I concede that ARIA doesn't have the same clout that the RIAA has, but the story doesn't just end there.
As has already been mentioned, ARIA and the RIAA have essentially the same members. I think we all agree that the RIAA has a *lot* of political clout in America (too much, but then, so do most commercial industries in the US). Ruddock (the Australian Attorney General) is on the record as saying that the Australian government wants to make its copyright and IP laws as compatible as possible with America.
It therefore seems reasonable to assume that if a company can't get its own way in Australia through the ARIA, they can apply pressure through the RIAA to make it happen anyway. It just takes longer.
From the article: The number of Web sites using PHP has risen to 23 million today from zero in 2000, according to surveys by the Internet analysis firm Netcraft.
Where does Netcraft get these figures? I decided to learn PHP in 1998 because every second site I visited was serving up PHP pages. Large corporations were only just starting to use PHP back then, but this is a far cry over the "zero" quoted above. Looks like a convenient exaggeration.
OTOH, I never saw Java-backed web pages as being too popular. They've certainly diminished, but were there ever that many? I've never seen Java and LAMP as being in serious contest with each other.
While.Net is certainly eating into some of Java's space, it is really only getting seriously deployed on Windows, which is limiting its penetration (particularly on servers).
I'm more than happy to use any system, so I don't think I'm too biased. But my reading of this article is as a manipulative marketing exercise. It's a shame it's getting so much press.
There are very few problems in classical computing that work efficiently with Quantum Computing. That's because Quantum Computing is very good at problems that are NP complete, while classical computing avoids NP complete problems (because they can't be done is a reasonable amount of time).
So computer scientists rarely come up with applications that are appropriate for Quantum Computing. But that doesn't mean that these applications do not exist! Calculating protein folding and factoring large numbers are just two appropriate examples.
The best general purpose Quantum Computing algorithm is Lov Grover's database search algorithm. This can be applied to most problems, and is provably the most efficient implementation for many situations.
The article seems misleading in its wording. It says "read the value of a qubit without changing its value." This can't mean that it doesn't change the original quantum value, as this makes no sense. Reading a quantum value (a qubit) collapses the probability to the value read, by definition. This means that the value is no longer quantum. The original probability function cannot be read (though it can be calculated).
The statement without changing its value must refer to reading the value reliably. When reading the state of an individual subatomic particle it is extremely easy to have the result perturbed by noise. Given that there is a probability of reading an alternative value, then it is not normally possible to tell when the wrong value was read. It appears that this makes the process much more reliable.
IAAQP (I Am A Quantum Physicist). Though I could still learn a thing or two about subatomic physics.
I'm Australian, and officially we use British spelling. That means we use "gaol" instead of "jail", "colour" instead of "color", and (mostly) use the letter "s" in words where America uses "z". "license" is the verb, and "licence" is the noun, while America just has "license". But I write code (and documentation) for an exclusively American market. So I have to change to American standards throughout all of my work.
But what should I do when I blog about my work? It's a personal thing for me, and read by people in Australia, Europe and North-America. Do I talk about "synchronisation primitives"? Do I "serialise" data? Should I use a GPL "licence"? Or should I be using the American standards?
Idioms are another issue. They tend to migrate from America to Australia, but it sounds strange to adopt it too early. "Should of" is just one example of language changing. 10 years ago, no one in Australia said "Cool". Now we all do. 50 years ago "got" meant "received" and nothing else. Now it is used as a superfluous adjunct to "have" (I've got to speak correctly). Which of the changes in language are appropriate to use?
Maybe my blog shouldn't matter, but it turns out that most of my clients read it, and it has led to several offers of work. I don't want to appear illiterate in it!
Larry seems to be saying that Open Source and commercial interests are at odds with one another, and it's only when a company tries to see beyond its profits that it participates in a compromise with OS software. But that's not what OS is about. Just ask RMS. His original intent was for companies to make all software free (as in "open"). The volunteer world of OS sort of came out of that idea, but I don't recall any histories describing volunteer work as a deliberate goal of GNU.
