OOP isn't fundamental, but many concepts that are are missing from your list.
A good course in programming would be one based on learning Scheme to start with. After that, and in conjuction with ML, Lisp, Prolog etc., courses can cover Java and/or C#.
A fine post until it starts turning into C Sharp evangelism. There's absolutely no fundamental OO or other computing concept in C Sharp that represents an improvement over Java. The language itself is embarrassingly derivative of Java, only the VM diverges in moderately interesting ways, although whether these differences, such as the absence of run-time type information, are really improvements is rather in the eye of the beholder.
As for OO itself, the post assumes a degree of acceptance and universality on the virtues of OO that is questionable. OO databases never took off, and OO itself is really rather a simplistic approach to structuring code.
The last thing I'd want is for my kid to be taught that C Sharp represents some kind of revolution and should therefore be regarded as the ultimate in programming languages.
Only slightly worse in that it's more specific to a specific platform and vendor, but more importantly Java and C Sharp are a lot worse than something like Scheme for the real fundamentals.
Switching from Lisp to Java for AI was pretty stupid too. Is it possible that AI is dumbing down?
To learn the fundamentals of programming, it helps if the environment is as open as possible. Seeing resultant machine code, tracing through code in a debugger, dumping data structures in a compiler, running interactively via a REPL - a lot of these aspects are best done with "toy" languages or environments such as DrScheme.
In my day, Pascal (which I think used a compiler from the University of Waterloo), C, C++, Lisp, Prolog and Smalltalk were the mainstream languages and all except the last ran on a positively eclectic variety of machine environments.
Now in other courses where languages are not the focus, C, C++, Java and C Sharp might have a place. However, there is a risk of such languages or associated tools and libraries being tied to particular products. This is not much of a risk with C on its own, only a little more with C++, a risk with Java and a certainty with C Sharp.
Why only a 'risk' with Java? Well, although commercial, Java is fully implemented by a number of vendors - IBM and Sun, of course, but also BEA and a large group producing VMs for phones, PDAs and other devices. When you leave college and start buying products, there's only a modest chance that they'll come from Sun (unfortunately for Sun). More likely is that you $$ will go to Borland or IBM, with some small % going to Sun for things like their certification tools.
There are really two important points to keep in mind - that learning is about ideas, and for computing fundamentals this are best facilitated by specific pedagogical tools; and that academic institutions should be as independent as possible to preserve their own reputations and to avoid arbitrary constraints on a student's future.
I use Java all the time but replacing Scheme with it for AI makes no sense - not much different really from replacing Lisp with Pascal 20 years ago [shows age], and why would anyone do that?
Logically, I should feel that there's a place for C Sharp somewhere in the courses just as much as Java - after all it's a standard, right? But this feels worse, and a successful Mono won't make me feel much better.
I installed Win2K on a recent Toshiba laptop and there were a phenomenal number of reboots involved. First there were the basic hardware drivers (about 5 reboots for chipset, network - not recognised, video, sound, modem and Toshiba's ACPI addon), then SP2, IE6 and about a dozen pre-SP3 fixes - another 4 or 5 boots ISTR.
However, the real killer was that one of these fixes stopped the modem from working. Now if you go to Add/Remove Programs, you'll see that each fix is listed separately with its Q number. With my install, there were about 22 of these and, you guessed, after removing each one you must reboot!
Needless to say the whole thing was a nightmare compared to SuSE 8. SuSE also had the advantage of recognizing my internal network port straight off, so avoiding fiddling around trying to transfer driver files via infrared as I did with Win2K. On the down side, it doesn't work with the internal Intel Winmodem - understandable but still annoying of course.
Americans suffer from an excess of corporate influence on the web as well as anyone else - it's absurd and very counterproductive to present this as a Europe vs. USA thing.
This chap seems to have as much difficulty as many US lawmakers in appreciating the logical fact that you cannot have international cooperation (over DNS or whatever) without ceding some sovereignty - it's impossible to have a net that simultaneously respects a bunch of contradictory rules.
I'm sure many/.ers will then take the libertarian angle and argue that the minimum amount of regulation is fine and that, for example, allocating a lot of top-level domains will allow each country, religion etc. to have their own version of www.truth.com or whatever.
