It's a good point but then shouldn't you extend that thought to all proprietary software? Without the source code any software could be doing a search of your files without you knowing. It seems strange to say you will boycott google search but not other proprietary products just because google's product is software that performs searches. It really depends on how much you trust the vendor when it comes to proprietary software.
Perhaps more unified could be a benefit for MS though. If 90% of people get their Linux through one or two companies then MS knows who to attack using their usual tactics. At the moment there are so many different distros being used that they can't attack them all. Maybe they just want to stream line the majority of Linux use through RedHat and Ubuntu so they can destroy them somehow in their next line of attack...
Each tag gets a URL. If you know the URL, you can use that. How is this not the same as having a meaningful name? Since when was a URL not a name?
Yes a URL is a name but it is a name used in a different context as far as the user is concerned. A URL specifies the branch to operate on and a revision identifier specifies the revision within that branch. They are 2 different namespaces and they should be kept that way for any kind of sane use. It does not make sense to commit to a tag so a tag should not be a URL. It is also easier for the user if tags are real tags instead of branches as this is how one thinks of them. Users are used to passing revision ids to commands and they just have to learn that a tag name can be substituted for a revision id wherever a revision id is needed or asked for.
Uhh yes there is. Subversion tagging mechanism does not do the main thing that a tagging mechanism should do: that is define a meaningful name to a revision in the repository which can be used in any command in place of that revision. If subversion really supported tagging I should be able to do this:
Microsoft did a lot of things much worse than Java though.
For example, C# removed checked exceptions which is one of the most useful features of the Java language. Most people don't seem to understand them though and think that they are just an annoyance and keep adding throws XXXXException to their methods. This totally removes the power of checked exceptions and is absolutely the wrong thing to do.
Checked exceptions are useful when you deal with IO in programs (and any useful program has to do some form of IO). Basically, you can be sure that the programmer has had to think about all IO problems (eg network cable removed, disk full, access denied, file does not exist etc) and decide if a given method is the correct place to handle it. If not then it needs to propagate the exception to a higher level - most likely wrapping it as the cause of a higher level exception. Thus higher up levels have to deal with only one kind of exception that is abstracted to their needs (e.g getCustomerData(String name) throws CustomerQueryException) and not many implementation specific exceptions (i.e NOT getCustomerData(String name) throws FileNotFoundException, PermissionDeniedException, FileNotReadableException, etc...)
So, because access to databases in C# do not throw checked exceptions unless the programmer remembers to handle every possible kind of exception (with no help from the compiler) the program will most likely crash with a meaningless error message. In Java the exceptions would get wrapped up the call chain and so the top level method that must handle it has enough information to present a meaningful error message to the user. eg:
Could not get the bank details for customer X. The was an error contacting the server. The connection to the server timed out.
Each line representing an exception that was wrapped in a higher level exception. Of course writing error handling code is annoying and not much fun but it is essential and very important. In Java one can be sure that the programmer has had to think about every problem that could occur and decide where it is best handled. As there as so many problems that can occur in any kind of complex application this is an amazing help from the compiler and leads to much cleaner designs when used properly. C# got this so wrong...
That's the person it can benefit the most IMHO! Learning to touch type on Dvorak is significantly easier than with Qwerty because within minutes you can be typing real words instead of jkjkjkjkfdfdfdfd like in qwerty typing lessons. This gives you a sense of progress and since much more typing is done on the home row with Dvorak you can type full sentences soon after. To do that in qwerty you have to be familiar with the whole layout and go through tedious and boring lessons.
If you can't touch type then I recommend you learn - you will never look back. And if you want to learn then it's much easier on Dvorak. If you use KDE then ktouch is a good program to learn with and of course it's free too...
Speed is not the main reason to switch to Dvorak. It's a nice benefit but the main reason for me is the comfort due to less finger stretching and movement and the even distribution of typing alternating between both hands.
The only downside I have found is the movement of shortcut keys. The main one being the X,C,V for cut, copy, paste. But I bought a typematrix http://typematrix.com/ keyboard that has separate keys for these operations anyway. Overall the benefits far outweighed the drawbacks for me....