I also take personal exception to the lack of innovation comment. I'm doing postgraduate research in an area where no acceptable solutions exist... and everything I'm doing is open source. If I can't prove that I'm creating something innovative then I'll won't be awarded my degree. So far my university believes that I'm track.
Sure, innovation is hard, which is why we don't see a lot of it, either commercially or in the volunteer OS world. But just because we don't see it much is no reason to claim that it does not exist.
The original question suggested mentioning people in your own company who push to open source a project against all opposition. Andrew is passionately devoted to open source, and without him the Kowari RDF database would never have been released, or become the project that it is.
Lots of us write open source code, but it takes people who'll lay it on the line like Andrew did to make a real difference.
Last year we took on a Windows programming contract, so I went ahead and bought an MSDN subscription. In order to log into the online stuff I needed a.Net passport, and this required an email address.
The address I gave had been around for 3 years and had never received more than a couple of spam messages a week. Within 24 hours of getting the.Net passport that email address was getting over 20 spams a day, and it has grown significantly since then. (Thank goodness it wasn't my primary email account!)
Conclusion: either the passport user list is being sold, or security is nonexistent. Either way this is not a system anyone sane person would subscribe to!
Is this the incentive for the US to change it's currency? Most countries change their notes eventually anyway, so maybe America should consider doing it sooner rather than later.
The UK has that fancy bit of shiny foil woven into the paper that is easy to spot, and Australia uses polymer notes with transparent windows in them (these last longer than paper too). There are lots of alternatives available that a simple printer could not copy.
OTOH, as Bruce Schneier pointed out in Secrets and Lies, sometimes the cost of addressing the problem is more than the cost that the problem causes you.:-)
That's because it is. You're looking mostly at the contents of Michi's slides (with a few notes from Glynn for good measure). Michi spends several minutes talking during each slide, filling in a lot of details which were missed here.
> Having said that, I disagree with his blanket argument.
It only looks like a blanket argument because you're seeing everything in point form. Michi points out that most users don't know or care about source code... and he's right. Sure, you and I care about source code, but we aren't the controlling influences in the market. But even though I read and modify source code I had to confess that I've never done a security audit of any network server I rely on. Michi was trying to point out that we say Open Source is a wonderful thing, yet not many people take advantage of any of the benefits we all talk about.
I'm really pleased that these notes are up, but they really don't do the talk justice. It was bl**dy funny.:-)
I injured my lower back a few years ago, so the hours I spend on a keyboard really take their toll. I attend a chiropractor, and he offers massive relief, but it's not perfect.
However, if I swim regularly all the pain disappears. Recently I'd been working on a large project, with long hours, plus I caught a nasty cold, so I stopped swimming for about 3 weeks. Bingo. Pain came back - had to go see a chiropractor. Started swimming again. Within a week the pain is gone.
So yeah. I'd recommend swimming for anyone with back pain. Besides, it feels great to be fit.:-)
Scientific American isn't a truely in-depth journal for science. It provides a fairly high overview of the material it presents. As others have suggested, Nature is much better for this purpose.
As for previous articles on this, IEEE Spectrum ran a story on this in January. If you have an IEEE account (or know someone who does) then you can see it here. This is a short "news in brief" style of article, but it still does a reasonable job of explaining the effect. The thrust of the article was actually about the potential use of this effect in semiconductor lithography (used for printing ICs).
Spectrum also references the original paper on this effect, which appeared in Physical Review Letters last October. This paper was written by John Pendry from the Blackett Laboratory at Imperial College, London, UK. His work was preceded by David Smith and Sheldon Schultz at the University of California when they built some of this so-called left-handed material using a "metamaterial". In fact, the theoretical background for left-handed material has been around since 1968 when the Russian physicist Victor Veselago first looked at it.
So this material has been around for a little while now. You just have to know the right places to look.:-)
It may not be of interest to those in the US, but I thought I could offer a different perspective.
