language, but Java clearly is not because it is not a good language for scientific and numerical computing, and because it does not interface well with existing C/C++ code.
Slashdot Mirror
You are quite right.
But that's not some gratuitous dislike of things commercial, it has concrete technical reasons. As long as Sun was going to go through a public standardization process, its many design flaws were fixable, its bloat could have been limited, and other people could have created better implementatiosn than Sun. As soon as Sun made Java proprietary-with-community-input, all that went out the window.
As for VB, you are confusing the language and its environment. VB-the-language is pretty awful--some of Microsoft's main VB hackers have left the company over that. What has made VB succeed is the programming environment and the copious documentation, training, and marketing. VB-the-language succeeded because a big company kept spending on making it succeed despite the pretty awful nature of the language itself.
So, "free" matters, because something that is "free" usually involves much more community input in its creation, and something that is "free" usually has to make it on its own merits. Java and VB are, and will remain, deeply flawed languages precisely because they are not free: as long as Microsoft and Sun can spend money to paper over the technical problems with those languages, and as long as the community itself doesn't have the final say in what happens to those languages independent of the corporate interests of their owners, those languages will remain sub-par.
(As for VB-like environments for Linux, Python, Glade, and your favorite Python IDE give you equivalent functionality and a better language. What they don't give you is the documentation, standardization, and name recognition of VB.)
NFS is just a protocol for clients accessing servers. Many forms of replication and migration can be supported even with just NFSv2, purely on the server side. And NFSv4 has direct support for these features (and I'm not at all convinced that adding all that junk to NFS was a good idea either).
Overall, basically, you are so excited about a bunch of gimmicks that you completely overlook the day-to-day reality of AFS: poor large file support, poor large volume support, exceptionally poor implementation of UNIX file system semantics, and poor handling of error conditions.
(And migrating users duing the day is just silly, even if your network file system nominally supports it.)
Yes, and that is one of the problems with AFS. A network file system for UNIX should foremost try to be as compatible as possible with UNIX, not to invent a completely different set of behaviors.
It's a reasonable choice for mine (a fairly small [approx 40-user] software company with lots of servers and few Windows clients -- and lots of potential for need to scale).
Sounds like your needs are really modest--just about anything would probably work for you.
NFS may be capable of scaling up to my network -- but it sure as hell couldn't scale up to IBM's.
What is "scaling up to IBM's network" supposed to mean? Sure, unlike NFSv2, AFS has reasonable support for security and user ids, but that's been fixed in current versions of NFS, and it doesn't really work all that well in practice in AFS either anyway. In terms of performance, AFS is usually worse. And the much touted global name space for AFS amounts to little more than "/net/host/..." under NFS. What kills AFS is that it has a steep learning curve, is inflexible, and, most of all, doesn't even try to support UNIX semantics.
That's what NIS is for. Furthermore, the flexibility of being able to set up machines with different views of the network is crucial in many applications. None of my workstations or servers actually have the same mount tables: they all get some stuff via NIS, and some stuff is modified locally. The restrictions AFS imposes are just unacceptable.
AFS makes administration tremendously easier after one's scaled the initial learning curve.
AFS is an administrative nightmare. Apart from the mess that ACLs cause and the problems of trying to fit real-world file sharing semantics into AFS's straightjacket, just the number of wedged machines due to overfull caches and its complete disregard for UNIX file system semantics cause no end of support hassles. Then, there is the poor support for Windows clients. We started out using AFS because it sounded good on paper, but it was a disaster in terms of support, and we got rid of it again after several years of suffering.
It performs far, far better than NFS on large networks (and merely somewhat better on smaller ones).
AFS's caching scheme works better than what NFS is doing for small files, but that case is fast and easy anyway. AFS's approach falls apart for just the kind of usage where it would matter most: huge files accessed from many machines.
Both NFS and AFS have very serious problems. But between the two, NFS is far simpler than AFS, is easier to administer in complex real-world environments, respects UNIX file system semantics better, and works better with large files. I can guardedly recommend NFS or SMB ("there is nothing better around, so you might as well use it"), but I can't imagine any environment for which AFS is a reasonable choice anymore. The only thing AFS had ever going for it as far as I'm concerned is that it was fairly secure at a time when NFS had no security whatsoever, but that is not an issue anymore.
In the medium term, however, I think WebDAV will become a better option, because it can be served and accessed with standard web servers and clients, in addition to being mappable onto the file system.
The Linux kernel already has WebDAV support (CODA hooks plus some user-mode process), although I'm not sure how well it works.
Of course it is. It gives you a single, unified view of a file system tree that can span many machines.
