The amount of code should be measured over the whole system. I doubt you could gain more than a few percentage points by including operator overloading in a language, even in math intensive programs.
Clear and consistent syntax is the whole reason + doesn't mean whatever you define it to be. To me a.add(b) is not any less clear than a + b, + a b or a b + which are all equivalent expressions.
Outside the domain of math heavy code, there is no good reason for operator overloading. Inside this domain you probably should be using other languages. If you are doing thousands of lines of codes with only formulas in a language like Java or C++ you are doing something wrong. Very few Java programs are actually math heavy so for most of these programs operator overloading can only decrease readability. Most of the math heavy programs have all the math isolated in a handfull of classes that are relatively stable.
Well change lanes earlier and the system probably won't kick in. Of course the whole point of this system is to smooth the whole driving experience so rather than creeping up to someones bumper, changing lanes at the last moment and accelerating to clear the only slightly slower moving traffic you either accept the (temporary) small decrease in speed, or change lanes and leisurely overtake the other car without your foot ever touching either brake or speed control.
At least it is time for something more efficient. More energy seems to go into heat and noise than into actual forward motion. It should be possible, using todays technology, to do this more efficiently.
CrazyObjectNumber c = a.clone(); c.multiply(b); c.add(53);
or (using static methods)
CrazyObjectNumber i = CrazyObjectNumber.multiply(a,b) CrazyObjectNumber c = CrazyObjectNumber.add(i, 53);
Nobody forces you to put things on one line in Java. Inlining happens automatically when it makes sense. Code completion and refactoring will help you avoid most of the typing.
http://bugzilla.mozilla.org/show_bug.cgi?id=217527 , according to this, it has been fixed. I'm running the release candidate now and everything looks ok. It never really bothered me anyway.
People just assumed that it wouldn't work. This assumption has now been proven wrong. I just watched the stream and it is pretty impressive for 32 bit/s. I kept an eye on memory usage and cpu usage. Memory usage is very acceptable (65 MB for a firefox nightly build + jvm + applet). It consumes about 40% cpu on my amd 2200+. The same applet in a freshly started IE is 46 MB. The difference is probably due to the fact that I had firefox open for the whole morning already (it tends to accumulate some weight over time).
Actually there is a lot of 'soft' censoring going on where mass media do not report or downplay facts that might harm political interests. Stuff gets labeled unpatriotic; foxnews estimates several tens of thousands demonstraters hit the streets in new york when other newssources are reporting several hundreds of thousands; CNN literally quoting some defense department monkey (50 terrorists have died, no mention of the dozens of civilians that were in the area). That sort of thing.
Frequently, the facts are picked up by mass media anyway after they've been exposed sufficiently by other media. But very often facts are succesfully hidden/misrepresented. Photos from dead US soldiers are rare. On a few occasions such photos made frontpage news but considering the amount of casualties there have actually been few of these reports. The US government discourages such reports and the media comply.
A disturbing recent trend is to label anything out of line with the republican party's vision as unpatriotic and liberal. The latter used to be a compliment but somehow the reality distortion field that covers the US nowadays has turned this into something evil. It's really amusing to watch the 'land of the free' become scared of 'liberal' opinions. The US is 'at war with terrorism' and anybody who says otherwise is a dangerous leftwing extremist.
Where does this notion that databases are slow come from? Databases are optimized for storing, finding and retrieving data as fast as possible without compromising the so called ACID properties. Filesystems are optimized to retrieve data that you already know the location of. These features are not mutually exclusive. However, implementing the features of one on top of the other is not necessarily an optimal solution (though it should be feasible).
That's why searching existing filesystems is slow, very limited and generally not worth the trouble beyond very simplistic queries. The search mechanisms on top of them are totally inadequate and slow. Similarly, databases are not very good at serving up large amounts of small files or at streaming big datablobs at a high bit/s ratio. That's why we still have hierarchical filesystems.
Hierarchical filesystems (even the ones that can store custom file attributes) do not exploit the enormous amount of meta data that is known (or can be extracted from) data. It would be enormously useful if I could update a version table with a database trigger whenever I hit the save button on some document. Trivial to implement on a dbfs, only feasible using hacks or custom tooling on a normal fs.
Using triggers, external datasources, manually entered data etc. you can associate very rich metadata with your data very easily. Obviously you don't want to that for system files (as the author of the article points out).
In fact, unlike traditional encyclopedias, wikipedia has both the manpower (all users, anyone who cares to contribute) and the room to provide exhaustive information.
Traditional encyclopedias are constrained in the amount of writers they can afford, the amount of research they can do and the amount of paper available for a single article. In a time where traditional encyclopedias are losing marketshare (thanks to internet, cdroms and other sources of information), cost savings are likely to put pressure on all three factors.
Wikipedia is a cumulative effort. If an article is not good, it can be fixed. If there are multiple views/interpretations on a topic, there is room to highlight both sides of the debate.
The longer the process goes on the better it becomes. Of course malicious people can insert information but sooner or later people will find out and fix it. You can put screening processes and peer reviews on wikipedia just like you can on a traditional encyclopedia.
Probably wikipedia's largest problem is not the process but the fact that it is accumulating information much faster than all other encyclopedias.
Now this guy has done something clever. He has made some small changes that would pass a first glance unless you already knew the facts. The problem is that he jumped to the wrong conclusion and never actually wondered how many people ever saw the changes. Since he only left the changes for 20 hours to max 5 days (!) and the articles do not exactly qualify as hot information, probably noone or at most a handful of people read the article. The changes obviously passed the vandalism procedures (for e.g. excessive changes in short periods of time) and nobody bothered to verify the information right away. The latter is actually the whole point of criticism. Wikipedia cannot be authorative because not all information is verified right away.
