The bias of the "good doctors" is that they are allowing themselves to be distracted by Java's uses and how it is being taught, rather than anything about the language in particular.

Java is being used everywhere, not just in Web apps.

-Tom

If you follow the link to their company's website, you will find this quote about ADA:

"A major improvement [to ADA] is that Java-like interfaces are introduced thereby permitting simple multiple inheritance; null procedures have also been introduced as a category of operation."

'Edsger Dijkstra described those exposed to Basic as "mentally mutilated beyond hope of regeneration"'.

I'd say it goes double for ASP numpties... :-)

More seriously, though, it isn't just the choice of language or how the language is taught. It has a lot more to do with the curiculum as a whole. How much and what quality of Maths are they taught. Do they get taught these subjects (Discrete Maths, Computer Electronics and Programming) in isolation, or are they taught in a way that shows how each afects the other.

Un fourtunately, in many schools, the Maths department teaches maths as though no other subject exists and the Computing Departments and Electronics departments are guilty of the same.

Back when I was in uni those profs lambasted C while extolling the virtues of Pascal/Modula 2.

I would have expected them to at least praise something like Haskell.

Mike

"While the good doctors acknowledge that "real programmers can write in any language", they specifically laud the virtues of C, C++, Lisp and Ada.

All of which makes the criticism of Java somewhat odd. Java syntax is derived from C++ ..."

Actually, no, it doesn't make the criticism odd unless the criticism was specifically about syntax and I sincerely doubt that. You do realize that there's slightly more to programming languages than syntax, don't you? And that beyond syntax Java and C are about as far apart from each other than two imperative languages can be?

It's funny to criticise Java for working so well on the web. Most applications are run on the web these days - Google, Ebay, Amazon, Facebook, Myspace, Wikipedia (can I mention that here?). Modern JITs are efficient enough to allow even the most challenging of applications - real time video editing and playback.

Software is there for its consumers, not its programmers. So let's get away from purism and observe that, in the real world, if you want real time safe, secure and accessible software, Java works best.

Christ, it's spot on. The list of "languages that matter" - nothing missing, in the right order, on the list for the right reasons. You can hate academics all you want but the author of this particular critique seems to have a valid point to make. And yes there's something unique about C as the layer between the machine and the programmer/compiler designer. I never cease to amaze at the foresight (or luck) of those two guys working for AT&T in the mid-seventies.

But what do I know, I just design instruction set architectures, implement them with pipelining, multithreading and support for loosely-coupled memory interconnects, supervise compiler projects serving them and finally write a simple operating system and play a game of tetris running in VHDL simulation once all the pieces are there. And yes I'm even being paid for all that. :-)

I kid you not we have just had a 12 week course on... Haskell.

Talk about a massive waste of tuition fees.

Back to Fortran (which is what I'm using at this very moment).

Yes, java is terrible. So is (IMHO) C++.

People should learn all about command line interfaces before doing much else. Leave that GUI stuff to silly artist types.

The main criticisms of Java in this article would seem to be a mixture of unfamiliarity with the language and poor teaching material.

Unfamiliarity

============

The two statements:

- Students found it hard to write programs that did not have a graphic interface

- its restricted use in the development of web applications

are mutually contradictory since web applications typically run on servers and never have a graphical interface.

Poor teaching material

=====================

If students don't understand the relationship between the source program and what code is generated or can't write command line utilities they simply haven't been taught properly.

====

It seems to me that Java is a nearly ideal first language since it gives excellent training in creating modular, well structured programs. This can be carried forward into other languages. In addition, it has such a rich set of standard classes that non-trivial assignments can be written much faster than would be possible in, say, C.

Teach Java first, then C or Ada and leave C++ for last. If you want students to understand the hardware, then teach them assembler right after C.

I've worked as tech support at a dev base for the HBOS - building of two to three hundred programmers and developers meant I didn't get any software calls as they dealt with them their selves (where feasible, access restrictions do get in the way but those calls I was merely there to type in the password ;) ) but the hardware fault list went through the roof as most of these programmers didn't know the basics of how a computer ran - which always bemused me how the can write efficient clean code if they didn't know how the system works?