1. Enumerate all the subclasses of a given class, or classes that implement a particular interface, including those supplied in plug-ins, at runtime.
This is not directly possible in Java API but can be done with a small utility function that uses reflection.
2. Call methods by name.
Again this can be done by reflection. You loose type safety so it's not encouraged.
3. Query whether a delegate object implements a given method, allowing for informal protocols.
Again this can be done through reflection. As Java tries to be type safe this is not part of the language syntax. You should use interfaces in this case. But it is possible to do this using reflection API as well.
4. Handle the case where an object tries to call a method on my object that doesn't exist, to allow the simple creation of generic proxy objects.
This is something Java and most statically type safe languages try to avoid as in most cases this is a programming error and it is better to catch at compile time. Using reflection you could check if the object supported the method or not though.
5. Add methods to a class, even if it's part of the standard library and I don't have the source code (I can even do this at runtime, although it's messier, and I haven't ever needed to).
This can be done with AspectJ.
6. Separate the allocation and initialisation of an object into separate methods, to allow different allocation policies to be implemented (e.g. pools for commonly re-cycled objects) transparently to users of the class.
Java implementations try to detect this automatically. In fact I think some implementations of Java can allocate objects faster than a malloc because they do pooling for you. But it would be nice if this could be done in Java I guess.
oint 12: They seem to be complaining about how hard it is to find individual windows for an application. Haven't they seen Expose? No? How about splat-` to cycle through the windows of the current application?
This is one of the worst and most inconsistent features I found in Mac OS/X. The cycle applications allows you to flick between 2 recently used apps with Apple Tab (I think - it was a while ago I use one) but when you flick between windows of the same app using the splat-` you have to cycle through them all. I had a text editor that had about 30 windows open (I am a programmer) and when I wanted to flick between the 2 most recently used windows of the text editor I expected this shortcut to behave the same as the apple tab to flick between the 2 recent apps. But no I had to press it 30 times to flick between the 2 most recently used files. Retarded and inconsistent! Maybe there is a way to do this and I am not aware but I could not find out how...
Commercial interests are driven by consumers so if we all act together we can help change this. It makes little sense to completely blame commercial entities for the situation. A company will try to maximise profits and if that means being more environmentally friendly because that is what consumers want then that is exactly what they will do. Unfortunately, few companies are rewarded by their customers for being environmentally friendly. This requires good education about the issues and ideas on what we can all do to help change this. I'm hopeful that this can change - for example, if I see a product that has carbon neutral emissions I would buy that over one that isn't and I walk or cycle instead of taking the car for the majority of my journeys. There are many small things we can all do that would make a huge difference if everyone tried.
I think we're more likely to make a significant difference this way than to rely on the governments to enforce behaviour by law - it is obvious that they are doing too little and too late...
I personally do not find eating meat unethical but I do find modern farming practices and factoring farming unethical. Treating an animal as though it were a machine is very different to how animals have been farmed for thousands of years. In modern times there is a strong emphasis on making profits with very little consideration for the animals or the environment in the process. People have become almost completely detached from the food production process and I'm sure most people would be horrified if they knew what went into the production of the food they eat. Farms used to be small scale with cows fed on grass and you knew the person who raised it - this is almost never true anymore.
Isn't role based security similar to what sudo has provided for a long time? You can configure sudo to allow a user to run only specific programs that require root priviledges. These programs are then the roles (eg apt would define the role of application installation).
Or am I missing something?
Many religious people believe that the people involved in translating and wording the bible were guided by God and so, therefore, were these translators. The bible always passed through people (even the very first version was written by people!) They do not think very rationally and will easily dismiss this issue of translation using this argument so that they can go on believing in the bible as the word of God. Dismissing evolution though is something that is not so easy for them to do.
