I view school as bootstrapping a person to learn how to learn, and for teaching them the things that are timeless.
I heartily agree. I'm largely a self taught programmer (a few evening classes in C and Assembly, that's about it academically), having come from a background of history and languages. I now feel that my degree course was really a grounding in assessing the worth of source material and how to learn rather than having any direct vocational value. I still wish I had done a good CompSci degree instead though!
As for me, if you mentioned Java in the interview I'd charge 50% more. Its hazard pay.
And if I was interviewing you you wouldn't get the job, although as you state your experience is C, C++ and Perl you shouldn't even be in the running for a Java job anyway. If you were a Java coder, but felt the need for "hazard pay" then my instinct would be that your not a competent Java coder. Having done a bit of C++ before I moved from C to Java, I found I had to "unlearn" a lot of C++ in order to fit in with the more Smalltalk or Objectice-C way of doing things.
Java GUI motivates some students to get more involved
I'd suggest making a subset of the projects non-GUI though (in other words, disallow use of the java.awt.* and javax.swing.* classes), just so the students don't become totally distracted by interface issues. I'd also stress the importance of understanding MVC for all projects - separating the core logic of a program from the presentation. This will definitely benefit them if the students become professional Java coders, as I see way too much code that isn't modular and is tied to either Servlets or Swing, thereby hindering testability and reuse.
Having made the transition from C to Java, I don't understand why an experienced procedural programmer would have any more difficulty learning OO abstractions than a college student. The only thing I can think of is that the student is in an environment where they should have the time and resources (including, I hope, competent teachers) to learn whereas a professional programmer may not. This is reinforced by my experience at one company where a bunch of Fortran and C programmers were simply told that they would be programming in Java as of that moment. No training or time to learn, they were thrown immediately into a large Java project which was fairly complex stuff with a large amount of CORBA! Under those circumstances it isn't surprising that the code turned into a mess, but I fail to see how grounding college students in Assembly and C means they will struggle with OO languages later.
As for Java being a bad choice for a computer science course, I agree if it is used as the only programming language. Even if you never plan on coding professionally in a language without access to generic data structures (such as Java's Collection classes) or garbage collection, it is still beneficial to have an idea of what's gonig on under the hood. Occasionally I still write my own data structure classes, even though I primarily work in Java, because I have an understanding of how to measure performance and choose alternative algorithms. Even if I was to only use the Collection classes, an understanding of algorithms is essential (the frequency with which I've seen LinkedList used where ArrayList or HashSet is more appropriate concerns me a lot). And teaching in Java doesn't mean that an algorithms course can't be taught with it - take a look at "Data Structures and Algorithms Analysis in Java" by Weiss for example.
Would you care to list them? Apart from errors caused by people using old ext3 journals after they'd accidentally mounted a partition as ext2 and modified it, I'm aware of no recent reports of file corruption bugs. There was an invalid inode size problem about a year ago that could cause a system hang, but as I understood it ext2 and ext3 couldn't create that invalid size itself.
I've got a Zaurus - the C3200 model - and an Asus Eee. The Zaurus is way better than any phone or PDA that I've used, as it really was the miniature laptop that a lot of people were looking for. However, compared to the Eee it's not so hot. The keyboard is really difficult to use, and the lack of power from the USB port means you need a powered hub to use an external keyboard. The available Linux distros for the Zaurus have small developer teams and are very unstable - they generally turn my machine into a brick whenever I try to configure wireless networking or perform an update. The Eee on the other hand has a usable keyboard, Pentium processor and conventional BIOS. This means a plain x86 Linux distro or BSD will install and run on the machine with no difficulty. The Xandros based distro that the Eee comes with is very nice when you actually use it rather than just criticising it based on the desktop theme as some people have done, and it's easy to strip the machine down if you want to (my Eee now runs NetBSD for example).
