I have a computer with a pre-installed version of Windows. It's a package deal. This Is Significant And Important (TM). I donate the PC to a school. The Windows license must accompany it.
Nonsense. I am under no obligation of donating the Windows license. The license may or may not be transferable to another machine, but I can certainly just throw it out. And that's, in fact, what I do with any unwanted Windows license; I see no point in polluting the minds of little children with Microsoft software.
I tend to agree that it isn't entrapment. But depending on what they do, it may be dangerously close and set a bad precedent for other areas of law. Also, if they leave the key in the ignition, they will probably end up catching many more low-level crooks that might otherwise not steal cars.
But I hope you're not trying to argue that a simple data access technique like the one MySQL uses can scale indefinately!
Why not? What kind of magic do you think happens when problems get big? The most scalable and efficient database applications use the simplest representations: fixed size records addressed by absolute location on a raw device. They access any record with at most one, carefully scheduled seek, with no queries, no buffer copies, and no context switch. Using any kind of additional data structure would only make them slower.
Which is why Interbase has an excellent reputation as an engine that supports high performance with low overhead.
That's roughly like saying that the C-shell is faster than the Bourne shell. It may be true, and someone who only knows shell programming may care. But if performance really matters, you don't write shell scripts, you program in assembly or C.
But to generalize from that to the believe that MySQL is the fastest engine for all applications is very naive.
If you actually read what I wrote, you'll see that I didn't make such a general claim.
Despite its SQL front end, MySQL is just a simple indexed file engine.
And how do you think that imposing the additional requirements of full SQL semantics makes a database faster?
Here's the nasty truth: for most large-scale applications, MySQL is very slow.
Now, who is talking in vague generalities? If MySQL matches the application, it tends to do pretty well in my experience, even for large databases. Slashdot looks to me like it ought to be a pretty good match to MySQL.
The nasty truth is that many real-world, high-performance transactional systems, scientific databases, and web sites don't use SQL or relational databases at all because they are simply too slow. Go and actually look at IBM's high-end transaction processing systems or the architectures behind major web search engines.
What do we need this law for? It seems invasive and just gives police additional tools to pressure people they don't like. Just wait: if you have a web cam pointing out your window and the police don't find the illegal copies of MS Office that the BSA claims you have based on the tip from your annoying former roommate, they'll just accuse you of video voyeurism.
I say: if you don't want to get taped, just close the curtains. If you keep your windows open, don't complain if other people look in or even tape it. Current privacy laws already protect you from other broadcasting their recordings. Sorry, we don't need a new law for this, and criminalizing video cameras is a very bad idea.
It would have been nice for Borland to find the contributions of the open source community useful enough to stick with keeping Interbase open. But the fact is that between MySQL and PostgreSQL, there is probably not much need for yet another open source database.
MySQL is indeed not a "real DBMS", but that fact makes it faster, not slower. If Slashdot had a serious problem with data corruption, you'd have a point. But in terms of performance, moving to a "real DBMS" would make the situation worse.
Sun is still calling all the shots when it comes to Java. Sun should follow through on their original promise: to make the core of the Java platform an ISO or ECMA standard that anybody can implement and build on without any obligation to Sun whatsoever.
Sun's excuses for not following through on their commitments don't hold water. Sun claims that their tight control over Java is necessary to ensure compatibility, but the Java APIs already fail to guarantee cross platform compatibility, and some of Sun's APIs are so deficient that third party APIs have already become a de-facto standard--on some platforms.
In different words, as far as I'm concerned, the JCP changes don't address the fundamental problems with Java and where it is going.
I would ask the opposite question: can companies like Microsoft or IBM ever produce usable software? You see, they are motivated by selling their wares and getting you hooked on their software. Whether their software is actually efficient or well-designed doesn't matter because their users rarely become proficient at any alternative and rarely have a basis for comparison.
It is particularly ironic for people from IBM to comment on free software usability. IBM's usability designers have often produced complete junk. Take a look at the interface hall of shame, for example.
Free software UIs are the way they are because the people who use them like them that way, not because of some hypothetical problems with the open source development process. The open source projects with the worst UIs (from the point of view of many open source users) tend to be the ones that try to duplicate Windows or have "professional" UI designers involved in them.