But then again I'm just a box swapper and don't code :D

To me the lambasting that keeps on keeping on is more to do with the dilemma of how much a college should be making its students "commercially ready" and how much they should be getting into the nitty gritty computer science. This sort of rant from college people happens on a somewhat regular cycle.

That being said, note that these people have their own language bias towards ADA.

it was the first language I ever learned, on a ZX81. I followed it up with BBC BASIC and 6502 assembly language.

I currently work as both a web developer (using multiple scripting languages) and as an Oracle certified developer, having previously worked with IBM Assembly Language and COBOL on IBM mainframes.

Yep. BASIC sure screwed up my chances of being an IT professional.

Prepare for some fanboyism...

What a load of utter crap. What they're saying is, programmers aren't as hardcore and uber-nerdy about piddly little details as they used to be, and these old-timers don't like it.

So what they're really saying is, someone's made a decent language where you don't HAVE to care about that shite, and that's a bad thing. Well, sorry lads, some of us got bored of having to worry about constant pointer issues and memory leaks.

I've used many many languages over the years, starting with BASIC and Pascal, moving on through web languages like PHP to C++ and finally to Java (via a gazillion others). And of them all, Java is the one that impresses me the most. Contrary to the opinions of these morons, it's flexible beyond belief. Want an app for a mobile phone? Perhaps your wristwatch? A website? Or maybe you'd like to code up a scalable enterprise banking system instead?

Sure, you don't get down to the really hardcore stuff as much, but that's kind of the point of Java - why the hell would you want to when someone's written a perfectly good library already? With the wonders of proper OO, you can just drop in a free-licence library and save yourself a ton of time and effort. How can that not be brilliant?

And how can using a very well structured, strongly typed, flexible language undermine good practice? If you don't follow good practice in Java, your program's likely to be crap. Or not even compile, for that matter.

I'm sorry, but their remarks are nonsense. Especially the one about "not knowing how to build programs without a graphical interface." That one made me laugh. On my University course they taught us how to code servlets and the like before they taught us how to code with Swing.

Truth is, take any student that has sat a "programming course," having never done programming before or outside the course, and they will still know sod all about programming in most languages when they come out. There's only so much you can teach, and there's such a vast amount to learn. Of course they don't know everything - that comes with experience. And obviously the course tutors are going to settle on a language that a) they like, and b) will be useful to the students in the future. And that's Java. (No matter how a certain over-enthusiastic lecturer at Leeds Uni wishes it was Python. Don't worry, we're getting him treatment.)

I wouldn't mind them teaching C++. Really, I wouldn't. It's not a crap language. It's just that...well...Java's better. By miles.

Right, rant over. Back to fiddling with PHP 4 and wishing it was PHP 5.

Edsger Dijkstra was (is? He's dead to me) a narrow-minded, bigoted specialist.

My second programming language was BASIC (Tiny-BASIC, to be precise). Hardware constraints forced me to teach myself assembly language, and, since no compiler available would fit in the memory I had available (this was the Z80 era, 4K RAM was expensive and 16K was luxury beyond belief), I also had to hand-compile it. Prior to that, I had learned enough FORTRAN to make it through my local community college's CompSci 101 course (all programs to be punched on lovely cards, sorted, and run through numerous torture devices before they could be submitted to the High Priests who determined exactly what was to be run, and what was to be discarded, presumably by reading the entrails of a freshman).

I later learned C, PowerBASIC (a lovely compiled BASIC that rivaled C for final code execution speed), C++, and several other "languages" which are Web-related and which I therefor consider less qualified to be called "programming" languages. I've been published, worked in Intel's Network Products Division as a build engineer, and moved on to enjoying the fruits of others' programming rather than writing much more than bash scripts these days.

The leeches deprecating Java are not wholly incorrect, but they have placed emphasis on the language which they prefer for venal reasons, rather than logical ones. Students should learn to program in pseudo-code, hand-compile, and run the results, before they are ever "taught" a high-level language such as Fortran, BASIC, or Java - or Ada. I wasn't fortunate enough to go that route myself, but I was sensible enough to go seek out the knowledge I needed when I needed it. I know from my experience as a supervisor that not everyone will do that.

Naturally, I don't post with my real name.

At last ... java should 'av been char broiled years ago.