Despite his authoritarianism, Fascism still cannot be attributed to Franco. His political convictions were much more in keeping with the monarchist and staunchly religious groups that backed his uprising - groups antithetical to the Fascism of the Falange, which is why Franco was keen to utilise its popularity but sideline its leaders. Fascism, as defined by the its early protaganists, was intended to be a modernising force and as such quite anti-monarchist and anti-clerical. The authoritarian aspects reflected a belief that democracy was too weak to cope with events such as global economic depression, and that a strong leadership was needed to counter Bolshevism. With the Nazi regime colouring our perception of the politics of inter-war Europe, it's difficult for many people to appreciate now that many felt then that democracy was a bankrupt concept, and a clash of ideologies was imminent. It was in this atmosphere that conservatives in Germany as well as Spain allied themselves with the populist Fascists. The difference was that in Germany it was the extreme right wingers that managed to end up manipulating the conservatives.
As for anarchism, my knowledge of that is limited to groups such as the CNT who were so badly treated by Stalin's stooges during the Spanish Civil War. Socialism on the other hand, is close to what I'm currently studying, as I'm hoping to do a part-time masters degree on the Paris commune in the not too distant future.
Franco was not a fascist - he was a conservative with strong religious convictions. He co-opted the Falange movement as a convenient political front for what was simply a military coup. Franco made no attempt to save Primo de Rivera (the leader of the Falange imprisoned and sentenced to death by the Republicans), as it was not in his interest - Franco was an uncharismatic figure, and saw Primo de Rivera as a potential opponent. I suggest you read a good book on the Spanish civil war (such as Anthony Beevor's "The Battle Of Spain") and also look at the post war political history of Spain. You might also benefit from reading about Fascism rather than making the common mistake of confusing it with regimes of Hitler and (beyond the first few years at least) Mussolini. Just as Stalin's regime is not a Marxist one, Franco's was not a Fascist one.
You make it sound like FreeBSD, which scares the shit out of me.
Yeah, it's a bit of a coincidence that Net will be following in the footsteps of Free by releasing a whole load of locking changes as prt of version 5.0. However, NetBSD -current already has the main pieces in place, and is shaping up to be a massive improvement in terms of SMP performance. Check Andre Doran's posts to the tech-kern mailing list over the last few months for the details.
In a sense NetBSD event ran Linux before the Xen support - well, Linux applications at least. There's Linux emulation built into the kernel that allows it to trap system calls from Linux binaries and translate them to NetBSD equivalents. Before the availability of a native Sun JDK, this was the way to run Java on NetBSD.
While the sharing is great and goes both ways, it's a little unfair to imply that all the new features in NetBSD 4.0 came from OpenBSD. The bluetooth stack was written specifically for NetBSD, for example. As for pf, the version in the NetBSD tree has been considerably modified from the 3.7 version, so merges from OpenBSD are made on a fix or per-feature basis as the porting is quite hard to do. Not unexpected, seeing as pf touches some quite low level aspects of the kernel. As for the release, I'll be upgrading my main machine from RC3 as soon as possible, but I'm really looking forward to 5.0 - which will include a new threading implementation along with massive improvements to kernel locking granularity and SMP support.
There was some sort of "developer release candidate", but I can't find it on the Apple dev site anymore, and have seen blog comments that suggest it's been pulled.
So in other words, Apple definitely did something dumb here, but it's not too hard to see what their motivations were.
The deprecation of the Java-Cocoa libraries is linked to Apple's support for Objective-C. At one point Apple was saying that Objective-C would be gradually replaced by C++ at one extreme and Java at the other. However, they seem to have realised that Objective-C and the tarted up OpenStep libraries are nice enough to work with that no one really wants to use C++, and that Java developers are going to stick to the plain ole JDK for their apps (either through familiarity or not to sacrifice portability).
I suppose I better expand on my "sudo is a waste of time" comment.
Sudo is generally configured out of the box to allow root access, making it little more than an alias for su. Actually configuring sudo to allow limited access to certain commands is fiddly, and often misses things (try running a root/bin/sh from sudo - works almost every time). Sudo is also a poor alternative to ACLs, or just setting up groups to control access to certain device files (which is often what a presumed need for sudo boils down to). For instance, perhaps you want an unprivileged user to be able to burn CDRs on a workstation install of Linux. Simply create a suitable group and set things up so that the unprivileged user is a part of that group. Then alter the group permissions on the CD burner's device file. This is far more fine grained and easier to configure than sudo.