However, he misses the point. If brittanica has a mistake you might be tempted to write to the editor and maybe in a next edition it would be fixed. But most people probably don't. If you spot an error in wikipedia, you can just fix it. The more articles are referred to, the more authorative and informative they become. Especially the 'hot' articles on politics, famous people, etc are likely to be read, scrutinized and edited very often. Messing up 'empuries' is easy but try inserting false data under 'George Bush' and see how quickly that is corrected.
Actually that involves keeping up with the rest of the field as well. Not every feature MS adds to their OS should be duplicated. But some features are useful and should be considered.
MS has basically announced/demonstrated most of the new features that are in longhorn. Effectively that has given the linux community two years to come up with competing features. Adding database features to a filesystem makes sense, beos has demonstrated that you can do some nifty stuff with it and both apple and MS have anounced to do this.
The linux community however is divided. You can install reiserfs, maybe develop some tools that use some of its more advanced features but that doesn't fundamentally change anything if openoffice, KDE and Gnome and other programs don't coevolve to use the new features.
The same goes for stuff like avalon. While everybody is still talking about how such technology might be used in OSS projects like mozilla, Gnome, MS is well on their way of implementing something that may actually work.
Filesystems with rich metadata were already a good idea ten years ago. The OSS community has talked about them where others have implemented them. Two years of more talking would be fairly consistent. IMHO the OSS community is underperforming in picking up new technology and good ideas.
Traditional encyclopedias are also written by random people. The only guarantees you have about their quality comes from their reputation and the hope that the publisher won't cut to much cost on quality control. Traditional encyclopedias have another disadvantage: room for argumentation and literature references is constrained. Articles are kept short to make them fit in the dead tree version at a reasonable cost. Wikipedia has no such limitations.
Literature references can be added and as I understand are being added when appropriate. A good researcher would never depend on vague formulations in an encyclopedia anyway but either back them up with more references or more evidence.
Now when it comes to references, you can judge the quality of a scientific article by looking at the references. If it only includes some obscure references (and maybe a handfull of wikipedia references) the author probably didn't do his homework. This is the way I used to review articles when I was still in academia: read the abstract, skim through the reference list and then the article. Usually my opinion after reading the abstract was confirmed by the reference list and argued by reading the rest of the article (I usually stopped reading after a few pages if it was really bad).
Reviewers have the liberty and the obligation to lookup references if that is essential to the argumentation of an article. If some author would make some vague claim that is essential to whatever he is trying to argue and would point to wikipedia for more material that would be suspicious already. A reviewer should then at least look up the wikipedia article and review that.
Now unlike a traditional encyclopedia, both author and reviewer can also use their knowledge to improve the wikipedia article if it would need improvement. For instance a reviewer might actually agree with the wikipedia article but add some footnote with a reference to some article to strengthen its argument and then continue to slap the author (of the reviewed article, not the wikipedia article) on his wrist for not arguing his point properly.
Now citing wikipedia articles might be a bit more problematic because the wikipedia article might change over time. Basically you have no guarantee that the version of an article you look up is the same as the version that was cited. Version history is the solution to that.
Because you don't call votes this way only a randomly selected subset of boxes to confirm that the machine count matches the paper output. Only in case of a recount based on solid information that in fact something is wrong with the machines (e.g. because) you proceed to count all votes this way.
Consider it as a backup option, something you guys don't have now. Something worth paying for and something that maybe would have resulted in Al Gore beying a democratically elected president instead of George Bush being a court appointed one right now.
It's impossible to assess whether George Bush was democratically elected now because all the evidence was burned shortly after the courts put an end to the whole recount circus. Whatever your position on this matter is, I don't think either party would have protested the outcome of the elections without any evidence of vote machine failure to register votes properly.
BTW fair elections are fundamentally not a technical issue. Elections are bought not won these days. Without a big bag of money for advertising you don't stand a chance of winning the elections. In order to obtain the necessary money you need to be corrupt (i.e. promise to use your power in someones favour in return for money). So whomever wins the election is guaranteed to be corrupt. Given that there are only two parties to choose from (and some independent candidates who have no chance whatsoever to win), you basically have to choose who is the least corrupt candidate or the best candidate in spite of the corrupt tendencies rather than being able to vote on a person you actually trust. I never liked Al Gore and I despised Bush in 2000. If I were American (which I'm not) I would have voted Gore though. I don't like Kerry and Edwards either and Bush exceeded all my expectations in being far more corrupt and stupid than I believed him to be in 2000. So despite me not liking Kerry, I'd vote Kerry nevertheless because he can't possibly be worse than four more years of Bush.
MS stock value is based on the concept of their revenue growing exponentially. It stopped doing that a few years ago (it still grows though) and the stockvalue has not been corrected yet to reflect that new reality. As soon as investors figure out that the stock price is too high in relation to the actual marketvalue of the company, the stockprice will go down (because investors buy stock with the hope to sell it at a higher price later on).
By opening new markets and thus growing revenue and perhaps even profit margins, MS hopes to delay this as long as they can. For the past few years the perspective on these yet to realize profits has kept investors happy. However, so far their efforts to acually grow profit have not been very succesful and investors have started to take note of that. Some of their 'new' products barely break even, some are not breaking even and some bring in a (modest) profit. Basically all of the new productlines pale in comparison to their traditional moneymakers (office & windows) MS has compensated this by increasing the margins on their products (to the point where these are actually losing marketshare). Predictably this strategy runs out of steam quickly when it comes to increasing revenue. Also OSS has put an enormous pressure on their prices and has eroded their traditionally very strong developers community.