If "ordinary users" find Windows easy to use, great! Let them pay for it. Open source efforts have no obligation to cater to preferences that their contributors and participants don't share.
The "unfriendly" administration interface of Linux programs is one of the main reasons I like to use Linux. You see, I rather prefer editing a text file once than going through endless configuration wizards and dialog boxes on Windows. Let's please not port Windows stupidity to Linux.
There is nothing wrong with reasonable volume charges: volume is what costs the ISP, and it makes sense to charge more for more usage.
However, "reasonable" is important. Average monthly payments should remain the same or go down, and off-hour volume charges should be steeply discounted or zero.
The links in question are apparently instructions about how to sabotage transports of nuclear materials in Germany, which happen to be done by rail there.
In the US, nuclear materials are transported by road. Imagine, for a moment, what would happen if you posted accurate information on the route information, security procedures, and instructions on how to sabotage such a transport here. Do you really believe the FBI wouldn't be knocking on your door? In the current climate, you'd probably simply disappear in some US "holding cell" somewhere, not to be heard from for months or years.
Microsoft can't "license" a published specification. If they publish it, it's protected by copyright, nothing else. That allows them to tell you what you can do with the specification itself to some limited degree, it doesn't give them any rights to tell you what you can do with the ideas contained in that document.
Microsoft can only restrict what you can do with the ideas if it involves patents they own or if it is unpublished, trade-secret information and you have an explicit contract with them. Neither of those applies in this case (they have some CIFS-related patents, but you can work around those).
Note that this is completely different from the GPL/LGPL. The GPL/LGPL is a license that gives you additional rights to the copyrighted material (the code) itself, rights that you wouldn't otherwise have. Microsoft is trying to take away rights to material that they don't have a copyright on based on fair use of their published, copyrighted material; sorry, but that just doesn't work. I'd love to see them try enforce this in court.
Right now, the model is pretty simple: you connect to the Internet through some big US ISP and you pay for traffic in both directions. Only a few, special entities have the leverage to get peering arrangements.
This arrangement isn't particularly nice; for example, a public service site with popular content still ends up paying for their bandwidth, even though it might actually make the Internet more attractive to other customers and hence contribute positively to the bottom line of the ISP it is already paying. Trouble is: nobody has figured out a good way of measuring that, and trying to measure it would impose enormous overhead and complexity. In the end, we'd likely all pay more, and Africans still wouldn't get their peering arrangement because of their asymmetric traffic patterns.
If Africa ever becomes a popular Internet destination, they will have the leverage to negotiate a peering arrangment. Until then, Africans can enjoy very cheap bandwidth, and they have a choice of where to hook up. They could hook up through a European ISP and negotiate a special deal if they don't have much US traffic.
On a serious note, it shows that we can do things under linux that happen in Windows; the OLE model in Windows has allowed things like this for years, and it's about time we had a similar model in the *nix world.
I find it absolutely fascinating that people think that the UNIX world lacks features found in Windows because, well, because of what? Do you seriously believe that people didn't have the resources to create this for UNIX?
The fact is that OLE (and its successors) are unreliable kludges. They grew out of some hacks trying to make Office components talk to one another, for which they were sort-of OK. But when applied to arbitrary applications, the GUI merging and process model they give you result in lousy user interfaces and unreliable applications. UNIX didn't get them because UNIX users didn't see a pressing need for them; other mechanisms work better.
Even if you think that what OLE was trying to do is actually useful, OLE and its successors are about the worst way of implementing it: unportable, low-level hacks that are difficult to version and don't protect programs from one another. OLE is really mainly an attempt to give C++ programs at least a little bit of what systems like Smalltalk and Java give you for free. And, surprise, Microsoft finally caught on and replaced the whole OLE/COM/ActiveX mess with C#/CLR (of course, there is some backwards compatibility and they use the OLE/COM terminology to explain the new functionality in C#/CLR).
If there was something genuinely new in these web standards, I could see at least having a debate about whether patents on them are reasonable. But most of the web stuff is just doing with HTTP and XML what has been well-known since the 1970's and 1980's in terms of binary RPC mechanisms; patents on those mostly have expired or will expire soon.
Even if there were some real inventions in that space, the older mechanisms, translated to HTTP and XML, are perfectly adequate for serving the needs of most business applications. So, there is no reason for the W3C to standardize anything that involve Microsoft or IBM patents (or patents by anybody else, for that matter).