So rather than needing to record two passwords if your machine doesn't allow remote root access, the keylogger only has to record one. When are idiots like you going to realise that sudo is a waste of time?
Local weather is a fundamental consideration in architecture.
Unless you're one of the "pop stars" of architecture like Gehry. He has consistently ignored environmental conditions when designing buildings, leading to titanium clad buildings that melt asphalt walkways on sunny days and many others that cannot tolerate rain or snow. For those posters who claim that it's the civil engineers responsibility to ensure the buildings are structurally sound, let me just say that when you're working on something as prestigous as a Gehry or Norman Foster project compromise is rarely possible. Often a civil engineer or contractor will point out a flaw in the design only to have the architect threaten to walk away from the project, with possible legal consequences. The sponsors of the project then take sides with tha architect, not wanting to put themselves into a bad PR position. Happens all the time - for example the Gherkin in London where I briefly worked is a Norman Foster ego-fest. Too hot in Summer, too cold in Winter and one of the supposedly indestructible windows fell out not long after completion. Add in the remarkably low floor space inside, thanks to the massive central column along with a large number of interior balconies, and you've got a building that's an over priced status symbol.
Unfortunately Python's threading is a bit of a joke at the moment, but work is being done to sort this out. Until then, Python's performance is pretty poor compared to Java, although if I had to use a "traditional" scripting language Python and Ruby would be my first two choices with Tcl a distant third and Perl a definite no.
Utter bullshit. I develop on NetBSD, using a native build of Sun's JDK from the SCSL sources. I deploy to Linux on x86, x86_64, ppc32 and ppc64 as well as Windows, Solaris and Mac OS X. I've never once run into a problem with portability with my own code, and only once in someone else's code (which used com.sun classes - which is frowned on anyway). Compared to coding in C for POSIX platforms, Java's a fskcing dream when it comes to portability.
Simply don't use anonymous inner classes. They're the kind of stuff (like the Reflection API) that appeals too much to the former Perl programmer in us all. They may seem to be an elegant feature, but they're more of a hack that's been made to popular by the frequency of their use in example code.
NetBeans wasn't written by Sun, they bought out the original developers, by which time it was already a clusterfuck. There's an article by Gosling (can't find the link at the moment, but that's beside the point) where he describes it as a good example of how not to write a Swing app. The trouble is that people see the Reflection API and don't think how inefficient it can be. This is largely Sun's own fault for emphasising the use of Reflection when they started pushing the whole Bean concept.
The answer is not to ship a JRE with your application if it's likely that the user already has one. Alternatively, just ship the most recent JRE, as backwards compatability is pretty darn good with Java and well written code should just run irrespective of whether it was written for 1.2 through 1.6. An example of badly written code is anything that uses the "hidden" sun.com classes, although I have used them myself for signal support.
It contained all of the JFC that was written in Java, the idea being that this was the stuff to look at for best practices in Java coding. The native code was released under the SCSL (Sun Community Source Code License) which is what kick started the Blackdown project and ultimately brought decent Java support to Linux. The native code stuff was largely uninteresting to the average Java coder so I can understand why it wasn't shipped as part of the JDK.
doesn't the US constitution say something about the right to bear arms?
The US constitution declares that states may raise armed militias. This reflects reasons and means by which the US gained independence from Britain, and a desire to not have a centralised government that could ride roughshod over the individual states. This was a careful balancing act to unify the states in a federal system, despite the considerable cultural and political differences between them. However, the passage in the constitution has bee deliberately misrepresented and quoted out of context to suggest that individuals have the right to bear arms. This is not an inevitable consequence of having a militia (such as the National Guard), as such organisations normally have depots for their weapons, to be issued when the militia is called out. It was certainly not intended to allow Joe Sixpack to wander around the US with a firearm tucked underneath his jacket.
Home made guns, or those made in dubious machine shops, have occasionally appeared on the black market in the UK. However, anyone in their right mind wouldn't use them to actually fire live ammunition as the materials aren't strong enough to take the pressure of a round firing without the gun blowing apart. In other words, you'd be better off bluffing with a replica. Machining suitable quality steel for handguns is an expensive business that requires much better tools than you'd find in a typical garage or machine shop.