So MS sees that google is worth billions and MS feels that they are in a position where they can effectively compete with google and add some billions to their company value. Whether this is true remains to be seen but they certainly have the resources to invest in the necessary technology. Making their technology attractive for businesses is easy (just go below whatever google charges). However, the service is free for endusers so they will (also) need to compete on quality rather than price to actually gain marketshare over google.
Even if MS succeeds in pushing google out of the market, it is debatable whether MS will be able to delay the inevitable downgrade of their stockvalue much longer. Googles profits are very modest when compared to MS and the (expected) market value is based on yet to be realized revenue streams.
CD keys are a bit too effective. Most games have no online community to speak of because the number of players with a key and the game installed drops rapidly after a few months. This affects sales too because the lack of an online community becomes a reason not to buy the game pretty soon after its release. Games like quake 1 and 2 were largely driven by online gaming and continued to sell well long after the release. The same goes for the earlier versions of unreal & unreal tournament.
The problem with video on demand is that the infrastructure is not widely deployed. Most tvs don't support any of the digital tv standards so you need a settopbox. The situation is different from country to country with many local monopolists competing (i.e. you don't have much choice in selecting cable operators). The network quality is mediocre at best with unpredictable bandwidth, latency and availability. You need uninterrupted downstream bandwidth of at least a few megabits per second for good quality video.
I agree with Bill Gates that this is going to be different in ten years. By then most homes will have some form of broadband, mobile telephone networks will have been deployed that support broadband services. In other words, pretty much anywhere you go there will be some form of broadband that is good enough for high quality streaming video.
Then it is just a matter of offering the content and using the bandwidth. However, before that happens a number of legal issues will need to be resolved. Also there will need to be some standards (as in not owned by Microsoft or any other company). And finally the media companies will need to get involved. All that can happen in ten years but I don't see much happening yet.
The media companies are still clutching to their existing revenue streams. At the same time they seem to be only frustrating attempts to move beyond physisical media. It took a company like Apple to convince the whole industry that online content is a viable revenue stream and that was only this year. The same could happen for other content then audio in a few years.
Given open, widely supported standards and given the wide availability of networks this could happen. The latter is on schedule to being solved be 2014. However, ten years is a very short time to change an industry that depends on proprietary, closed standards.
We're talking about a list of specific issues that go beyond just biomedical research, signed by a number of scientists including some who have won a nobel prize. I'm normally sceptical too but this is simply too much too simply dismiss as propaganda. I'm well aware how the conservative right has dominated the political agenda the past few years and how it has effected society in numerous ways.
About point 2, this style of politics seems to have replaced reason and common sense in the US. Both parties know most people will never get a chance to examine evidence in detail so instead they depend on making lots of noise in the media.
But still, when a large group of respected, smart and well educated people supports these accusations I think that is more credible than the white house telling us everything is fine. Getting Nobel prize winners to support this means that a few very smart people made a balanced judgement and came to the conclusion that they wanted to support this.
It takes some enormous wisdom or stupidity to dismiss such a thing. I'm afraid there's plenty of stupid americans who will just do that. I'm as pessimistic to believe that the current US government is actually so stupid that they actually believe they know better.
Izotope ozone is a (non free) winamp/directx plugin that emulates some of the distortion effects that sixties amplifiers produce using tubes. I've been using it for quite some time and it really enhances the listening experience. I can recommend it and it sure is worth the small license fee (which is peanuts compared to what you would need to invest in hardware otherwise). I haven't found any other plugins that produce a similar improvement in sound. There are many plugins that just beef up the bass a bit or add cheap 3d effects. Izotope Ozone is in a different league.
The plugin clearly demonstrates that the distortions (when used with care) can really enhance music. It also demonstrates that you can get the same effect by processing the sound digitally instead of with tubes. Izotope ozone actually goes way beyond what traditional tubes can do because it doesn't have the physical limitations.
Of course most commercial rock and pop music is processed and filtered in the studio before it is put on cd whereas older music (or indie records) tend to sound better when played back on equipment that adds the distortion effects. Of course the amount of distortion is a matter of personal taste and I find that I enjoy my music more with a little bass compression and a bit of sparkle in the higher ranges. Studios tend to optimize for cheap equipment (i.e. it has to sound nice on cheap radios) so you can gain a lot by adding some distortions.
You can also use sound distortion to compensate for lossy compression or lousy speakers. Just boost the bass digitally for the frequency range that your subwoofer can actually handle; add a little sparkle to compensate for loss of higher frequencies during the mp3 compression; add some overdrive on a guitar track. Distortion is not necessarily about reproducing sound as it was when it was recorded but about making it sound as nice/pleasing as possible. Much of the distortion effects in sixties equipment is deliberate and not accidental. Electrical guitars are a good example of how distortion can be used to produce a wide range of sounds.
I don't know any desktop users who actually wait for stable debian releases. If you are using stable, you are using KDE 2.x or Gnome 1.x.
Most users trade stability for features because stable releases are obsolete the day they are released (this is intentional because nothing can possibly get into stable in less than a few months). Major packages like KDE or Gnome which have several minor releases per year are usually obsolete by the time they reach stable. This was really apparent with the last stable release where KDE 3 was already out and debian shipped with 2.x. Even testing is usually struggling to keep up with minor releases (which usually fix lots of bugs).
Most commercial debian releases are therefore not based on stable but on testing and unstable packages (obsolete is hard to sell in the desktop market). Stable is good for servers, not for desktops. On a server you get all the stable packages you need and maybe a handfull of handpicked non stable packages with things you really need.