Personally I would prefer some Least Common Denominator regulation of content, practices, privacy etc. as well as raw technical standards, but only on the basis of strict democracy and not via governments - we don't want the Chinese vetoing the Taiwan country domain.
There are a few transnational democratic bodies in the professions, the European Parliament and (sort of) the International Criminal Court. The ICANN successor and related bodies should be elected by users. Nothing could be simpler in practice - it's the principle that national governments might find hard to swallow.
If such bodies were established the real issue is then whether Washington is able to accept any external authority, democratic or not - unfortunately the immediate track record is not encouraging but you never know, on this issue things might work out differently.
This stuff goes way back to before the EU was the EU, or the EC, or even the EEC, so Canada is not really an anomaly. Actually I'm surprised the Aussies and Kiwis aren't in on the act, or maybe they are...?
No, I'm asking for the whole system to be secured. That's where your approach comes unstuck.
Having pointers does not prevent you from guaranting integrity of the application...
You're using integrity in a much broader sense that most people do, particularly with regard to common security exploits. The advance that Java represented over traditional code models has been documented at length by academics - no other language in common use comes close, certainly not Perl, Scheme or Common Lisp.
The bottom line is that Java and Dotnet offer a a value proposition that no security-minded developer or user can afford to ignore, and for practical purposes no amount hand-waving about formal proof systems is likely to deliver anything remotely comparable.
I think you're failing to distinguish between fundamental architecture principles and gossip.
Dotnet does protect against viruses - it allows precisely the appropriate level of privileges to be accorded to something like an email attachment. There may continue to be exploits, but unlike existing systems suffering from buffer overflows etc. the exploits can be fixed in the platform rather than the applications. A secure platform is a whole new ballgame over POSIX and Win32.
XP's own exploits are no more relevant to Dotnet than they are to Java - Dotnet has its own security model, and it is perfectly possible to put a secure environment on top of an insecure one, providing that the secure interface is the only public one.
You still seem to be confused about what a secure platform does. Essentially, Microsoft can ship you any 'binary' and you have the power to limit what it does (TCPA lock-downs excluded, of course - use Java as an example if you prefer). Therefore the issue of backdoors in an app need never arise - just deny it permission to open a socket.
On a positive note, I agree that having the source is the ultimate guarantee, though ideally you want both the source and a secured platform. The amazing thing is that Java has achieved this and put it on a plate for everyone - it's almost impossible to ship a Java application without allowing the source to be recreated - but still people persist in generating user applications with antedeluvian C tools. Java technologies and Open Source goals complement each other in fundamental ways, but the importance and potential of this has never been recognised.
I was hoping someone would step in with these assertions - it's really time such complacency was examined:
1. Java runs on Linux. So it does, and will be well supported not by one product but by three (Sun JVM, IBM JVM, BEA JRockit). What a pity, then, that this commitment by the vendors is not matched by the Linux community at large. No significant GNOME or KDE application is written in Java, and KDE does not even have a LGPL Java binding!
2. Dotnet "being implemented" on Linux. Coincidentally, starting this morning, Dotnet is being implemented by me on my toaster, and I can say with confidence that with regard to the major APIs such as Windows Forms, both efforts have made similar progress, i.e. none. And if Mono / DotGNU does make progress, I'm betting they get sued for patent infringement.
3. User mode Linux 'superior' to Java or Dotnet. This is, frankly, a ridiculous statement. Firstly, Java and Dotnet provide guaranteed integrity of the application, something that's impossible to achieve when you have pointers floating about. Secondly, the rest of the system can be structured into a highly organized form where just the necessary amount of information is shared and rights given. For example, I can apply access controls to individual methods of an application, but a particular application may not itself even be able to read those controls, so it has no idea of who can do what.
The problem is really the same one that has bedevilled Linux already - choice is good when you can make it (KDE or GNOME), not so good when it is made for you via a chain of dependencies (Kmail requiring KDE requiring Qt). Coordination and control applied only to the kernel is no longer sufficient to warrant Linux being called a platform - unless enough momentum is built up behind some combination such as Kernel + IBM VM + SWT + Qt + KDE the platform will become so diverse and complicated that vendors will be unable to port their applications to it.
Good point, but it's perhaps a little optimistic to assume that Linux will have this edge.