It doesn't support any: replication, disconnection, sharing, distribution.
Sure it does. Some of that functionality requires more than a vanilla NFS server, but that can be transparent to clients.
It is centralised, requires the same user names|numberpace and security.
Older versions did, current versions don't.
Don't get me wrong, NFS has lots of problems in many environments. But for networking a handful of machines in a home environment, it is nearly perfect. In fact, NFS version 2 is probably the best choice (the old one without all the security stuff). Furthermore, the alternatives (AFS, SMB, CODA, etc.) are harder to administer, perform worse, and have lots of problems with UNIX file system semantics.
Research versions of UNIX were based on bits and pieces from BSD, but they involved removing a lot of functionality, so by looking at the documentation as well as their follow-on, Plan 9, you can get a pretty good idea of what they considered good and bad.
Based on conversations I have had with the Bell Labs folks over Plan 9, I suspect that they probably wouldn't want to take responsibility for the OS X kernel.
Generally, Gigabit Ethernet is more flexible, easier to maintain, and has more third party hardware available for it, but if you have a motherboard with FW and are setting up a special-purpose, low-cost cluster, IEEE1394 or USB2 networking may be a reasonable choice.
Some of the Sharp laptops (though not the ultra-small MM10) and the Sony PictureBook seem to be alternatives and have also been reported to run Linux.
If you have unearthed credible reports of Linux installations on the MM10, please share them with the rest of us. Until then, the MM10 should probably be considered a "Windows-only" machine.
Another interesting question about the machine that has, so far, been unanswered, is whether its use as a dockable disk-drive is OS dependent. That is, is the USB mass storage implemented by hardware that is separate from the main processor, or is it implemented in the BIOS, or is it implemented by some piece of code running under Windows?
As for the software, OS X is not Linux; there is far less UNIX software available for OS X than for Linux and porting software to it can be a pain. PPC Linux works surprisingly well, but it still is second rate compared to the x86 versions of Linux.
Buy an iBook if you like OS X and want to run OS X applications. For anybody looking for a Linux laptop, there are better choices out there.
In any case, I kind of don't see the point. Why not just a a wireless repeater? With 802.11a/g that's a lot faster than this, and it means you don't have two separate networking technologies to hassle with. And usually, you want to cover an area completely anyway, so all wireless access points need to overlap, which mean that they can act as repeaters.
Apple didn't invent color management or color science either. They simply happened to have a large graphics arts user community, so they were the first to incorporate this stuff into their systems.
Microsoft, Apple, and Linux are each years behind the state of the art in many areas. Windows XP is some VMS work-alike on steroids, and OS X is a warmed-over version of NeXTStep. And Linux still gives you that warm-and-fuzzy UNIX feeling from, oh, 15 years ago. That's the way it is with commercial or real-world systems. Just because Apple happens to incorporate some feature into their system first doesn't give them claim to it in perpetuity.
Wrong. The Apple Lisa had a pre-emptive multi-tasking OS with protected memory, but the hardware cost too much (the Motorola 68k in particular had a paging bug at the time that required them to use their own MMU). Yes, and the Lisa itself was a clone of a similarity priced Xerox machine, which also failed. It's not like Apple invented the technology behind either the Lisa or the Mac. The Mac won in the marketplace over the Lisa, therefore it can be argued that co-operative multi-tasking and a simple memory model are better. And Windows won in the marketplace over Apple. It can therefore, by your reasoning, be argued that Windows is better. Right?
Letters that are that important should be sent by registered mail.
to wreak havoc on the postal system for political reasons,
Provided the US government isn't subsidizing junk mail (if they are, they should stop), every piece of junk mail that is sent makes the USPS a small profit. Well, then let them "wreak" away.
or even worse, to serve as a diversion for a terrorist act, such as the mailing of a contaminated letter.'
I somehow have a hard time seeing how this is a serious risk, over and above the general risk of "contaminated letters".
Remember that security consultants and "experts", like politicians, have a tendency to create unnecessary fear in order to hype up their own importance.
Repeating made up numbers doesn't make them true. The fact is that nobody knows how many people are using Linux. It actually seems likely to me that there are more Linux desktop users out there than there are OS X users (many of them dual-booting Windows).
In any case, no matter what the market share is, Apple needs to realize that Linux users can cause them real problems: as this shows, the Linux community is capable of developing codecs themselves if Apple doesn't supply them. In the long run, that erodes Apple's control over a market segment that they desperately want to control. Linux MPEG-4 encoders are already among the best and fastest in the business, for example.
Of course, that's what I would like to see happen anyway: so, keep up the good work--keep NOT shipping proprietary codecs for Linux, we don't want them anyway.