OOOh pointer arithmetic is error prone ... lets not doo that!!! but the OS is mostly assembly & C, so are most of the third party apps ARRRGHHHH ... quick JNI to the rescue. That whole pointer thing is a joke, most of optimised assembly code I wrote is not even reentrant.

Wot's happening in the JVM ... who noze??? Lets instrument ... but you'll have to write some C++ to get the full (alleged) benefits of JVM TI ... rubbish!!!

Have a look at the Hot Spot JVM source code ... when I started as a low grade C++ code pee'er I would have been thrown out of the window for writing that kind of code.

Virtual Machine is not new and some I used (Smaltalk & Forté 4GL) were said to be slow ... lightning fast compared with the Gosling T21somic child, they started that crap in the early 90's.

I'll shut up the day a Doom/Quake like game written entirely in Java will perform the same on the same machine. If it's not fit for a million dollar game how can it be for a billion dollar bank app?

Good experiment but please Ditch the stuff!

Very poor form Phil! If you read the article again, it does not attack Java at all. It simply states the fact, obvious to anyone who has tried to hire fresh young programmers with sparkling new degrees, that Java is not a great language to teach the fundamentals in (flamers note: the fundamentals). Neither is COBOL for that matter.

It's not academic snootiness but a simple fact of life. I have found to my cost that trying to alter the mindset of Java-only programmers is likely to be a very expensive re-training exercise.

By the way, neither of the two best programmers I have ever worked with had formal qualifications of any sort and were entirely self-taught. One ended up as the principal partner of one of the most specialised software companies in the Old Dart. This may say something about formal education.

Your characterisation of Tony Hoare as an ivory tower egghead is so far off the mark that I must assume that "you are making with the little jokes, si?". A more practical minded guru there never was.

But keep up the good work writing about languages. It's always interesting.

Cobol? Fortran?

The way basic CS courses are taught are more important than the language itself. On the other hand it is quite hard to motivate students to learn languages such as ADA, Lisp, Prolog and such.

Yet another case of "I hate it, refuse to learn it therefore it is crap"?

Oh, PL-1, wherefore art thou?

Rafael

Based on personal experience working with "new graduates", those who went through the theoretical programming courses (ie, basics of OS, memory management, set theory) where more useful than those who simply picked up a language where all the "dirty work" (garbage collection, memory handling, etc) was done for them. The first group, being cognisant of the "theory" of programming could pick up new languages in a very short time (all programming follows certain basic patterns). Those in the first group were lost once they were asked to move outside of the language they were taught.

And that's the problem: the first group were taught to programme, the second group were taught Java/.Net/etc... Big difference.

Don't get me wrong - Java (for example) is a great language *if you already know the theory of programming*. I remember recoiling in horror when I attended a Java class (i needed a couple of credits to finish my post-grad) and a student asked why Java arrays start at 0... and the lecturer could not answer! Going straight into Java, .Net and their ilk produces a generation of "button pushers" who have no idea of what is actually going on inside the HW (even virtually) and are totally lost once they have to think "outside the IDE".

As far as I'm concerned, don't let *anyone* near 3.5G language until they've learnt the theory...

"restricted use in the development of web applications".

I can only think that they are talking about the core language and not the profusion of APIs and framework that have sprung up around it to simplify web development.

However, learning LISP and ADA are hardly going to help putting graduates in a good position to apply for technical positions on leaving education. Learning a well adopted language in the business sector such as C++, Java and C# will.

And what is wrong with the console for feedback anyway?

I grew up with Commodore BASIC, and learned 6502, 68000 and some ARM and 8086 assembler, before even starting to learn C. So far, I've managed to avoid touching Java too much ... and despite its exposure to several strains of BASIC, my brain is still quite healthy (actually, it made learning Perl a lot easier for me than for my fiancée, who had never touched BASIC).

The real point here is that few computer scientists who now come out of university really understand anything deep about computers and the operating systems that run on them. I know several people who have CS degrees, yet couldn't understand terms like cache coherency, spin locks, endianism or semaphores (nor could they spot the odd one out in that list without having to resort to guessing). Many more have never ever understood interrupts, let alone written code that makes use of them. And we wonder why skills in *really* high-level tasks (like multithreaded programming) is sorely lacking, these days?