If you haven't noticed, many banks, government agencies, and other institutions that you deal with almost daily have no qualms requiring you to use Microsoft products. They will have no qualms exposing their services using technology that involves Microsoft and IBM patents either.
So, as long as you only run open source software talking to other open source software, it may not matter to you. But if you want to interact electronically with the rest of the world, this matters a lot.
The point is, it's not obvious what all these buttons do, they're hard to press, and *SETTING* a watch should be a fairly obvious function, and simple to do.
Different devices for different people. Crowns are easy to set--if you have the dexterity. Four buttons may be a whole lot easier to some people, if they bother to read the instructions.
See, that's the problem with a lot of UI design: people assume that's what's easy for one group of people is easy for everybody.
Sun has promised standardization for Java, efficient support for numerical computations, fast startup and reduced memory footprint, availability of multiple independent implementations, and a whole lot of other things.
What has Sun delivered? Well, they have been doing a good job on the JIT. But beyond that, the Java APIs have become huge, there is no new support for numerical computations, the genericity support is flaky, VM sharing and fast startup are missing, and Sun is suing or picking a fight with whoever shows any initiative in the Java space.
C# isn't a lot better than Java, although it has a few important additional features. But, amazingly, Microsoft has been doing a much better job opening up C# in the form of ECMA C# than Sun. Maybe, in the long run, we may just have to thank Sun for providing the competition that caused Sun to open up.
No, Norman is right. (Disclaimer, I saw Norman's keynote at UIST '94).
m
I didn't say he was wrong, I said many HCI researchers are addressing irrelevant problems.
And I hate DST time change, because I have no clue how to set/change the time on my daughters' digital watches (four unlabeled buttons, -- too small to really press properly, none of which has the obvious function of time set).
If you can't deal with it, why did you buy that watch? Or if your daughter bought it, why do you set it for her? It's here problem. There are plenty of analog watches with a crown; I have one. Other people may want the features that are accessed through those buttons.
If we are to enter the era of what Norman calls "ubiquitous computing",
The era is already here, and Norman didn't even invent the term.
Xerox invested heavily in bringing several machines based on the work developed at PARC to market, so I think this was something Xerox was very much involved in.
Look at the list of things he cares about. DVD menus are too complex? People can't deal with home theaters? Who cares? Why is that even worth spending any time on?
Most people pop DVDs into their player and play them. Some people play around with the menus, many people don't. It's a leisure activity. It doesn't have to be efficient. It doesn't have to be simple. And if some people don't get it, it doesn't matter. If it's quirky, that's part of the charm. If customers don't like it, the manufacturer's focus groups will let them know.
Tog had a similarly irrelevant column tearing apart the MacOSX dock. Come on, what's the problem? It looks nice, people like it, and anybody with an IQ greater than 80 can use it. Optimizing it doesn't save anybody any real amount of time.
Guys: spend some time thinking about some real stuff, stuff that matters. Saying semi-obvious things about trivial little features really isn't interesting.
I have followed HCI research on and off for the last decade, and I think it's largely missing the point. Just look at some really long-lived and successful real-world user interfaces: musical instruments, typewriters, cars, bicycles, electronic devices, etc. What makes a user interface successful is a very complex mix of factors. Being intuitive and efficient, two criteria that are the focus of much HCI research, are only two minor factors; factors like style, design, power, simplicity, and physical constraints are often much more important--and they should probably be for computer interfaces as well.
Or, in different words, if musical instruments were designed like software, instead of violins and pianos, we'd probably only be getting those electronic children's books that play a melody when you touch different parts on the page. Kind of intuitive and easy, but not exactly very powerful or interesting.
Slashdot Mirror
Nonsense. I am under no obligation of donating the Windows license. The license may or may not be transferable to another machine, but I can certainly just throw it out. And that's, in fact, what I do with any unwanted Windows license; I see no point in polluting the minds of little children with Microsoft software.
I tend to agree that it isn't entrapment. But depending on what they do, it may be dangerously close and set a bad precedent for other areas of law. Also, if they leave the key in the ignition, they will probably end up catching many more low-level crooks that might otherwise not steal cars.