It would be helpful though if the debian developers rethought their release strategy for desktop usage. I don't agree that stable needs to equate to obsolete (which apparently is their interpretation). Releasing more often would actually improve stability for most users because they would no longer have to rely on testing and unstable (which generally work out as the name suggests). I can symphatize with having a really stable version for server usage (so why not label it server). Testing is quite good most of the time but not always. So why not mark testing as stable three or four times a year? Use a few weeks to fix/remove the show stopping packages and rubber stamp it as stable.
I think the market is pretty much non existent outside the US. At least I'm not aware of any supported software by my bank (Abn Amro, one of the major banks in the Netherlands) other than their own web based software.
And judging from the review, we aren't missing much. I pay my bills online and haven't seen the inside of a bank for several years.
Tedious is standard with linux because of its packagemanagement tools which are incompatible with each other and even with earlier versions of the same tools. That is the main reason why most commercial software either supports only a very small subset of linux distributions (typically only specific versions of Red Hat) or bypass the packagemanagers entirely. That's not a Java problem. If you don't like tedious, use something else to run Java on. Linux is entirely optional. Of course if you are competent enough to get linux up and running you are more than qualified to follow the very small set of simple instructions to download and install the JRE. If you are still on a modem of course any download seems huge. But really, 10MB shouldn't be an issue any more these days, even if you are on a modem.
Of course if you download the jdk instead of just the runtime environment and an IDE you are looking at maybe 90 MB for the latest version of eclipse and another 40MB for the j2sdk.
Then the freedom issue. Yes, the SDK from SUN is not open source (though it is available for free for both commercial and non commercial use). However, open source JDKs and JREs are available (and are shipped with several linux distributions). If it makes you feel good, you can compile your gnu java with the gnu compiler and deploy it on your gnu server.
There is the small matter of the Java trademark which is indeed owned by Sun which and you may not stick on software without their permission (Just like the Mozilla foundation owns & controls the trademark for Mozilla). In practice this means that software that does not pass SUN's tests cannot be called Java. These tests are somewhat controversial and if that is what you meant by "not free" I don't disagree with you.
So unlike C, C++, packagemanagers, web browsers. The Java API is well defined (by SUN and several hundreds of major industrial and non industrial partners through the Java Community Process), and compliant open source and commercial implementations are widely available. Many of the reference implementations of J2EE APIs are open source (e.g. apache Tomcat is the reference implementation for the servlet specification). In short if you are looking for open source success stories, look no further than Java.
Finding a C++ compiler that complies 100% with the ANSI standards is nearly impossible (which is why porting applications between compilers is a major PITA) and why a minor compiler version increment requires you to update most of your linux software to versions that A) compile with the new version and B) have been compiled with the new version. 2.x to 3.x of the GNU compiler was very disruptive IMHO.
There are no standards for packagemanagers and the W3C publishes interesting, recommendations which nobody, including the intended audience of webbrowser developers, seems to be able to interpret in a consistent fashion.
Legally, the minister has voted on behalf of the dutch government. All actions of the government are ultimately controlled by the dutch parliament. Now, the parliament has decided that the minister should excercise a legal loophole and revoke the the vote.
Technically the minister can decide not to do this in which case a vote of no confidence may follow (after a debate with the minister on his reasons for not revoking his vote). However, given the political consequences of this scenario, this is unlikely to happen. A vote of no confidence would be rather damaging to all three of the coalition parties (given the recent polls) since it would likely result in new elections and result in a loss of the majority they are currently enjoying.
Given this context there are two scenarios. Either the parliament will look the other way if the minister ignores them (this will be settled behind closed doors and the opposition will make lots of noise) or the minister will revoke his vote. Given his earlier comments, the minister will favour the first option and will put lots of pressure on the three coalition parties to look the other way.
The only way to prevent this from happening is media attention. So far this has been pretty much a non issue in the media (few people understand what this debate is about). If this doesn't change, parliament won't work too hard on this issue. On the other hand if there is too much pressure on the minister, the coalition parties will likely back off to prevent a vote of no confidence.
The reason gets continues to be used is because it is in many textbooks and by default none of the IDEs, compilers, etc. seem to warn against its usage. Any security measure that requires education in order to work is fatally flawed. Security should be fool proof. Tools are your first line of defense and in many programming languages the only line of defense. If GCC doesn't tell you the code is flawed, in many cases you will never find out until it crashes.
Code that utilizes gets or similarly flawed legacy parts of C libraries should simply not compile without using a --insecure flag or something. If you want potentially insecure code to compile at least it should be a conscious choice. Of course gets isn't really the problem anymore (at least I'm sure that somebody has grepped sourcecode of e.g. sendmail for occurences of gets and buffer overflows continue to be a problem there).
Personally I find it amazing that buffer overflows still are an issue. The technology to prevent them has been around for decades. I never experience buffer overflows in Java. I do run into bugs in the JVM on very rare occasions (which is written in C). Also sometimes I discover memory or resource leaks in Java software (which eventually prevent continued operation of the software). Java is not immune to that sort of problems but buffer overflows are simply not an issue.
Considering that buffer overflows are the #1 security problem today, there is something to say for managed code environments. Personally, I believe that running non managed code on a server is irresponsible. The one certainty you have about software packages like sendmail, apache, IIS oracle, mysql, postgresql, openssh, etc. is that the latest buffer overflow won't be the last one to be found.
Slashdot Mirror
The amount of code should be measured over the whole system. I doubt you could gain more than a few percentage points by including operator overloading in a language, even in math intensive programs.