In Australia we have only 2 possible cable providers, and they are both capped. ADSL is also capped, but it is only being offered by one provider. This is provided for about $75AUD/month (about $36USD at the moment, but it was more like $50USD a few months ago).
I'm not sure of ADSL, but the cap on my cable is 512k (it was lower, but it was recently raised. I believe it was to make them more competitive). Ironically, this cap is not a huge deal because the majority of content is from overseas, and the bottlenecks are rarely in the cable. It's annoying when I'm accessing a local mirror though.
I believe the justification for the cap is because localities all share bandwidth on the cable. You don't want a few individuals sucking up all the bandwidth. This hasn't been an issue though, since cable and ADSL are not really marketed at all in Australia, so nobody I know seems to know about them (unless they are technically oriented).
If you want to do it optically, then you really need to combine the optical signal directly. We currently have no technology which would allow us to transport the phase of photons at visible wavelengths over that kind of distance. Even in the Keck observatory the optical systems are tricky enough, and that signal gets to stay in the same building!
Radio interferometry is much more possible, but still difficult. The bandwidth required to link 2 completely seperate observatories in real time is not really available (though theoretically possible - IIRC someone may have even done it), but to do it with multiple observatories is really out of our league. One inventive idea was to record an analogue radio signal along with precise timestamps onto high bandwidth tapes. The tapes were then brought together and interferometry performed on the signals offline with analogue circuits. This was very difficult (especially with the timing part) but it apparently worked.
I'm hoping that in the future, high speed digital data aquisition from multiple sites using RSFQ circuits might allow digital processing of these signals offline.
We haven't yet been able to see a planet around another star. We believe these planets exist because we see the star's brightness change and because we see a slight wobble in the star's position. However, we can't say we absolutely know there is a planet there unless we see it (we're just 99% sure). These telescopes will hopefully allow us to confirm their existence.
Also, the above methods are very difficult and time consuming (ie. expensive). Observing planets directly will be a lot faster, and allow our other astronomical resources to concentrate on other things. So not only do we get to see things we otherwise couldn't have seen, we actually get to save time and money for other observatories.
So whether you agree with the reasons or not, there really was a point to building it.:-)
You typically want more than one baseline in radio interferometry so you can determine (or even steer) the direction by the phase difference. Fortunately they already know the direction because each telescope is directional.
An array would probably help, but the surface of each telescope mirror acts a little like an array all on it's own. Since each telescope is individually one of the largest in the world then they get a very effective array without needing lots and lots of telescopes. I think that a couple of large mirrors will be more effective that lots of small ones. They would also make it easier to combine the optics.
OTOH, the more the merrier! Maybe someday a much larger array will be built which has individual scopes this large.
Well I'm about to start my postgraduate study in this area, so I can't say that I'm an expert yet... but I have to say, that's a darn good idea! If you don't mind, I might even do some research into this.
Being a computer engineer I've had the opportunity to build physical devices, though these days I program for a living (for some reason people think that if you have the word "computer" in your degree then you know how to program. Go figure. I won't complain - it pays better;-)
I'd have to say that writing a well crafted piece of software is extraordinarily satisfying, but it doesn't come close to the satisfaction of having built a physical device. Unfortunately, physical devices are more expensive and time consuming to build. The way I see it, hardware design is the same as programming, only in a different medium. The advantage is the device you can show others when you've finished.
Of course, probably the most satisfying thing I've ever done was to build a device based on a microprocessor, and to then write all the software that ran on it.:-)
This could have important ramifications for quantum computing. The biggest obstacle to quantum computing is decoherence as the particles encoding the quantum state interact with nearby particles. At temperatures this low the interaction is much, much less.
Maybe we'll be able to build quantum computers sooner than we thought?
I wouldn't rule out interpreted languages, because you're not going to be doing any "heavy lifting" in them anyway.
This is a really good point.