Why not? What kind of magic do you think happens when problems get big? The most scalable and efficient database applications use the simplest representations: fixed size records addressed by absolute location on a raw device. They access any record with at most one, carefully scheduled seek, with no queries, no buffer copies, and no context switch. Using any kind of additional data structure would only make them slower.
Which is why Interbase has an excellent reputation as an engine that supports high performance with low overhead.
That's roughly like saying that the C-shell is faster than the Bourne shell. It may be true, and someone who only knows shell programming may care. But if performance really matters, you don't write shell scripts, you program in assembly or C.
If you actually read what I wrote, you'll see that I didn't make such a general claim.
Despite its SQL front end, MySQL is just a simple indexed file engine.
And how do you think that imposing the additional requirements of full SQL semantics makes a database faster?
Here's the nasty truth: for most large-scale applications, MySQL is very slow.
Now, who is talking in vague generalities? If MySQL matches the application, it tends to do pretty well in my experience, even for large databases. Slashdot looks to me like it ought to be a pretty good match to MySQL.
The nasty truth is that many real-world, high-performance transactional systems, scientific databases, and web sites don't use SQL or relational databases at all because they are simply too slow. Go and actually look at IBM's high-end transaction processing systems or the architectures behind major web search engines.
I say: if you don't want to get taped, just close the curtains. If you keep your windows open, don't complain if other people look in or even tape it. Current privacy laws already protect you from other broadcasting their recordings. Sorry, we don't need a new law for this, and criminalizing video cameras is a very bad idea.
It would have been nice for Borland to find the contributions of the open source community useful enough to stick with keeping Interbase open. But the fact is that between MySQL and PostgreSQL, there is probably not much need for yet another open source database.
MySQL is indeed not a "real DBMS", but that fact makes it faster, not slower. If Slashdot had a serious problem with data corruption, you'd have a point. But in terms of performance, moving to a "real DBMS" would make the situation worse.
Sun's excuses for not following through on their commitments don't hold water. Sun claims that their tight control over Java is necessary to ensure compatibility, but the Java APIs already fail to guarantee cross platform compatibility, and some of Sun's APIs are so deficient that third party APIs have already become a de-facto standard--on some platforms.
In different words, as far as I'm concerned, the JCP changes don't address the fundamental problems with Java and where it is going.
It is particularly ironic for people from IBM to comment on free software usability. IBM's usability designers have often produced complete junk. Take a look at the interface hall of shame, for example.
Free software UIs are the way they are because the people who use them like them that way, not because of some hypothetical problems with the open source development process. The open source projects with the worst UIs (from the point of view of many open source users) tend to be the ones that try to duplicate Windows or have "professional" UI designers involved in them.
If "ordinary users" find Windows easy to use, great! Let them pay for it. Open source efforts have no obligation to cater to preferences that their contributors and participants don't share.
The "unfriendly" administration interface of Linux programs is one of the main reasons I like to use Linux. You see, I rather prefer editing a text file once than going through endless configuration wizards and dialog boxes on Windows. Let's please not port Windows stupidity to Linux.
However, "reasonable" is important. Average monthly payments should remain the same or go down, and off-hour volume charges should be steeply discounted or zero.
In the US, nuclear materials are transported by road. Imagine, for a moment, what would happen if you posted accurate information on the route information, security procedures, and instructions on how to sabotage such a transport here. Do you really believe the FBI wouldn't be knocking on your door? In the current climate, you'd probably simply disappear in some US "holding cell" somewhere, not to be heard from for months or years.
Microsoft can only restrict what you can do with the ideas if it involves patents they own or if it is unpublished, trade-secret information and you have an explicit contract with them. Neither of those applies in this case (they have some CIFS-related patents, but you can work around those).
Note that this is completely different from the GPL/LGPL. The GPL/LGPL is a license that gives you additional rights to the copyrighted material (the code) itself, rights that you wouldn't otherwise have. Microsoft is trying to take away rights to material that they don't have a copyright on based on fair use of their published, copyrighted material; sorry, but that just doesn't work. I'd love to see them try enforce this in court.
This arrangement isn't particularly nice; for example, a public service site with popular content still ends up paying for their bandwidth, even though it might actually make the Internet more attractive to other customers and hence contribute positively to the bottom line of the ISP it is already paying. Trouble is: nobody has figured out a good way of measuring that, and trying to measure it would impose enormous overhead and complexity. In the end, we'd likely all pay more, and Africans still wouldn't get their peering arrangement because of their asymmetric traffic patterns.