If MS provide the tools someone else can easily finish the job, e.g. providing a secure email program with workflow scripting, without imposing nasty stuff like proprietary file formats.
That's right - a dumb setuid program can always be created. However, Dotnet (and Java) represent a quantum leap in the "securability" of a platform and one to which Linux has no answer, so I'm afraid the firmament is decending with undiminished velocity.
...which is one reason why secured platforms like Java and Dotnet are so important to the future of IT.
(The other is the availability of cross [hardware] platform applications - and no, compiling code yourself is not a substitute).
Linux has about another two years to continue poking fun at Windows security defects, then Dotnet will be in place and the party will be well and truly over.
Really? I use Borland JBuilder and WebLogic Workshop - both 100% Java GUI apps, and performance is fine. Actually, JBuilder 4 performance was fine 2.5 years ago on a Pentium 266 with Java 1.2.
And of course on the server side (servlets vs. CGI) there's no contest.
Ah, one of my favourite quotes. Well, it sounds better of course because then you get the full "demented whinge" effect, plus lots of harrumphing "hear hear" agreement from the old farts in the 'audience'.
I think this is a parody of some predecessor to Question Time, a current affairs panel series, or possibly The [David] Frost Show which was inflicted on the US as well.
Actually, the 'high-end features' of messaging and cluster-based replication may not be all they're cracked up to be.
In my experience
pulling transaction records from a remote database results in a considerably superior solution to that obtainable with any messaging middleware product
using duplicated "hot-standby" systems is more manageable and efficient than data replication.
So please don't dismiss PostgreSQL and SAP DB on the basis of checkboxes for features you might not want to use.
Slashdot Mirror
I fear that you are talking gibberish, and random references to Occam, Python and semicolons may not be sufficient to restore your credibility.
For the record, here's how to serialize a Java object to XML (yes, it uses reflection, but Java hides that bit from you):
it's a full blown modern language
Well, as modern as Simula 67 is, I suppose.
OOP isn't fundamental, but many concepts that are are missing from your list.
A good course in programming would be one based on learning Scheme to start with. After that, and in conjuction with ML, Lisp, Prolog etc., courses can cover Java and/or C#.
A fine post until it starts turning into C Sharp evangelism. There's absolutely no fundamental OO or other computing concept in C Sharp that represents an improvement over Java. The language itself is embarrassingly derivative of Java, only the VM diverges in moderately interesting ways, although whether these differences, such as the absence of run-time type information, are really improvements is rather in the eye of the beholder.
As for OO itself, the post assumes a degree of acceptance and universality on the virtues of OO that is questionable. OO databases never took off, and OO itself is really rather a simplistic approach to structuring code.
The last thing I'd want is for my kid to be taught that C Sharp represents some kind of revolution and should therefore be regarded as the ultimate in programming languages.
Only slightly worse in that it's more specific to a specific platform and vendor, but more importantly Java and C Sharp are a lot worse than something like Scheme for the real fundamentals.
Switching from Lisp to Java for AI was pretty stupid too. Is it possible that AI is dumbing down?
To learn the fundamentals of programming, it helps if the environment is as open as possible. Seeing resultant machine code, tracing through code in a debugger, dumping data structures in a compiler, running interactively via a REPL - a lot of these aspects are best done with "toy" languages or environments such as DrScheme.
In my day, Pascal (which I think used a compiler from the University of Waterloo), C, C++, Lisp, Prolog and Smalltalk were the mainstream languages and all except the last ran on a positively eclectic variety of machine environments.
Now in other courses where languages are not the focus, C, C++, Java and C Sharp might have a place. However, there is a risk of such languages or associated tools and libraries being tied to particular products. This is not much of a risk with C on its own, only a little more with C++, a risk with Java and a certainty with C Sharp.
Why only a 'risk' with Java? Well, although commercial, Java is fully implemented by a number of vendors - IBM and Sun, of course, but also BEA and a large group producing VMs for phones, PDAs and other devices. When you leave college and start buying products, there's only a modest chance that they'll come from Sun (unfortunately for Sun). More likely is that you $$ will go to Borland or IBM, with some small % going to Sun for things like their certification tools.
There are really two important points to keep in mind - that learning is about ideas, and for computing fundamentals this are best facilitated by specific pedagogical tools; and that academic institutions should be as independent as possible to preserve their own reputations and to avoid arbitrary constraints on a student's future.