As an example, our local university is doing research in quantum computing, and is using a large beowulf cluster for simulations. All the libraries they use are written in C/Assembler, but the tasks they write are all in Python. This combination gives them fast execution, but the programs are fast and easy to code.
OK, for raw speed I'm going to keep clear of scripted languages and leave that to others...
Yes, Java can be very fast, especially with some of the latest Hotspot JVMs. In fact Java can be faster than C++.
C is generally much faster than Java. C++ can be just as fast, providing you don't use certain constructs, such as virtual functions. If you do then you will be slower than Java on Hotspot.
Of course Assembler is the fastest if you concentrate on the coding (as others have pointed out). However, this becomes quickly infeasible if a group of programmers are working on the project. A good compiler is MUCH better to use. Hence you should just use assembler for bottlenecks (meaning that you're pretty much stuck with C/C++. Let's not get into JNI).
As for the framework, others have pointed out the cost of CGI. mod_almostanything will probably be faster than the tightest assembler (depending on the ratio of hits to the complexity of the task). For sheer invocation speed, you can't go past modules. That usually means Java or a scripting language, but there's nothing to prevent you from writing the program as a binary module.
So it's an engineering decision between trade-offs.
If the program is long and complex enough then the cost of CGI is probably small. If instead, it is a simple program then modules will allow for more hits.
If you want a module then is the time and complexity of writing a binary module in C or C++ worth the speed advantage provided over scripted or Java modules?
If you go with a module framework then will Java really provide the speed increase over mod_perl, mod_php, etc, that you want? Some tests here are in order.
If you use CGI (because the application is complex enough that invocation cost is negligible) then will the OO advantages of Java and C++ make it worthwhile for the team to program in these languages? Otherwise C is the fastest option (with assembler tuning of bottlenecks).
If using CGI and you choose Java or C++, then are you after code-reuse, etc? If so then you're possibly going to use the C++ features that make it slower than Java. Alternatively, C++ also allows for easy Assembler optimisation of bottlenecks.
If speed outweighs everything, including complexity, development time, cross-platform issues, and cost, then I recommend building your own Apache module in C and optimising bottlenecks in assembler. If complexity, development time, corss-platform, etc, are still important, then I'd probably go with Java with a fast module, but I'd do some performance tests first.
All of these languages are great (including the scripting languages I haven't mentioned), but they need to be used where they are most appropriate.
IBM have a product that they used to call TopPage and is now called WebSphere. It may not provide all the professional functionality that you need, but it would be worth a look anyway. I tried it on some pages a few months ago, and it was quite useable (though I'm no web designer, and I prefer raw html).
They've ported it from Windows using Wine. Unfortunately, to guarantee that the libraries will remain consistent they needed to distribute their own wine libraries (Doh!). They've also chosen to distribute as RPMs. Not really a big deal, but it can be annoying for those of us who use Debian.
The downside is that it costs $69 (which doesn't seem too bad to me) and of course it is closed source (but then, so are your other GUI options). There's a 60 day trial available for download so you can see if it is suitable before buying it.
Many companies don't want to know anything about Linux (even when their employees have seen the light), so they would see this as an unnecessary expense.
I agree that it would be a trivial expense. It would even be cheaper to just buy the thing than to pay for the time taken for a developer to argue the case! But management and bean-counters don't seem to view the world rationally.:-)
Slashdot Mirror
I concede that ARIA doesn't have the same clout that the RIAA has, but the story doesn't just end there.
As has already been mentioned, ARIA and the RIAA have essentially the same members. I think we all agree that the RIAA has a *lot* of political clout in America (too much, but then, so do most commercial industries in the US). Ruddock (the Australian Attorney General) is on the record as saying that the Australian government wants to make its copyright and IP laws as compatible as possible with America.
It therefore seems reasonable to assume that if a company can't get its own way in Australia through the ARIA, they can apply pressure through the RIAA to make it happen anyway. It just takes longer.