Slashdot Mirror
I view school as bootstrapping a person to learn how to learn, and for teaching them the things that are timeless.
I heartily agree. I'm largely a self taught programmer (a few evening classes in C and Assembly, that's about it academically), having come from a background of history and languages. I now feel that my degree course was really a grounding in assessing the worth of source material and how to learn rather than having any direct vocational value. I still wish I had done a good CompSci degree instead though!
As for me, if you mentioned Java in the interview I'd charge 50% more. Its hazard pay.
And if I was interviewing you you wouldn't get the job, although as you state your experience is C, C++ and Perl you shouldn't even be in the running for a Java job anyway. If you were a Java coder, but felt the need for "hazard pay" then my instinct would be that your not a competent Java coder. Having done a bit of C++ before I moved from C to Java, I found I had to "unlearn" a lot of C++ in order to fit in with the more Smalltalk or Objectice-C way of doing things.
Java GUI motivates some students to get more involved
I'd suggest making a subset of the projects non-GUI though (in other words, disallow use of the java.awt.* and javax.swing.* classes), just so the students don't become totally distracted by interface issues. I'd also stress the importance of understanding MVC for all projects - separating the core logic of a program from the presentation. This will definitely benefit them if the students become professional Java coders, as I see way too much code that isn't modular and is tied to either Servlets or Swing, thereby hindering testability and reuse.
Having made the transition from C to Java, I don't understand why an experienced procedural programmer would have any more difficulty learning OO abstractions than a college student. The only thing I can think of is that the student is in an environment where they should have the time and resources (including, I hope, competent teachers) to learn whereas a professional programmer may not. This is reinforced by my experience at one company where a bunch of Fortran and C programmers were simply told that they would be programming in Java as of that moment. No training or time to learn, they were thrown immediately into a large Java project which was fairly complex stuff with a large amount of CORBA! Under those circumstances it isn't surprising that the code turned into a mess, but I fail to see how grounding college students in Assembly and C means they will struggle with OO languages later.
As for Java being a bad choice for a computer science course, I agree if it is used as the only programming language. Even if you never plan on coding professionally in a language without access to generic data structures (such as Java's Collection classes) or garbage collection, it is still beneficial to have an idea of what's gonig on under the hood. Occasionally I still write my own data structure classes, even though I primarily work in Java, because I have an understanding of how to measure performance and choose alternative algorithms. Even if I was to only use the Collection classes, an understanding of algorithms is essential (the frequency with which I've seen LinkedList used where ArrayList or HashSet is more appropriate concerns me a lot). And teaching in Java doesn't mean that an algorithms course can't be taught with it - take a look at "Data Structures and Algorithms Analysis in Java" by Weiss for example.
Would you care to list them? Apart from errors caused by people using old ext3 journals after they'd accidentally mounted a partition as ext2 and modified it, I'm aware of no recent reports of file corruption bugs. There was an invalid inode size problem about a year ago that could cause a system hang, but as I understood it ext2 and ext3 couldn't create that invalid size itself.
I've got a Zaurus - the C3200 model - and an Asus Eee. The Zaurus is way better than any phone or PDA that I've used, as it really was the miniature laptop that a lot of people were looking for. However, compared to the Eee it's not so hot. The keyboard is really difficult to use, and the lack of power from the USB port means you need a powered hub to use an external keyboard. The available Linux distros for the Zaurus have small developer teams and are very unstable - they generally turn my machine into a brick whenever I try to configure wireless networking or perform an update. The Eee on the other hand has a usable keyboard, Pentium processor and conventional BIOS. This means a plain x86 Linux distro or BSD will install and run on the machine with no difficulty. The Xandros based distro that the Eee comes with is very nice when you actually use it rather than just criticising it based on the desktop theme as some people have done, and it's easy to strip the machine down if you want to (my Eee now runs NetBSD for example).