True that evolution does not disprove God but it does, in my opinion, disprove many religions. I think the reason most religous people are offended by it is that it makes their Bible, Koran etc look somewhat silly. The thing is that they regard the bible as 100% correct and as the word of God. One only has to find one inacurracy and then you have shown that the bible is not the word of God (how could He be so careless to overlook such an issue and lie to us in the process?). If the bible is not the word of God and has such huge inaccuracies then that somewhat diminishes its respect to the point that it is not worthy as the foundation for belief.
Evolution is not a minor issue and cannot be easily explained away and contradicts the bible in so many ways.
What were MS thinking? It looks like they tried to create something with a wow factor and didn't really pull it off. Most of the effects look pointless and actually look like they would be anti-productive. For example, the window borders are semi transparent. What a rediculous idea - I mean the user wants to quickly and easily distinguish window borders and making them transparent makes this much harder to do. In one of those screenshots the inside of the window has such strong contrast and the border is transparen that my eye naturally goes for the inside of the window for where I would expect the title bar to be because of this.
I don't mind eye-candy but it should be used to boost productivity and clarity and not to reduce it. Oh well...
The article mentions that XGL made some architectural decisions that some people disagree with, and that AIGLX is a more "incremental" design. Does this mean that AIGLX is Yet Another Extension Bolted On to X?
That's not my understanding, anyway. AIGLX is not an extension to an X Server but an X server implementation. In fact, AIGLX and XGL are both implementations of an X server. An X server is a program that recieves requests from other programs (called X clients) to draw things on the screen, open windows etc. The X server also tells the clients about mouse moves and key presses etc. The X protocol defines how a client talks to the X server to achieve these things.
So AIGLX and XGL both listen to the same requests from clients using the same protocol. The difference between the 2 is in how they convert these requests to something the graphics card can understand. XGL converts all these drawing requests to OpenGL commands whereas AIGLX does not. Theoretically, you should be able to use Compiz or Metacity with either of these servers (it wouldn't know the difference as both speak the same protocol).
Cairo is a 2D graphics library for applications to draw things on the screen and Xgl is an X server that can process the cairo requests (usually through the XRender extension) and accelerate them through the graphics hardware. So they are completly different things and there is no waste of resources. In other words Xgl is an implementation of the X Window System and Cairo can output to many different windowing systems as well as X (eg MS Windows or Mac OS X).
Good points! I don't think a blanket tax on fuel for everyone would be a good idea. However, there is a middle ground. Tax subsidies on fuel could be given to farmers and others whose business depends on transportation. The majority of people use their cars to go buy a litre of milk at the grocery store and these people could be taxed in a higher bracket. The government already has lots of tax brackets so that not everyone pays the same tax rate. Couldn't they do a similar thing for fuel tax? This would discorage unnecessary use of a car but would not then penalize a plumber or farmer whose livelihood depends on transportation of mass goods. Elderly people could also be put in a lower fuel tax bracket. It may be difficult to implement effectively but if it could be implemented it could really help the environment and cut down on unnecessary fuel usage.
Really you should try it. They may have a slow web page but that has absolutely no reflection on the quality and speed of their product. It has to be seen to be believed. I have run programs over a modem in 24 bit (16 million colours), full screen 1400x1050, over a modem connection and it was practically indistiguishable from running the program locally! You can also run single apps and not just a full desktop like most other remote display protocols.
The advantage of this approach is that you can publish an app on the web that has an interface as rich and as interactive as a desktop app with no complicated browser compatibility problems. The development is greatly simplified and the user gets the best interactive interface across any network and any OS.
I think a much better approach for cross platform interactive web applications is through the NX technology from Nomachine http://nomachine.com/
It allows the full richness of a desktop app over a modem connection with easy deployment through a browser plugin. The app doesn't even need to be especially written for the web and the performance is simply amazing. It is also completely cross platform as the app runs on the server and remote displays to the client.
There is also a free server and all the smart compression libraries are under GPL.
Slashdot Mirror
I'm sure that is just as convenient to use as well and wouldn't get tedious...