I use Java all the time but replacing Scheme with it for AI makes no sense - not much different really from replacing Lisp with Pascal 20 years ago [shows age], and why would anyone do that?
Logically, I should feel that there's a place for C Sharp somewhere in the courses just as much as Java - after all it's a standard, right? But this feels worse, and a successful Mono won't make me feel much better.
I installed Win2K on a recent Toshiba laptop and there were a phenomenal number of reboots involved. First there were the basic hardware drivers (about 5 reboots for chipset, network - not recognised, video, sound, modem and Toshiba's ACPI addon), then SP2, IE6 and about a dozen pre-SP3 fixes - another 4 or 5 boots ISTR.
However, the real killer was that one of these fixes stopped the modem from working. Now if you go to Add/Remove Programs, you'll see that each fix is listed separately with its Q number. With my install, there were about 22 of these and, you guessed, after removing each one you must reboot!
Needless to say the whole thing was a nightmare compared to SuSE 8. SuSE also had the advantage of recognizing my internal network port straight off, so avoiding fiddling around trying to transfer driver files via infrared as I did with Win2K. On the down side, it doesn't work with the internal Intel Winmodem - understandable but still annoying of course.
Americans suffer from an excess of corporate influence on the web as well as anyone else - it's absurd and very counterproductive to present this as a Europe vs. USA thing.
/.ers will then take the libertarian angle and argue that the minimum amount of regulation is fine and that, for example, allocating a lot of top-level domains will allow each country, religion etc. to have their own version of www.truth.com or whatever.
This chap seems to have as much difficulty as many US lawmakers in appreciating the logical fact that you cannot have international cooperation (over DNS or whatever) without ceding some sovereignty - it's impossible to have a net that simultaneously respects a bunch of contradictory rules.
I'm sure many
Personally I would prefer some Least Common Denominator regulation of content, practices, privacy etc. as well as raw technical standards, but only on the basis of strict democracy and not via governments - we don't want the Chinese vetoing the Taiwan country domain.
There are a few transnational democratic bodies in the professions, the European Parliament and (sort of) the International Criminal Court. The ICANN successor and related bodies should be elected by users. Nothing could be simpler in practice - it's the principle that national governments might find hard to swallow.
If such bodies were established the real issue is then whether Washington is able to accept any external authority, democratic or not - unfortunately the immediate track record is not encouraging but you never know, on this issue things might work out differently.
Do I detect a certain jealousy of our arrogance?
This stuff goes way back to before the EU was the EU, or the EC, or even the EEC, so Canada is not really an anomaly. Actually I'm surprised the Aussies and Kiwis aren't in on the act, or maybe they are...?
Sounds great!
Remind me again, this delivers precisely what benefits over using the well-established Java VMs on Linux?
1) Deliver high media profile for Miguel de Icaza
2) ?
If you're asking for a platform to be secured...
No, I'm asking for the whole system to be secured. That's where your approach comes unstuck.
Having pointers does not prevent you from guaranting integrity of the application...
You're using integrity in a much broader sense that most people do, particularly with regard to common security exploits. The advance that Java represented over traditional code models has been documented at length by academics - no other language in common use comes close, certainly not Perl, Scheme or Common Lisp.
The bottom line is that Java and Dotnet offer a a value proposition that no security-minded developer or user can afford to ignore, and for practical purposes no amount hand-waving about formal proof systems is likely to deliver anything remotely comparable.
I think you're failing to distinguish between fundamental architecture principles and gossip.
Dotnet does protect against viruses - it allows precisely the appropriate level of privileges to be accorded to something like an email attachment. There may continue to be exploits, but unlike existing systems suffering from buffer overflows etc. the exploits can be fixed in the platform rather than the applications. A secure platform is a whole new ballgame over POSIX and Win32.
XP's own exploits are no more relevant to Dotnet than they are to Java - Dotnet has its own security model, and it is perfectly possible to put a secure environment on top of an insecure one, providing that the secure interface is the only public one.
You still seem to be confused about what a secure platform does. Essentially, Microsoft can ship you any 'binary' and you have the power to limit what it does (TCPA lock-downs excluded, of course - use Java as an example if you prefer). Therefore the issue of backdoors in an app need never arise - just deny it permission to open a socket.