Clear and consistent syntax is the whole reason + doesn't mean whatever you define it to be. To me a.add(b) is not any less clear than a + b, + a b or a b + which are all equivalent expressions.
Outside the domain of math heavy code, there is no good reason for operator overloading. Inside this domain you probably should be using other languages. If you are doing thousands of lines of codes with only formulas in a language like Java or C++ you are doing something wrong. Very few Java programs are actually math heavy so for most of these programs operator overloading can only decrease readability. Most of the math heavy programs have all the math isolated in a handfull of classes that are relatively stable.
Well change lanes earlier and the system probably won't kick in. Of course the whole point of this system is to smooth the whole driving experience so rather than creeping up to someones bumper, changing lanes at the last moment and accelerating to clear the only slightly slower moving traffic you either accept the (temporary) small decrease in speed, or change lanes and leisurely overtake the other car without your foot ever touching either brake or speed control.
At least it is time for something more efficient. More energy seems to go into heat and noise than into actual forward motion. It should be possible, using todays technology, to do this more efficiently.
why not do
r c = CrazyObjectNumber.add(i, 53);
CrazyObjectNumber c = a.clone();
c.multiply(b);
c.add(53);
or (using static methods)
CrazyObjectNumber i = CrazyObjectNumber.multiply(a,b)
CrazyObjectNumbe
Nobody forces you to put things on one line in Java. Inlining happens automatically when it makes sense. Code completion and refactoring will help you avoid most of the typing.
http://bugzilla.mozilla.org/show_bug.cgi?id=217527 , according to this, it has been fixed. I'm running the release candidate now and everything looks ok. It never really bothered me anyway.
People just assumed that it wouldn't work. This assumption has now been proven wrong. I just watched the stream and it is pretty impressive for 32 bit/s. I kept an eye on memory usage and cpu usage. Memory usage is very acceptable (65 MB for a firefox nightly build + jvm + applet). It consumes about 40% cpu on my amd 2200+. The same applet in a freshly started IE is 46 MB. The difference is probably due to the fact that I had firefox open for the whole morning already (it tends to accumulate some weight over time).
Actually there is a lot of 'soft' censoring going on where mass media do not report or downplay facts that might harm political interests. Stuff gets labeled unpatriotic; foxnews estimates several tens of thousands demonstraters hit the streets in new york when other newssources are reporting several hundreds of thousands; CNN literally quoting some defense department monkey (50 terrorists have died, no mention of the dozens of civilians that were in the area). That sort of thing.
Frequently, the facts are picked up by mass media anyway after they've been exposed sufficiently by other media. But very often facts are succesfully hidden/misrepresented. Photos from dead US soldiers are rare. On a few occasions such photos made frontpage news but considering the amount of casualties there have actually been few of these reports. The US government discourages such reports and the media comply.
A disturbing recent trend is to label anything out of line with the republican party's vision as unpatriotic and liberal. The latter used to be a compliment but somehow the reality distortion field that covers the US nowadays has turned this into something evil. It's really amusing to watch the 'land of the free' become scared of 'liberal' opinions. The US is 'at war with terrorism' and anybody who says otherwise is a dangerous leftwing extremist.
Where does this notion that databases are slow come from? Databases are optimized for storing, finding and retrieving data as fast as possible without compromising the so called ACID properties. Filesystems are optimized to retrieve data that you already know the location of. These features are not mutually exclusive. However, implementing the features of one on top of the other is not necessarily an optimal solution (though it should be feasible).
That's why searching existing filesystems is slow, very limited and generally not worth the trouble beyond very simplistic queries. The search mechanisms on top of them are totally inadequate and slow. Similarly, databases are not very good at serving up large amounts of small files or at streaming big datablobs at a high bit/s ratio. That's why we still have hierarchical filesystems.
Hierarchical filesystems (even the ones that can store custom file attributes) do not exploit the enormous amount of meta data that is known (or can be extracted from) data. It would be enormously useful if I could update a version table with a database trigger whenever I hit the save button on some document.
Trivial to implement on a dbfs, only feasible using hacks or custom tooling on a normal fs.
Using triggers, external datasources, manually entered data etc. you can associate very rich metadata with your data very easily. Obviously you don't want to that for system files (as the author of the article points out).
In fact, unlike traditional encyclopedias, wikipedia has both the manpower (all users, anyone who cares to contribute) and the room to provide exhaustive information.
Traditional encyclopedias are constrained in the amount of writers they can afford, the amount of research they can do and the amount of paper available for a single article. In a time where traditional encyclopedias are losing marketshare (thanks to internet, cdroms and other sources of information), cost savings are likely to put pressure on all three factors.
Wikipedia is a cumulative effort. If an article is not good, it can be fixed. If there are multiple views/interpretations on a topic, there is room to highlight both sides of the debate.
The longer the process goes on the better it becomes. Of course malicious people can insert information but sooner or later people will find out and fix it. You can put screening processes and peer reviews on wikipedia just like you can on a traditional encyclopedia.
Probably wikipedia's largest problem is not the process but the fact that it is accumulating information much faster than all other encyclopedias.
Now this guy has done something clever. He has made some small changes that would pass a first glance unless you already knew the facts. The problem is that he jumped to the wrong conclusion and never actually wondered how many people ever saw the changes. Since he only left the changes for 20 hours to max 5 days (!) and the articles do not exactly qualify as hot information, probably noone or at most a handful of people read the article. The changes obviously passed the vandalism procedures (for e.g. excessive changes in short periods of time) and nobody bothered to verify the information right away. The latter is actually the whole point of criticism. Wikipedia cannot be authorative because not all information is verified right away.