It's a good point but then shouldn't you extend that thought to all proprietary software? Without the source code any software could be doing a search of your files without you knowing. It seems strange to say you will boycott google search but not other proprietary products just because google's product is software that performs searches. It really depends on how much you trust the vendor when it comes to proprietary software.
Perhaps more unified could be a benefit for MS though. If 90% of people get their Linux through one or two companies then MS knows who to attack using their usual tactics. At the moment there are so many different distros being used that they can't attack them all. Maybe they just want to stream line the majority of Linux use through RedHat and Ubuntu so they can destroy them somehow in their next line of attack...
Yes a URL is a name but it is a name used in a different context as far as the user is concerned. A URL specifies the branch to operate on and a revision identifier specifies the revision within that branch. They are 2 different namespaces and they should be kept that way for any kind of sane use. It does not make sense to commit to a tag so a tag should not be a URL. It is also easier for the user if tags are real tags instead of branches as this is how one thinks of them. Users are used to passing revision ids to commands and they just have to learn that a tag name can be substituted for a revision id wherever a revision id is needed or asked for.
Uhh yes there is. Subversion tagging mechanism does not do the main thing that a tagging mechanism should do: that is define a meaningful name to a revision in the repository which can be used in any command in place of that revision. If subversion really supported tagging I should be able to do this:
svn co -r tag:VERSION_1_1 svn://host/project
or this:
svn diff -r tag:VERSION_1_0 tag:VERSION_1_1
that they fix the bug where you can't disconnect USB drives in Feisty using unmount or eject. Otherwise end users will get the impression that Mark is trying to avoid: that is Ubuntu will look like a cheap Windows.... https://bugs.launchpad.net/ubuntu/+source/gnome-vo lume-manager/+bug/63090
Microsoft did a lot of things much worse than Java though.
For example, C# removed checked exceptions which is one of the most useful features of the Java language. Most people don't seem to understand them though and think that they are just an annoyance and keep adding throws XXXXException to their methods. This totally removes the power of checked exceptions and is absolutely the wrong thing to do.
Checked exceptions are useful when you deal with IO in programs (and any useful program has to do some form of IO). Basically, you can be sure that the programmer has had to think about all IO problems (eg network cable removed, disk full, access denied, file does not exist etc) and decide if a given method is the correct place to handle it. If not then it needs to propagate the exception to a higher level - most likely wrapping it as the cause of a higher level exception. Thus higher up levels have to deal with only one kind of exception that is abstracted to their needs (e.g getCustomerData(String name) throws CustomerQueryException) and not many implementation specific exceptions (i.e NOT getCustomerData(String name) throws FileNotFoundException, PermissionDeniedException, FileNotReadableException, etc...)
So, because access to databases in C# do not throw checked exceptions unless the programmer remembers to handle every possible kind of exception (with no help from the compiler) the program will most likely crash with a meaningless error message. In Java the exceptions would get wrapped up the call chain and so the top level method that must handle it has enough information to present a meaningful error message to the user. eg:
Could not get the bank details for customer X.
The was an error contacting the server.
The connection to the server timed out.
Each line representing an exception that was wrapped in a higher level exception. Of course writing error handling code is annoying and not much fun but it is essential and very important. In Java one can be sure that the programmer has had to think about every problem that could occur and decide where it is best handled. As there as so many problems that can occur in any kind of complex application this is an amazing help from the compiler and leads to much cleaner designs when used properly. C# got this so wrong...
That's the person it can benefit the most IMHO! Learning to touch type on Dvorak is significantly easier than with Qwerty because within minutes you can be typing real words instead of jkjkjkjkfdfdfdfd like in qwerty typing lessons. This gives you a sense of progress and since much more typing is done on the home row with Dvorak you can type full sentences soon after. To do that in qwerty you have to be familiar with the whole layout and go through tedious and boring lessons.
If you can't touch type then I recommend you learn - you will never look back. And if you want to learn then it's much easier on Dvorak. If you use KDE then ktouch is a good program to learn with and of course it's free too...