If you *really* want to learn computer programming these days, you could do a lot worse than finding a C64 and 1541 disk drive, some blank floppies, a few copies of Input magazine - all on eBay or your favourite auction site - and start coding. I was annoying my A-level Computer Science teacher with advanced answers to system hardware questions, long before I met people coming out of University with CS degrees who didn't even know this stuff. Sorry if I come across as slightly arrogant, but I believe that if you go to University to learn a subject, you should learn about it in depth. Why else would you go to Uni?

Ivory towers or not, these guys are right. It's not just University professors bitching about course content these days, either: Companies have long hated the fact that the average CS graduate comes out of University completely unprepared for the real world (and of course, they have to pay to re-train them). This is partly why graduates from India and China are taking our jobs: They're willing to learn the skills - I can bet you *they* aren't sitting around on their bums all day learning just about Java.

Oliver.

They're attacking Java because it's what all the universities teach; if they taught C#, VB or Python I'm sure those would be under fire. And rightly so: CS students should learn the basics and the basics start with console applications written on *nix in C, using the standard libraries.

OK, so they complain of "the lack of mathematical rigor and formal techniques" and then propose C over Java? Remind me never to bother with any papers from these guys. Oh hang on, I've NEVER come across any papers from these guys before.

Your comment about their criticism of Java's "close association" to web applications is spot on. I'm a strong advocate of Java as a language for teaching computer science, and we never see a single GUI window in our class.

Juxtaposing these guys with Dijkstra and Hoare is an insult to computer scientists everywhere :)

While the syntax of Java is derived from C++, the semantics are very different, and I expect that that is what the two academics are uncomfortable with.

There's nothing wrong with teaching Java - I consider it a good programming language, with a good set of class libraries. But it is a much higher-level language than C, C++ and Lisp (I don't know Ada), in the sense that the language constructs do not map closely to operations in a typical physical machine. Java is designed to run in a virtual machine with complex semantics; a lack of understanding of lower-level concepts would make it hard to understand how this virtual machine can in principle be implemented on common hardware, and thus what the cost in time and space of Java constructs is likely to be.

One of the best books I read at university (early 90s) was 'Structured Computer Organisation' by Andrew Tanenbaum, which presented logic circuits, microcode, several assembly languages, and then C-like languages, showing how each level of abstraction could be built on an implementation in the level immediately below. Java could usefully be taught once these lower-level abstractions are understood. The student would then be equipped to understand which problems are suitable for tackling in a Java-like language, and which would be better implemented in something more like C++, or even machine code.

So, Java is a bad teaching language, but C++ is ok? Or Ada?

If the kiddies are learning bad programming with a lack of rigor, meybe they ought to teach good, rigorous programming. Java is a perfectly fine language for this purpose. If their courses are too narrowly focused on web apps, maybe they ought to change the content of their courses, what they themmselves designed and are teaching. If you get my drift.

The problem is not with java. It is with sausage-factory "higher education" which is nothing but vocational training. Take their money, teach 'em just enough so they can pursue a career, and to hell with learning.

Yes, it all comes down to money, and to the way universities really work these days.

IMO: programming-for-money ought to be taught in technical colleges. It is a trade - you are building something.

Joke: Assembler is the only true language! All hail assembly!

Not joke: BASIC was my first language, as the first computer I used was a PET 2001-N. Then I tried learning 6502 machine language, but it was too difficult without a book.

What is being missed is that the students are not taught design. They have no concept of design, so they blindly write something that kind of fits the language.

Of course basic computer use is essential. Microsoft routinely hires people who can't type, can't use a command prompt, don't know how to use Windows, and can't even hook the computer up to a KVM. This means that their productivity is nearly nil, and constantly complain that tasks are "too hard."

I chatted recently with a fellow who had a professor who taught Java. The prof in question was a Java-worshiper, who was completely dismissive of any language which could access a pointer. These instructors are damaging the students. The real power of a computer is only realized when you get close to the hardware. Everything else eats away at the full power of the system, especially throwing away CPU cycles to a virtual machine.

All languages have good parts and bad parts. What needs to come back into use is domain-specific languages. C and C++ are horrible for GUI work. Java is absolutely the wrong choice for writing a device driver. COBOL would be horrific for sending a man to the moon. Use the language that is appropriate for the problem at hand.