However, he misses the point. If brittanica has a mistake you might be tempted to write to the editor and maybe in a next edition it would be fixed. But most people probably don't. If you spot an error in wikipedia, you can just fix it. The more articles are referred to, the more authorative and informative they become. Especially the 'hot' articles on politics, famous people, etc are likely to be read, scrutinized and edited very often. Messing up 'empuries' is easy but try inserting false data under 'George Bush' and see how quickly that is corrected.
Actually that involves keeping up with the rest of the field as well. Not every feature MS adds to their OS should be duplicated. But some features are useful and should be considered.
MS has basically announced/demonstrated most of the new features that are in longhorn. Effectively that has given the linux community two years to come up with competing features. Adding database features to a filesystem makes sense, beos has demonstrated that you can do some nifty stuff with it and both apple and MS have anounced to do this.
The linux community however is divided. You can install reiserfs, maybe develop some tools that use some of its more advanced features but that doesn't fundamentally change anything if openoffice, KDE and Gnome and other programs don't coevolve to use the new features.
The same goes for stuff like avalon. While everybody is still talking about how such technology might be used in OSS projects like mozilla, Gnome, MS is well on their way of implementing something that may actually work.
Filesystems with rich metadata were already a good idea ten years ago. The OSS community has talked about them where others have implemented them. Two years of more talking would be fairly consistent. IMHO the OSS community is underperforming in picking up new technology and good ideas.
Traditional encyclopedias are also written by random people. The only guarantees you have about their quality comes from their reputation and the hope that the publisher won't cut to much cost on quality control. Traditional encyclopedias have another disadvantage: room for argumentation and literature references is constrained. Articles are kept short to make them fit in the dead tree version at a reasonable cost. Wikipedia has no such limitations.
Literature references can be added and as I understand are being added when appropriate. A good researcher would never depend on vague formulations in an encyclopedia anyway but either back them up with more references or more evidence.
Now when it comes to references, you can judge the quality of a scientific article by looking at the references. If it only includes some obscure references (and maybe a handfull of wikipedia references) the author probably didn't do his homework. This is the way I used to review articles when I was still in academia: read the abstract, skim through the reference list and then the article. Usually my opinion after reading the abstract was confirmed by the reference list and argued by reading the rest of the article (I usually stopped reading after a few pages if it was really bad).
Reviewers have the liberty and the obligation to lookup references if that is essential to the argumentation of an article. If some author would make some vague claim that is essential to whatever he is trying to argue and would point to wikipedia for more material that would be suspicious already. A reviewer should then at least look up the wikipedia article and review that.
Now unlike a traditional encyclopedia, both author and reviewer can also use their knowledge to improve the wikipedia article if it would need improvement. For instance a reviewer might actually agree with the wikipedia article but add some footnote with a reference to some article to strengthen its argument and then continue to slap the author (of the reviewed article, not the wikipedia article) on his wrist for not arguing his point properly.
Now citing wikipedia articles might be a bit more problematic because the wikipedia article might change over time. Basically you have no guarantee that the version of an article you look up is the same as the version that was cited. Version history is the solution to that.
Because you don't call votes this way only a randomly selected subset of boxes to confirm that the machine count matches the paper output. Only in case of a recount based on solid information that in fact something is wrong with the machines (e.g. because) you proceed to count all votes this way.
Consider it as a backup option, something you guys don't have now. Something worth paying for and something that maybe would have resulted in Al Gore beying a democratically elected president instead of George Bush being a court appointed one right now.
It's impossible to assess whether George Bush was democratically elected now because all the evidence was burned shortly after the courts put an end to the whole recount circus. Whatever your position on this matter is, I don't think either party would have protested the outcome of the elections without any evidence of vote machine failure to register votes properly.
BTW fair elections are fundamentally not a technical issue. Elections are bought not won these days. Without a big bag of money for advertising you don't stand a chance of winning the elections. In order to obtain the necessary money you need to be corrupt (i.e. promise to use your power in someones favour in return for money). So whomever wins the election is guaranteed to be corrupt. Given that there are only two parties to choose from (and some independent candidates who have no chance whatsoever to win), you basically have to choose who is the least corrupt candidate or the best candidate in spite of the corrupt tendencies rather than being able to vote on a person you actually trust. I never liked Al Gore and I despised Bush in 2000. If I were American (which I'm not) I would have voted Gore though. I don't like Kerry and Edwards either and Bush exceeded all my expectations in being far more corrupt and stupid than I believed him to be in 2000. So despite me not liking Kerry, I'd vote Kerry nevertheless because he can't possibly be worse than four more years of Bush.
MS stock value is based on the concept of their revenue growing exponentially. It stopped doing that a few years ago (it still grows though) and the stockvalue has not been corrected yet to reflect that new reality. As soon as investors figure out that the stock price is too high in relation to the actual marketvalue of the company, the stockprice will go down (because investors buy stock with the hope to sell it at a higher price later on).
By opening new markets and thus growing revenue and perhaps even profit margins, MS hopes to delay this as long as they can. For the past few years the perspective on these yet to realize profits has kept investors happy. However, so far their efforts to acually grow profit have not been very succesful and investors have started to take note of that. Some of their 'new' products barely break even, some are not breaking even and some bring in a (modest) profit. Basically all of the new productlines pale in comparison to their traditional moneymakers (office & windows) MS has compensated this by increasing the margins on their products (to the point where these are actually losing marketshare). Predictably this strategy runs out of steam quickly when it comes to increasing revenue. Also OSS has put an enormous pressure on their prices and has eroded their traditionally very strong developers community.