If Africa ever becomes a popular Internet destination, they will have the leverage to negotiate a peering arrangment. Until then, Africans can enjoy very cheap bandwidth, and they have a choice of where to hook up. They could hook up through a European ISP and negotiate a special deal if they don't have much US traffic.
I find it absolutely fascinating that people think that the UNIX world lacks features found in Windows because, well, because of what? Do you seriously believe that people didn't have the resources to create this for UNIX?
The fact is that OLE (and its successors) are unreliable kludges. They grew out of some hacks trying to make Office components talk to one another, for which they were sort-of OK. But when applied to arbitrary applications, the GUI merging and process model they give you result in lousy user interfaces and unreliable applications. UNIX didn't get them because UNIX users didn't see a pressing need for them; other mechanisms work better.
Even if you think that what OLE was trying to do is actually useful, OLE and its successors are about the worst way of implementing it: unportable, low-level hacks that are difficult to version and don't protect programs from one another. OLE is really mainly an attempt to give C++ programs at least a little bit of what systems like Smalltalk and Java give you for free. And, surprise, Microsoft finally caught on and replaced the whole OLE/COM/ActiveX mess with C#/CLR (of course, there is some backwards compatibility and they use the OLE/COM terminology to explain the new functionality in C#/CLR).
Even if there were some real inventions in that space, the older mechanisms, translated to HTTP and XML, are perfectly adequate for serving the needs of most business applications. So, there is no reason for the W3C to standardize anything that involve Microsoft or IBM patents (or patents by anybody else, for that matter).
So, as long as you only run open source software talking to other open source software, it may not matter to you. But if you want to interact electronically with the rest of the world, this matters a lot.
Different devices for different people. Crowns are easy to set--if you have the dexterity. Four buttons may be a whole lot easier to some people, if they bother to read the instructions.
See, that's the problem with a lot of UI design: people assume that's what's easy for one group of people is easy for everybody.
That's the point. Something that is hard to learn is often more useful/versatile and more rewarding than something that's easy to learn.
What has Sun delivered? Well, they have been doing a good job on the JIT. But beyond that, the Java APIs have become huge, there is no new support for numerical computations, the genericity support is flaky, VM sharing and fast startup are missing, and Sun is suing or picking a fight with whoever shows any initiative in the Java space.
C# isn't a lot better than Java, although it has a few important additional features. But, amazingly, Microsoft has been doing a much better job opening up C# in the form of ECMA C# than Sun. Maybe, in the long run, we may just have to thank Sun for providing the competition that caused Sun to open up.
It's malicious, it's destructive, and it attempts to install itself on every PC...
m I didn't say he was wrong, I said many HCI researchers are addressing irrelevant problems.
And I hate DST time change, because I have no clue how to set/change the time on my daughters' digital watches (four unlabeled buttons, -- too small to really press properly, none of which has the obvious function of time set).
If you can't deal with it, why did you buy that watch? Or if your daughter bought it, why do you set it for her? It's here problem. There are plenty of analog watches with a crown; I have one. Other people may want the features that are accessed through those buttons.
If we are to enter the era of what Norman calls "ubiquitous computing",
The era is already here, and Norman didn't even invent the term.
Xerox invested heavily in bringing several machines based on the work developed at PARC to market, so I think this was something Xerox was very much involved in.
Most people pop DVDs into their player and play them. Some people play around with the menus, many people don't. It's a leisure activity. It doesn't have to be efficient. It doesn't have to be simple. And if some people don't get it, it doesn't matter. If it's quirky, that's part of the charm. If customers don't like it, the manufacturer's focus groups will let them know.
Tog had a similarly irrelevant column tearing apart the MacOSX dock. Come on, what's the problem? It looks nice, people like it, and anybody with an IQ greater than 80 can use it. Optimizing it doesn't save anybody any real amount of time.
Guys: spend some time thinking about some real stuff, stuff that matters. Saying semi-obvious things about trivial little features really isn't interesting.
Or, in different words, if musical instruments were designed like software, instead of violins and pianos, we'd probably only be getting those electronic children's books that play a melody when you touch different parts on the page. Kind of intuitive and easy, but not exactly very powerful or interesting.