Despite his authoritarianism, Fascism still cannot be attributed to Franco. His political convictions were much more in keeping with the monarchist and staunchly religious groups that backed his uprising - groups antithetical to the Fascism of the Falange, which is why Franco was keen to utilise its popularity but sideline its leaders. Fascism, as defined by the its early protaganists, was intended to be a modernising force and as such quite anti-monarchist and anti-clerical. The authoritarian aspects reflected a belief that democracy was too weak to cope with events such as global economic depression, and that a strong leadership was needed to counter Bolshevism. With the Nazi regime colouring our perception of the politics of inter-war Europe, it's difficult for many people to appreciate now that many felt then that democracy was a bankrupt concept, and a clash of ideologies was imminent. It was in this atmosphere that conservatives in Germany as well as Spain allied themselves with the populist Fascists. The difference was that in Germany it was the extreme right wingers that managed to end up manipulating the conservatives.
As for anarchism, my knowledge of that is limited to groups such as the CNT who were so badly treated by Stalin's stooges during the Spanish Civil War. Socialism on the other hand, is close to what I'm currently studying, as I'm hoping to do a part-time masters degree on the Paris commune in the not too distant future.
Franco was not a fascist - he was a conservative with strong religious convictions. He co-opted the Falange movement as a convenient political front for what was simply a military coup. Franco made no attempt to save Primo de Rivera (the leader of the Falange imprisoned and sentenced to death by the Republicans), as it was not in his interest - Franco was an uncharismatic figure, and saw Primo de Rivera as a potential opponent. I suggest you read a good book on the Spanish civil war (such as Anthony Beevor's "The Battle Of Spain") and also look at the post war political history of Spain. You might also benefit from reading about Fascism rather than making the common mistake of confusing it with regimes of Hitler and (beyond the first few years at least) Mussolini. Just as Stalin's regime is not a Marxist one, Franco's was not a Fascist one.
You make it sound like FreeBSD, which scares the shit out of me.
Yeah, it's a bit of a coincidence that Net will be following in the footsteps of Free by releasing a whole load of locking changes as prt of version 5.0. However, NetBSD -current already has the main pieces in place, and is shaping up to be a massive improvement in terms of SMP performance. Check Andre Doran's posts to the tech-kern mailing list over the last few months for the details.
In a sense NetBSD event ran Linux before the Xen support - well, Linux applications at least. There's Linux emulation built into the kernel that allows it to trap system calls from Linux binaries and translate them to NetBSD equivalents. Before the availability of a native Sun JDK, this was the way to run Java on NetBSD.
While the sharing is great and goes both ways, it's a little unfair to imply that all the new features in NetBSD 4.0 came from OpenBSD. The bluetooth stack was written specifically for NetBSD, for example. As for pf, the version in the NetBSD tree has been considerably modified from the 3.7 version, so merges from OpenBSD are made on a fix or per-feature basis as the porting is quite hard to do. Not unexpected, seeing as pf touches some quite low level aspects of the kernel. As for the release, I'll be upgrading my main machine from RC3 as soon as possible, but I'm really looking forward to 5.0 - which will include a new threading implementation along with massive improvements to kernel locking granularity and SMP support.
There was some sort of "developer release candidate", but I can't find it on the Apple dev site anymore, and have seen blog comments that suggest it's been pulled.
So in other words, Apple definitely did something dumb here, but it's not too hard to see what their motivations were.
The deprecation of the Java-Cocoa libraries is linked to Apple's support for Objective-C. At one point Apple was saying that Objective-C would be gradually replaced by C++ at one extreme and Java at the other. However, they seem to have realised that Objective-C and the tarted up OpenStep libraries are nice enough to work with that no one really wants to use C++, and that Java developers are going to stick to the plain ole JDK for their apps (either through familiarity or not to sacrifice portability).
The "w3" web browser extension for Emacs can display images.
I suppose I better expand on my "sudo is a waste of time" comment.
Sudo is generally configured out of the box to allow root access, making it little more than an alias for su. Actually configuring sudo to allow limited access to certain commands is fiddly, and often misses things (try running a root /bin/sh from sudo - works almost every time). Sudo is also a poor alternative to ACLs, or just setting up groups to control access to certain device files (which is often what a presumed need for sudo boils down to). For instance, perhaps you want an unprivileged user to be able to burn CDRs on a workstation install of Linux. Simply create a suitable group and set things up so that the unprivileged user is a part of that group. Then alter the group permissions on the CD burner's device file. This is far more fine grained and easier to configure than sudo.