From the article: The number of Web sites using PHP has risen to 23 million today from zero in 2000, according to surveys by the Internet analysis firm Netcraft.
Where does Netcraft get these figures? I decided to learn PHP in 1998 because every second site I visited was serving up PHP pages. Large corporations were only just starting to use PHP back then, but this is a far cry over the "zero" quoted above. Looks like a convenient exaggeration.
OTOH, I never saw Java-backed web pages as being too popular. They've certainly diminished, but were there ever that many? I've never seen Java and LAMP as being in serious contest with each other.
While .Net is certainly eating into some of Java's space, it is really only getting seriously deployed on Windows, which is limiting its penetration (particularly on servers).
I'm more than happy to use any system, so I don't think I'm too biased. But my reading of this article is as a manipulative marketing exercise. It's a shame it's getting so much press.
So computer scientists rarely come up with applications that are appropriate for Quantum Computing. But that doesn't mean that these applications do not exist! Calculating protein folding and factoring large numbers are just two appropriate examples.
The best general purpose Quantum Computing algorithm is Lov Grover's database search algorithm. This can be applied to most problems, and is provably the most efficient implementation for many situations.
The statement without changing its value must refer to reading the value reliably. When reading the state of an individual subatomic particle it is extremely easy to have the result perturbed by noise. Given that there is a probability of reading an alternative value, then it is not normally possible to tell when the wrong value was read. It appears that this makes the process much more reliable.
IAAQP (I Am A Quantum Physicist). Though I could still learn a thing or two about subatomic physics.
This is actually a day-to-day problem for me.
I'm Australian, and officially we use British spelling. That means we use "gaol" instead of "jail", "colour" instead of "color", and (mostly) use the letter "s" in words where America uses "z". "license" is the verb, and "licence" is the noun, while America just has "license". But I write code (and documentation) for an exclusively American market. So I have to change to American standards throughout all of my work.
But what should I do when I blog about my work? It's a personal thing for me, and read by people in Australia, Europe and North-America. Do I talk about "synchronisation primitives"? Do I "serialise" data? Should I use a GPL "licence"? Or should I be using the American standards?
Idioms are another issue. They tend to migrate from America to Australia, but it sounds strange to adopt it too early. "Should of" is just one example of language changing. 10 years ago, no one in Australia said "Cool". Now we all do. 50 years ago "got" meant "received" and nothing else. Now it is used as a superfluous adjunct to "have" (I've got to speak correctly). Which of the changes in language are appropriate to use?
Maybe my blog shouldn't matter, but it turns out that most of my clients read it, and it has led to several offers of work. I don't want to appear illiterate in it!
Larry seems to be saying that Open Source and commercial interests are at odds with one another, and it's only when a company tries to see beyond its profits that it participates in a compromise with OS software. But that's not what OS is about. Just ask RMS. His original intent was for companies to make all software free (as in "open"). The volunteer world of OS sort of came out of that idea, but I don't recall any histories describing volunteer work as a deliberate goal of GNU.
I also take personal exception to the lack of innovation comment. I'm doing postgraduate research in an area where no acceptable solutions exist... and everything I'm doing is open source. If I can't prove that I'm creating something innovative then I'll won't be awarded my degree. So far my university believes that I'm track.
Sure, innovation is hard, which is why we don't see a lot of it, either commercially or in the volunteer OS world. But just because we don't see it much is no reason to claim that it does not exist.
The original question suggested mentioning people in your own company who push to open source a project against all opposition. Andrew is passionately devoted to open source, and without him the Kowari RDF database would never have been released, or become the project that it is.
Lots of us write open source code, but it takes people who'll lay it on the line like Andrew did to make a real difference.
Last year we took on a Windows programming contract, so I went ahead and bought an MSDN subscription. In order to log into the online stuff I needed a .Net passport, and this required an email address.
The address I gave had been around for 3 years and had never received more than a couple of spam messages a week. Within 24 hours of getting the .Net passport that email address was getting over 20 spams a day, and it has grown significantly since then. (Thank goodness it wasn't my primary email account!)