Speed is not the main reason to switch to Dvorak. It's a nice benefit but the main reason for me is the comfort due to less finger stretching and movement and the even distribution of typing alternating between both hands.
The only downside I have found is the movement of shortcut keys. The main one being the X,C,V for cut, copy, paste. But I bought a typematrix http://typematrix.com/ keyboard that has separate keys for these operations anyway. Overall the benefits far outweighed the drawbacks for me....
1. Enumerate all the subclasses of a given class, or classes that implement a particular interface, including those supplied in plug-ins, at runtime.
This is not directly possible in Java API but can be done with a small utility function that uses reflection.
2. Call methods by name.
Again this can be done by reflection. You loose type safety so it's not encouraged.
3. Query whether a delegate object implements a given method, allowing for informal protocols.
Again this can be done through reflection. As Java tries to be type safe this is not part of the language syntax. You should use interfaces in this case. But it is possible to do this using reflection API as well.
4. Handle the case where an object tries to call a method on my object that doesn't exist, to allow the simple creation of generic proxy objects.
This is something Java and most statically type safe languages try to avoid as in most cases this is a programming error and it is better to catch at compile time. Using reflection you could check if the object supported the method or not though.
5. Add methods to a class, even if it's part of the standard library and I don't have the source code (I can even do this at runtime, although it's messier, and I haven't ever needed to).
This can be done with AspectJ.
6. Separate the allocation and initialisation of an object into separate methods, to allow different allocation policies to be implemented (e.g. pools for commonly re-cycled objects) transparently to users of the class.
Java implementations try to detect this automatically. In fact I think some implementations of Java can allocate objects faster than a malloc because they do pooling for you. But it would be nice if this could be done in Java I guess.
oint 12: They seem to be complaining about how hard it is to find individual windows for an application. Haven't they seen Expose? No? How about splat-` to cycle through the windows of the current application? This is one of the worst and most inconsistent features I found in Mac OS/X. The cycle applications allows you to flick between 2 recently used apps with Apple Tab (I think - it was a while ago I use one) but when you flick between windows of the same app using the splat-` you have to cycle through them all. I had a text editor that had about 30 windows open (I am a programmer) and when I wanted to flick between the 2 most recently used windows of the text editor I expected this shortcut to behave the same as the apple tab to flick between the 2 recent apps. But no I had to press it 30 times to flick between the 2 most recently used files. Retarded and inconsistent! Maybe there is a way to do this and I am not aware but I could not find out how...
Commercial interests are driven by consumers so if we all act together we can help change this. It makes little sense to completely blame commercial entities for the situation. A company will try to maximise profits and if that means being more environmentally friendly because that is what consumers want then that is exactly what they will do. Unfortunately, few companies are rewarded by their customers for being environmentally friendly. This requires good education about the issues and ideas on what we can all do to help change this. I'm hopeful that this can change - for example, if I see a product that has carbon neutral emissions I would buy that over one that isn't and I walk or cycle instead of taking the car for the majority of my journeys. There are many small things we can all do that would make a huge difference if everyone tried. I think we're more likely to make a significant difference this way than to rely on the governments to enforce behaviour by law - it is obvious that they are doing too little and too late...
I personally do not find eating meat unethical but I do find modern farming practices and factoring farming unethical. Treating an animal as though it were a machine is very different to how animals have been farmed for thousands of years. In modern times there is a strong emphasis on making profits with very little consideration for the animals or the environment in the process. People have become almost completely detached from the food production process and I'm sure most people would be horrified if they knew what went into the production of the food they eat. Farms used to be small scale with cows fed on grass and you knew the person who raised it - this is almost never true anymore.
Isn't role based security similar to what sudo has provided for a long time? You can configure sudo to allow a user to run only specific programs that require root priviledges. These programs are then the roles (eg apt would define the role of application installation). Or am I missing something?
Many religious people believe that the people involved in translating and wording the bible were guided by God and so, therefore, were these translators. The bible always passed through people (even the very first version was written by people!) They do not think very rationally and will easily dismiss this issue of translation using this argument so that they can go on believing in the bible as the word of God. Dismissing evolution though is something that is not so easy for them to do.