So MS sees that google is worth billions and MS feels that they are in a position where they can effectively compete with google and add some billions to their company value. Whether this is true remains to be seen but they certainly have the resources to invest in the necessary technology. Making their technology attractive for businesses is easy (just go below whatever google charges). However, the service is free for endusers so they will (also) need to compete on quality rather than price to actually gain marketshare over google.
Even if MS succeeds in pushing google out of the market, it is debatable whether MS will be able to delay the inevitable downgrade of their stockvalue much longer. Googles profits are very modest when compared to MS and the (expected) market value is based on yet to be realized revenue streams.
CD keys are a bit too effective. Most games have no online community to speak of because the number of players with a key and the game installed drops rapidly after a few months. This affects sales too because the lack of an online community becomes a reason not to buy the game pretty soon after its release. Games like quake 1 and 2 were largely driven by online gaming and continued to sell well long after the release. The same goes for the earlier versions of unreal & unreal tournament.
The problem with video on demand is that the infrastructure is not widely deployed. Most tvs don't support any of the digital tv standards so you need a settopbox. The situation is different from country to country with many local monopolists competing (i.e. you don't have much choice in selecting cable operators). The network quality is mediocre at best with unpredictable bandwidth, latency and availability. You need uninterrupted downstream bandwidth of at least a few megabits per second for good quality video.
I agree with Bill Gates that this is going to be different in ten years. By then most homes will have some form of broadband, mobile telephone networks will have been deployed that support broadband services. In other words, pretty much anywhere you go there will be some form of broadband that is good enough for high quality streaming video.
Then it is just a matter of offering the content and using the bandwidth. However, before that happens a number of legal issues will need to be resolved. Also there will need to be some standards (as in not owned by Microsoft or any other company). And finally the media companies will need to get involved. All that can happen in ten years but I don't see much happening yet.
The media companies are still clutching to their existing revenue streams. At the same time they seem to be only frustrating attempts to move beyond physisical media. It took a company like Apple to convince the whole industry that online content is a viable revenue stream and that was only this year. The same could happen for other content then audio in a few years.
Given open, widely supported standards and given the wide availability of networks this could happen. The latter is on schedule to being solved be 2014. However, ten years is a very short time to change an industry that depends on proprietary, closed standards.
We're talking about a list of specific issues that go beyond just biomedical research, signed by a number of scientists including some who have won a nobel prize. I'm normally sceptical too but this is simply too much too simply dismiss as propaganda. I'm well aware how the conservative right has dominated the political agenda the past few years and how it has effected society in numerous ways.
About point 2, this style of politics seems to have replaced reason and common sense in the US. Both parties know most people will never get a chance to examine evidence in detail so instead they depend on making lots of noise in the media.
But still, when a large group of respected, smart and well educated people supports these accusations I think that is more credible than the white house telling us everything is fine. Getting Nobel prize winners to support this means that a few very smart people made a balanced judgement and came to the conclusion that they wanted to support this.
It takes some enormous wisdom or stupidity to dismiss such a thing. I'm afraid there's plenty of stupid americans who will just do that. I'm as pessimistic to believe that the current US government is actually so stupid that they actually believe they know better.
Izotope ozone is a (non free) winamp/directx plugin that emulates some of the distortion effects that sixties amplifiers produce using tubes. I've been using it for quite some time and it really enhances the listening experience. I can recommend it and it sure is worth the small license fee (which is peanuts compared to what you would need to invest in hardware otherwise). I haven't found any other plugins that produce a similar improvement in sound. There are many plugins that just beef up the bass a bit or add cheap 3d effects. Izotope Ozone is in a different league.
The plugin clearly demonstrates that the distortions (when used with care) can really enhance music. It also demonstrates that you can get the same effect by processing the sound digitally instead of with tubes. Izotope ozone actually goes way beyond what traditional tubes can do because it doesn't have the physical limitations.
Of course most commercial rock and pop music is processed and filtered in the studio before it is put on cd whereas older music (or indie records) tend to sound better when played back on equipment that adds the distortion effects. Of course the amount of distortion is a matter of personal taste and I find that I enjoy my music more with a little bass compression and a bit of sparkle in the higher ranges. Studios tend to optimize for cheap equipment (i.e. it has to sound nice on cheap radios) so you can gain a lot by adding some distortions.
You can also use sound distortion to compensate for lossy compression or lousy speakers. Just boost the bass digitally for the frequency range that your subwoofer can actually handle; add a little sparkle to compensate for loss of higher frequencies during the mp3 compression; add some overdrive on a guitar track. Distortion is not necessarily about reproducing sound as it was when it was recorded but about making it sound as nice/pleasing as possible. Much of the distortion effects in sixties equipment is deliberate and not accidental. Electrical guitars are a good example of how distortion can be used to produce a wide range of sounds.
Not really.
I don't know any desktop users who actually wait for stable debian releases. If you are using stable, you are using KDE 2.x or Gnome 1.x.
Most users trade stability for features because stable releases are obsolete the day they are released (this is intentional because nothing can possibly get into stable in less than a few months). Major packages like KDE or Gnome which have several minor releases per year are usually obsolete by the time they reach stable. This was really apparent with the last stable release where KDE 3 was already out and debian shipped with 2.x. Even testing is usually struggling to keep up with minor releases (which usually fix lots of bugs).
Most commercial debian releases are therefore not based on stable but on testing and unstable packages (obsolete is hard to sell in the desktop market). Stable is good for servers, not for desktops. On a server you get all the stable packages you need and maybe a handfull of handpicked non stable packages with things you really need.