Conclusion: either the passport user list is being sold, or security is nonexistent. Either way this is not a system anyone sane person would subscribe to!
Is this the incentive for the US to change it's currency? Most countries change their notes eventually anyway, so maybe America should consider doing it sooner rather than later.
The UK has that fancy bit of shiny foil woven into the paper that is easy to spot, and Australia uses polymer notes with transparent windows in them (these last longer than paper too). There are lots of alternatives available that a simple printer could not copy.
OTOH, as Bruce Schneier pointed out in Secrets and Lies, sometimes the cost of addressing the problem is more than the cost that the problem causes you. :-)
That's because it is. You're looking mostly at the contents of Michi's slides (with a few notes from Glynn for good measure). Michi spends several minutes talking during each slide, filling in a lot of details which were missed here.
> Having said that, I disagree with his blanket argument.
It only looks like a blanket argument because you're seeing everything in point form. Michi points out that most users don't know or care about source code... and he's right. Sure, you and I care about source code, but we aren't the controlling influences in the market. But even though I read and modify source code I had to confess that I've never done a security audit of any network server I rely on. Michi was trying to point out that we say Open Source is a wonderful thing, yet not many people take advantage of any of the benefits we all talk about.
I'm really pleased that these notes are up, but they really don't do the talk justice. It was bl**dy funny. :-)
However, if I swim regularly all the pain disappears. Recently I'd been working on a large project, with long hours, plus I caught a nasty cold, so I stopped swimming for about 3 weeks. Bingo. Pain came back - had to go see a chiropractor. Started swimming again. Within a week the pain is gone.
So yeah. I'd recommend swimming for anyone with back pain. Besides, it feels great to be fit. :-)
As for previous articles on this, IEEE Spectrum ran a story on this in January. If you have an IEEE account (or know someone who does) then you can see it here. This is a short "news in brief" style of article, but it still does a reasonable job of explaining the effect. The thrust of the article was actually about the potential use of this effect in semiconductor lithography (used for printing ICs).
Spectrum also references the original paper on this effect, which appeared in Physical Review Letters last October. This paper was written by John Pendry from the Blackett Laboratory at Imperial College, London, UK. His work was preceded by David Smith and Sheldon Schultz at the University of California when they built some of this so-called left-handed material using a "metamaterial". In fact, the theoretical background for left-handed material has been around since 1968 when the Russian physicist Victor Veselago first looked at it.
So this material has been around for a little while now. You just have to know the right places to look. :-)
In Australia we have only 2 possible cable providers, and they are both capped. ADSL is also capped, but it is only being offered by one provider. This is provided for about $75AUD/month (about $36USD at the moment, but it was more like $50USD a few months ago).
I'm not sure of ADSL, but the cap on my cable is 512k (it was lower, but it was recently raised. I believe it was to make them more competitive). Ironically, this cap is not a huge deal because the majority of content is from overseas, and the bottlenecks are rarely in the cable. It's annoying when I'm accessing a local mirror though.
I believe the justification for the cap is because localities all share bandwidth on the cable. You don't want a few individuals sucking up all the bandwidth. This hasn't been an issue though, since cable and ADSL are not really marketed at all in Australia, so nobody I know seems to know about them (unless they are technically oriented).
If you want to do it optically, then you really need to combine the optical signal directly. We currently have no technology which would allow us to transport the phase of photons at visible wavelengths over that kind of distance. Even in the Keck observatory the optical systems are tricky enough, and that signal gets to stay in the same building!
Radio interferometry is much more possible, but still difficult. The bandwidth required to link 2 completely seperate observatories in real time is not really available (though theoretically possible - IIRC someone may have even done it), but to do it with multiple observatories is really out of our league. One inventive idea was to record an analogue radio signal along with precise timestamps onto high bandwidth tapes. The tapes were then brought together and interferometry performed on the signals offline with analogue circuits. This was very difficult (especially with the timing part) but it apparently worked.