On a positive note, I agree that having the source is the ultimate guarantee, though ideally you want both the source and a secured platform. The amazing thing is that Java has achieved this and put it on a plate for everyone - it's almost impossible to ship a Java application without allowing the source to be recreated - but still people persist in generating user applications with antedeluvian C tools. Java technologies and Open Source goals complement each other in fundamental ways, but the importance and potential of this has never been recognised.
I was hoping someone would step in with these assertions - it's really time such complacency was examined:
1. Java runs on Linux. So it does, and will be well supported not by one product but by three (Sun JVM, IBM JVM, BEA JRockit). What a pity, then, that this commitment by the vendors is not matched by the Linux community at large. No significant GNOME or KDE application is written in Java, and KDE does not even have a LGPL Java binding!
2. Dotnet "being implemented" on Linux. Coincidentally, starting this morning, Dotnet is being implemented by me on my toaster, and I can say with confidence that with regard to the major APIs such as Windows Forms, both efforts have made similar progress, i.e. none. And if Mono / DotGNU does make progress, I'm betting they get sued for patent infringement.
3. User mode Linux 'superior' to Java or Dotnet. This is, frankly, a ridiculous statement. Firstly, Java and Dotnet provide guaranteed integrity of the application, something that's impossible to achieve when you have pointers floating about. Secondly, the rest of the system can be structured into a highly organized form where just the necessary amount of information is shared and rights given. For example, I can apply access controls to individual methods of an application, but a particular application may not itself even be able to read those controls, so it has no idea of who can do what.
The problem is really the same one that has bedevilled Linux already - choice is good when you can make it (KDE or GNOME), not so good when it is made for you via a chain of dependencies (Kmail requiring KDE requiring Qt). Coordination and control applied only to the kernel is no longer sufficient to warrant Linux being called a platform - unless enough momentum is built up behind some combination such as Kernel + IBM VM + SWT + Qt + KDE the platform will become so diverse and complicated that vendors will be unable to port their applications to it.
Good point, but it's perhaps a little optimistic to assume that Linux will have this edge.
If MS provide the tools someone else can easily finish the job, e.g. providing a secure email program with workflow scripting, without imposing nasty stuff like proprietary file formats.
That's right - a dumb setuid program can always be created. However, Dotnet (and Java) represent a quantum leap in the "securability" of a platform and one to which Linux has no answer, so I'm afraid the firmament is decending with undiminished velocity.
...or did you realize that? Sorry if I missed some irony in there.
...which is one reason why secured platforms like Java and Dotnet are so important to the future of IT.
(The other is the availability of cross [hardware] platform applications - and no, compiling code yourself is not a substitute).
Linux has about another two years to continue poking fun at Windows security defects, then Dotnet will be in place and the party will be well and truly over.
Really? I use Borland JBuilder and WebLogic Workshop - both 100% Java GUI apps, and performance is fine. Actually, JBuilder 4 performance was fine 2.5 years ago on a Pentium 266 with Java 1.2.
And of course on the server side (servlets vs. CGI) there's no contest.
See the "Petilon" critique.
Personally I think they're pretty even, which is something of an indictment of MS's strategic planning.
Autoboxing is of debatable utility - see this critique.
I think I've found a solution!
Use Jad first to turn the bytecode back into full source code, and then optimize that.
Brilliant, eh? I feel a software patent filing coming on...
Ah, one of my favourite quotes. Well, it sounds better of course because then you get the full "demented whinge" effect, plus lots of harrumphing "hear hear" agreement from the old farts in the 'audience'.
I think this is a parody of some predecessor to Question Time, a current affairs panel series, or possibly The [David] Frost Show which was inflicted on the US as well.
Are you sure? I thought Oracle replication was only for 7.x, and I can't remember any feature corresponding to my understanding of hot-standby.
A link or two might help to justify the rather high score awarded to these assertions.
In my experience
- pulling transaction records from a remote database results in a considerably superior solution to that obtainable with any messaging middleware product
- using duplicated "hot-standby" systems is more manageable and efficient than data replication.
So please don't dismiss PostgreSQL and SAP DB on the basis of checkboxes for features you might not want to use.