Any system whose growth rate is bounded away from zero grows exponentially.
And The Economist's useage of exponential within its economic context is entirely correct.
Of course it is. What isn't correct is the Economist's use of the term "not exponential" or its equivalents.
You seem to be a bit hazy on basic mathematics as well. You aren't an economist yourself by any chance?
Most mature industries grow exponentially because they grow with the economy as a whole. Any economist that would propose that the economy not grow exponentially would be lynched by politicians--our whole economic system assumes that we can sustain at least modest exponential growth indefinitely (whether that is reasonable is a different question).
The IT industry will continue to grow exponentially, just like almost any other sector . What it won't do is have growth rates that exceed that of the economy as a whole. That is probably what the Economist means.
The Economist just lost a lot of credibility in my eyes: misusing the term "exponential" in this way is something that just shouldn't get past the editors of any publication that claims any competence in economics.
Yes, Microsoft gives you consistently gray-and-blue dialog boxes. Lots of them. You can click around in them for hours trying to get anything done. But it's consistent.
and while the people exist who can administer those, the configuration for say, Apache, is wildly different to just about anything else, and anything else from each other. Just an observation.
And you think that the configuration of different Windows utilities is consistent just because you see them all through dialog boxes? Underneath that "consistent" veneer of Windows is the same variety of configuration semantics as exists for Linux. But, hey, it appears consistent.
The 2nd point is support. It's impeccable, and having guaranteed 24-hour help for those times when things foo bar up so badly we can't repair things is essential to running a service for our clients.
Support costs money. If you pay enough for Windows support, you get "impeccable" Windows support. If you pay enough for Linux support, you get "impeccable" Linux support. However, if you are any reasonably sized business, you are big enough to have enough Linux expertise in-house so that you don't need any outside support.
I won't comment on their business practices, but suffice it to say that's their choice.
I won't comment on your business practices. Suffice it to say that there are a lot of companies like yours that have way too much money and don't know what they are doing.
You get lots of nano-particles into your lungs every day, much of it from Diesel engines and coal power plants. Before worrying about NRAM, we should worry about those first.
Wow, I have to upgrade the entire CPU. Like, I get a new motherboard with 128Gbytes of nanotube memory, but I can't put bits and pieces of my old Opteron on it. Perfectly shocking. What was AMD thinking.
Palm is just like Microsoft, only that their software is even worse. As far as I'm concerned, Palm isn't cool, and my m505, which I bought a few years ago, is probably the last Palm handheld I will ever own.
You didn't understand the question. Yes, Sun has added generics to the language and compiler. But without changes to the JVM, Java generics are not type-safe across compilation boundaries.
Scientific and numerical computing is not general purpose.
No, and server-side computing, desktop application development, graphics, or any other specific application isn't "general purpose computing" either. What makes a language "general purpose" is that it has acceptable support for all those application areas, even if it doesn't excel at any of them.
In your market you need absolute controll over your app and java certainly does not allow this.
It is folly to think of numerical computing as being a separate "market". Do you want E-mail filtering? Do you want text-based retrieval? Do you want to browse images? Do you want to mine databases for patterns? Do you want to analyze business data? Do you want to edit video? Do you want nicely laid out text? Do you want great looking outline fonts? Do you want good process scheduling? Do you want good intrusion detection? Do you want smart opponents in games? Good, modern implementations of all of those have strong numerical components, in addition to requiring good UIs and other facilities. Java makes it a lot of unnecessary work right now to do a decent job at implementing their numerical components.
Why do you want to switch? Why not put the audio into an analog mixer and have all of them active simultaneously?
You are saying that Sun is extending the JVM to support generics in 1.5? Ummm--where exactly does it say that?
Value classes are still not there and I agree with the notion that operator overloading is evil and wish it was never invented. Its only purpose is to produce very bad code and create debugging nightmares. You can cast values to different data types when necessary. It beats using overloaded operators depending on value type because it can be confusing to other people looking at your code through a debugger. This is not the early 80's anymore where ram is a premium. Operator overloading was added when c++ programes were measured in kb's.
The applications we write barely fit into several Gigabytes of memory and they involve complicated mathematical formulas.
Basically, your attitude is typical, and it's ignorant: "I don't need it, so it must be useless". Fine, but as a result, Java is simply unsuitable for an increasing range of applications. And the fact that Java's C++ interface is so inefficient and cumbersome means that Java doesn't have an easy workaround either.
C# looks like a pretty general purpose, modern programming language, but Java clearly is not because it is not a good language for scientific and numerical computing, and because it does not interface well with existing C/C++ code.