I'm hoping that in the future, high speed digital data aquisition from multiple sites using RSFQ circuits might allow digital processing of these signals offline.
Also, the above methods are very difficult and time consuming (ie. expensive). Observing planets directly will be a lot faster, and allow our other astronomical resources to concentrate on other things. So not only do we get to see things we otherwise couldn't have seen, we actually get to save time and money for other observatories.
So whether you agree with the reasons or not, there really was a point to building it. :-)
An array would probably help, but the surface of each telescope mirror acts a little like an array all on it's own. Since each telescope is individually one of the largest in the world then they get a very effective array without needing lots and lots of telescopes. I think that a couple of large mirrors will be more effective that lots of small ones. They would also make it easier to combine the optics.
OTOH, the more the merrier! Maybe someday a much larger array will be built which has individual scopes this large.
Well I'm about to start my postgraduate study in this area, so I can't say that I'm an expert yet... but I have to say, that's a darn good idea! If you don't mind, I might even do some research into this.
I'd have to say that writing a well crafted piece of software is extraordinarily satisfying, but it doesn't come close to the satisfaction of having built a physical device. Unfortunately, physical devices are more expensive and time consuming to build. The way I see it, hardware design is the same as programming, only in a different medium. The advantage is the device you can show others when you've finished.
Of course, probably the most satisfying thing I've ever done was to build a device based on a microprocessor, and to then write all the software that ran on it. :-)
Maybe we'll be able to build quantum computers sooner than we thought?
This is a really good point.
As an example, our local university is doing research in quantum computing, and is using a large beowulf cluster for simulations. All the libraries they use are written in C/Assembler, but the tasks they write are all in Python. This combination gives them fast execution, but the programs are fast and easy to code.
Yes, Java can be very fast, especially with some of the latest Hotspot JVMs. In fact Java can be faster than C++.
C is generally much faster than Java. C++ can be just as fast, providing you don't use certain constructs, such as virtual functions. If you do then you will be slower than Java on Hotspot.
Of course Assembler is the fastest if you concentrate on the coding (as others have pointed out). However, this becomes quickly infeasible if a group of programmers are working on the project. A good compiler is MUCH better to use. Hence you should just use assembler for bottlenecks (meaning that you're pretty much stuck with C/C++. Let's not get into JNI).
As for the framework, others have pointed out the cost of CGI. mod_almostanything will probably be faster than the tightest assembler (depending on the ratio of hits to the complexity of the task). For sheer invocation speed, you can't go past modules. That usually means Java or a scripting language, but there's nothing to prevent you from writing the program as a binary module.
So it's an engineering decision between trade-offs.
If speed outweighs everything, including complexity, development time, cross-platform issues, and cost, then I recommend building your own Apache module in C and optimising bottlenecks in assembler. If complexity, development time, corss-platform, etc, are still important, then I'd probably go with Java with a fast module, but I'd do some performance tests first.
All of these languages are great (including the scripting languages I haven't mentioned), but they need to be used where they are most appropriate.
Just my 2 bits worth.
They've ported it from Windows using Wine. Unfortunately, to guarantee that the libraries will remain consistent they needed to distribute their own wine libraries (Doh!). They've also chosen to distribute as RPMs. Not really a big deal, but it can be annoying for those of us who use Debian.
The downside is that it costs $69 (which doesn't seem too bad to me) and of course it is closed source (but then, so are your other GUI options). There's a 60 day trial available for download so you can see if it is suitable before buying it.
Many companies don't want to know anything about Linux (even when their employees have seen the light), so they would see this as an unnecessary expense.
:-)
I agree that it would be a trivial expense. It would even be cheaper to just buy the thing than to pay for the time taken for a developer to argue the case! But management and bean-counters don't seem to view the world rationally.
Ahhh yes, but how would they make the call? :-)
Exactly the kind of thing we want. Can you supply the details please?