So rather than needing to record two passwords if your machine doesn't allow remote root access, the keylogger only has to record one. When are idiots like you going to realise that sudo is a waste of time?
Unless you're one of the "pop stars" of architecture like Gehry. He has consistently ignored environmental conditions when designing buildings, leading to titanium clad buildings that melt asphalt walkways on sunny days and many others that cannot tolerate rain or snow. For those posters who claim that it's the civil engineers responsibility to ensure the buildings are structurally sound, let me just say that when you're working on something as prestigous as a Gehry or Norman Foster project compromise is rarely possible. Often a civil engineer or contractor will point out a flaw in the design only to have the architect threaten to walk away from the project, with possible legal consequences. The sponsors of the project then take sides with tha architect, not wanting to put themselves into a bad PR position. Happens all the time - for example the Gherkin in London where I briefly worked is a Norman Foster ego-fest. Too hot in Summer, too cold in Winter and one of the supposedly indestructible windows fell out not long after completion. Add in the remarkably low floor space inside, thanks to the massive central column along with a large number of interior balconies, and you've got a building that's an over priced status symbol.
Unfortunately Python's threading is a bit of a joke at the moment, but work is being done to sort this out. Until then, Python's performance is pretty poor compared to Java, although if I had to use a "traditional" scripting language Python and Ruby would be my first two choices with Tcl a distant third and Perl a definite no.
Utter bullshit. I develop on NetBSD, using a native build of Sun's JDK from the SCSL sources. I deploy to Linux on x86, x86_64, ppc32 and ppc64 as well as Windows, Solaris and Mac OS X. I've never once run into a problem with portability with my own code, and only once in someone else's code (which used com.sun classes - which is frowned on anyway). Compared to coding in C for POSIX platforms, Java's a fskcing dream when it comes to portability.
Simply don't use anonymous inner classes. They're the kind of stuff (like the Reflection API) that appeals too much to the former Perl programmer in us all. They may seem to be an elegant feature, but they're more of a hack that's been made to popular by the frequency of their use in example code.
NetBeans wasn't written by Sun, they bought out the original developers, by which time it was already a clusterfuck. There's an article by Gosling (can't find the link at the moment, but that's beside the point) where he describes it as a good example of how not to write a Swing app. The trouble is that people see the Reflection API and don't think how inefficient it can be. This is largely Sun's own fault for emphasising the use of Reflection when they started pushing the whole Bean concept.
The answer is not to ship a JRE with your application if it's likely that the user already has one. Alternatively, just ship the most recent JRE, as backwards compatability is pretty darn good with Java and well written code should just run irrespective of whether it was written for 1.2 through 1.6. An example of badly written code is anything that uses the "hidden" sun.com classes, although I have used them myself for signal support.
It contained all of the JFC that was written in Java, the idea being that this was the stuff to look at for best practices in Java coding. The native code was released under the SCSL (Sun Community Source Code License) which is what kick started the Blackdown project and ultimately brought decent Java support to Linux. The native code stuff was largely uninteresting to the average Java coder so I can understand why it wasn't shipped as part of the JDK.
doesn't the US constitution say something about the right to bear arms?
The US constitution declares that states may raise armed militias. This reflects reasons and means by which the US gained independence from Britain, and a desire to not have a centralised government that could ride roughshod over the individual states. This was a careful balancing act to unify the states in a federal system, despite the considerable cultural and political differences between them. However, the passage in the constitution has bee deliberately misrepresented and quoted out of context to suggest that individuals have the right to bear arms. This is not an inevitable consequence of having a militia (such as the National Guard), as such organisations normally have depots for their weapons, to be issued when the militia is called out. It was certainly not intended to allow Joe Sixpack to wander around the US with a firearm tucked underneath his jacket.
Home made guns, or those made in dubious machine shops, have occasionally appeared on the black market in the UK. However, anyone in their right mind wouldn't use them to actually fire live ammunition as the materials aren't strong enough to take the pressure of a round firing without the gun blowing apart. In other words, you'd be better off bluffing with a replica. Machining suitable quality steel for handguns is an expensive business that requires much better tools than you'd find in a typical garage or machine shop.