It would be helpful though if the debian developers rethought their release strategy for desktop usage. I don't agree that stable needs to equate to obsolete (which apparently is their interpretation). Releasing more often would actually improve stability for most users because they would no longer have to rely on testing and unstable (which generally work out as the name suggests). I can symphatize with having a really stable version for server usage (so why not label it server). Testing is quite good most of the time but not always. So why not mark testing as stable three or four times a year? Use a few weeks to fix/remove the show stopping packages and rubber stamp it as stable.
I think the market is pretty much non existent outside the US. At least I'm not aware of any supported software by my bank (Abn Amro, one of the major banks in the Netherlands) other than their own web based software.
And judging from the review, we aren't missing much. I pay my bills online and haven't seen the inside of a bank for several years.
Tedious is standard with linux because of its packagemanagement tools which are incompatible with each other and even with earlier versions of the same tools. That is the main reason why most commercial software either supports only a very small subset of linux distributions (typically only specific versions of Red Hat) or bypass the packagemanagers entirely. That's not a Java problem. If you don't like tedious, use something else to run Java on. Linux is entirely optional. Of course if you are competent enough to get linux up and running you are more than qualified to follow the very small set of simple instructions to download and install the JRE. If you are still on a modem of course any download seems huge. But really, 10MB shouldn't be an issue any more these days, even if you are on a modem.
Of course if you download the jdk instead of just the runtime environment and an IDE you are looking at maybe 90 MB for the latest version of eclipse and another 40MB for the j2sdk.
Then the freedom issue. Yes, the SDK from SUN is not open source (though it is available for free for both commercial and non commercial use). However, open source JDKs and JREs are available (and are shipped with several linux distributions). If it makes you feel good, you can compile your gnu java with the gnu compiler and deploy it on your gnu server.
There is the small matter of the Java trademark which is indeed owned by Sun which and you may not stick on software without their permission (Just like the Mozilla foundation owns & controls the trademark for Mozilla). In practice this means that software that does not pass SUN's tests cannot be called Java. These tests are somewhat controversial and if that is what you meant by "not free" I don't disagree with you.
So unlike C, C++, packagemanagers, web browsers. The Java API is well defined (by SUN and several hundreds of major industrial and non industrial partners through the Java Community Process), and compliant open source and commercial implementations are widely available. Many of the reference implementations of J2EE APIs are open source (e.g. apache Tomcat is the reference implementation for the servlet specification). In short if you are looking for open source success stories, look no further than Java.
Finding a C++ compiler that complies 100% with the ANSI standards is nearly impossible (which is why porting applications between compilers is a major PITA) and why a minor compiler version increment requires you to update most of your linux software to versions that A) compile with the new version and B) have been compiled with the new version. 2.x to 3.x of the GNU compiler was very disruptive IMHO.
There are no standards for packagemanagers and the W3C publishes interesting, recommendations which nobody, including the intended audience of webbrowser developers, seems to be able to interpret in a consistent fashion.
Legally, the minister has voted on behalf of the dutch government. All actions of the government are ultimately controlled by the dutch parliament. Now, the parliament has decided that the minister should excercise a legal loophole and revoke the the vote.
Technically the minister can decide not to do this in which case a vote of no confidence may follow (after a debate with the minister on his reasons for not revoking his vote). However, given the political consequences of this scenario, this is unlikely to happen. A vote of no confidence would be rather damaging to all three of the coalition parties (given the recent polls) since it would likely result in new elections and result in a loss of the majority they are currently enjoying.
Given this context there are two scenarios. Either the parliament will look the other way if the minister ignores them (this will be settled behind closed doors and the opposition will make lots of noise) or the minister will revoke his vote. Given his earlier comments, the minister will favour the first option and will put lots of pressure on the three coalition parties to look the other way.
The only way to prevent this from happening is media attention. So far this has been pretty much a non issue in the media (few people understand what this debate is about). If this doesn't change, parliament won't work too hard on this issue. On the other hand if there is too much pressure on the minister, the coalition parties will likely back off to prevent a vote of no confidence.
The reason gets continues to be used is because it is in many textbooks and by default none of the IDEs, compilers, etc. seem to warn against its usage. Any security measure that requires education in order to work is fatally flawed. Security should be fool proof. Tools are your first line of defense and in many programming languages the only line of defense. If GCC doesn't tell you the code is flawed, in many cases you will never find out until it crashes.
Code that utilizes gets or similarly flawed legacy parts of C libraries should simply not compile without using a --insecure flag or something. If you want potentially insecure code to compile at least it should be a conscious choice. Of course gets isn't really the problem anymore (at least I'm sure that somebody has grepped sourcecode of e.g. sendmail for occurences of gets and buffer overflows continue to be a problem there).
Personally I find it amazing that buffer overflows still are an issue. The technology to prevent them has been around for decades. I never experience buffer overflows in Java. I do run into bugs in the JVM on very rare occasions (which is written in C). Also sometimes I discover memory or resource leaks in Java software (which eventually prevent continued operation of the software). Java is not immune to that sort of problems but buffer overflows are simply not an issue.
Considering that buffer overflows are the #1 security problem today, there is something to say for managed code environments. Personally, I believe that running non managed code on a server is irresponsible. The one certainty you have about software packages like sendmail, apache, IIS oracle, mysql, postgresql, openssh, etc. is that the latest buffer overflow won't be the last one to be found.
And there's your rounding error :-). Extrapolating to a meter, 100cm must obviously be 39.36 inches.
It would be hilarious if he did :-)