True that evolution does not disprove God but it does, in my opinion, disprove many religions. I think the reason most religous people are offended by it is that it makes their Bible, Koran etc look somewhat silly. The thing is that they regard the bible as 100% correct and as the word of God. One only has to find one inacurracy and then you have shown that the bible is not the word of God (how could He be so careless to overlook such an issue and lie to us in the process?). If the bible is not the word of God and has such huge inaccuracies then that somewhat diminishes its respect to the point that it is not worthy as the foundation for belief. Evolution is not a minor issue and cannot be easily explained away and contradicts the bible in so many ways.
What were MS thinking? It looks like they tried to create something with a wow factor and didn't really pull it off. Most of the effects look pointless and actually look like they would be anti-productive. For example, the window borders are semi transparent. What a rediculous idea - I mean the user wants to quickly and easily distinguish window borders and making them transparent makes this much harder to do. In one of those screenshots the inside of the window has such strong contrast and the border is transparen that my eye naturally goes for the inside of the window for where I would expect the title bar to be because of this. I don't mind eye-candy but it should be used to boost productivity and clarity and not to reduce it. Oh well...
The article mentions that XGL made some architectural decisions that some people disagree with, and that AIGLX is a more "incremental" design. Does this mean that AIGLX is Yet Another Extension Bolted On to X? That's not my understanding, anyway. AIGLX is not an extension to an X Server but an X server implementation. In fact, AIGLX and XGL are both implementations of an X server. An X server is a program that recieves requests from other programs (called X clients) to draw things on the screen, open windows etc. The X server also tells the clients about mouse moves and key presses etc. The X protocol defines how a client talks to the X server to achieve these things. So AIGLX and XGL both listen to the same requests from clients using the same protocol. The difference between the 2 is in how they convert these requests to something the graphics card can understand. XGL converts all these drawing requests to OpenGL commands whereas AIGLX does not. Theoretically, you should be able to use Compiz or Metacity with either of these servers (it wouldn't know the difference as both speak the same protocol).
Cairo is a 2D graphics library for applications to draw things on the screen and Xgl is an X server that can process the cairo requests (usually through the XRender extension) and accelerate them through the graphics hardware. So they are completly different things and there is no waste of resources. In other words Xgl is an implementation of the X Window System and Cairo can output to many different windowing systems as well as X (eg MS Windows or Mac OS X).
Good points! I don't think a blanket tax on fuel for everyone would be a good idea. However, there is a middle ground. Tax subsidies on fuel could be given to farmers and others whose business depends on transportation. The majority of people use their cars to go buy a litre of milk at the grocery store and these people could be taxed in a higher bracket. The government already has lots of tax brackets so that not everyone pays the same tax rate. Couldn't they do a similar thing for fuel tax? This would discorage unnecessary use of a car but would not then penalize a plumber or farmer whose livelihood depends on transportation of mass goods. Elderly people could also be put in a lower fuel tax bracket. It may be difficult to implement effectively but if it could be implemented it could really help the environment and cut down on unnecessary fuel usage.
Really you should try it. They may have a slow web page but that has absolutely no reflection on the quality and speed of their product. It has to be seen to be believed. I have run programs over a modem in 24 bit (16 million colours), full screen 1400x1050, over a modem connection and it was practically indistiguishable from running the program locally! You can also run single apps and not just a full desktop like most other remote display protocols. The advantage of this approach is that you can publish an app on the web that has an interface as rich and as interactive as a desktop app with no complicated browser compatibility problems. The development is greatly simplified and the user gets the best interactive interface across any network and any OS.
I think a much better approach for cross platform interactive web applications is through the NX technology from Nomachine http://nomachine.com/ It allows the full richness of a desktop app over a modem connection with easy deployment through a browser plugin. The app doesn't even need to be especially written for the web and the performance is simply amazing. It is also completely cross platform as the app runs on the server and remote displays to the client. There is also a free server and all the smart compression libraries are under GPL.