It's not really a withering attack, is it? Not on those responsible for the curriculum, and not even on Java. And I don't think that you'e chosen terribly good examples of ivory tower academics.

The argument that the authors put forward might seem a little out of date - or perhaps it's too positive about Ada, and not positive enough about Java - but the concerns are real enough.

They didn't slag off Java's syntax so why do you mention its similarity to the C/C++ syntax? Their complaint seems to be that Java hides what the actual machine is doing, pointers and so on, which is completely true, and that students don't seem to understand anything other than GUI programs (which seems unfair on the face of it; Java can write bad command-line apps as well as it can write terrible GUI apps).

Syntax is a very small part of languages, and just because Java gets it right by using a good syntax doesn't mean everything else is right with the language.

If you don't like academics in their ivory towers, read what Joel Spolsky said about Java, which seems to be along similar lines to what these two academics have said.

(At university I was taught functional programming followed by C, and then Java on a course that most would consider excellent. I already knew assembler so I guess I didn't need teaching about hardware and pointers but I'd say that, at the time at least, Java probably made sense as a teaching language for certain things, but certainly not everything. Any course which only uses one language is stupid if you ask me, but at the same time I would say that, these days, I'm not sure Java makes sense for anything as there are better langauges with a superset of its features and no relative downsides.)

When I were a lad, we wrote our programs using strange three-letter words and the occasional number. If we wanted a variable we 'ad to go diggin' through the stack. And we thought we were lucky!

Words? With letters? You were lucky! I 'ad to punch 'oles in little pieces of card and line up behind ten thousand other programmers to put the cards in the room-sized machine. And we thought we were lucky!

And you try and tell the young people of today that ... they won't believe you.

java was designed for multimedia TVs not web, it just keeps trying to reinvent itself.

java is a bit like the ugly girl at the dance, wearing one of those pink princess dresses, and unlike a fairytale there is no Godmother to wipe away those blemishes, it is permanent acne.

IMO Python is probably the best language to start on, and to use day to day. C should still be taught, as should C++. The shells are useful at least one should be known. Haskell for the functional. O'Caml is quite interesting, and of course Assembler should still be taught at degree level if not earlier. Ada is worth looking at; good programming practice can be drawn from that language.

The ugly sisters are things like Coldfusion, gopher, and modula, they have very little application, and if you do find someone specializing in them they have either never had an academic grounding in computer science or have never left academia.

I must admit I rather like Perl, but it is a connoisseur language like Lisp, unlike Lisp though a lot of people pick Perl up who perhaps need to start with something else.

PHP is popular but of course the language is not really academic, it is a quick hack really trying to up its respectability, but of course that is the language most ISPs include with their free web space, so there are quite a few younguns who grew up with it. PHP will always be Personal Home Page to me. :)

java is just a verbose mess, at least there exists Jython for those who cannot convince the 'powers that be' that the Java platform is not a smart move. It is especially seen in development life cycle; you could probably fit in three comparable Perl projects to one comparable java one. Python you will probably be looking at 2.5 projects but then at least you can revisit with a bit more ease. If you are developing using java you are probably actually coning the client a bit.

"Wherefore Art, Thou" by Larry Wall of Perl fame, does a good of job of describing the differences between some languages; java is muzak and that is a generous description of it. :)

Greg my dear boy, anyone who gets so worked up about a computer language has to be a great programmer and I’m sure you have built some terrific software in your time. But, psst…, I’ll let you in on a little secret. You know the mobile phone and the wristwatch, which you think you can program so well. You may not believe this - at any rate you tell us emphatically that you don’t care - but some poor sod who actually understands the hardware and all the other “piddly little details” of the device, wrote some nerdy embedded stuff which makes it possible for you and other programmers working at your elevated level to think that you are in control of the thing.

If he or she was good at their work and virtualised the underlying levels well enough, you may never ever have to think about the piddles yourself. But eventually something will come and bite you and when that happens you won’t have the foggiest clue what to do. You know why? Because by your own admission you can’t really program a computer at all. Not really really. You need at least six layers of other people’s work between you and that “sub ax,ax”.

Now, that does not mean that you are stupid or incompetent or in any way inferior. It means that whatever you do, and probably do well, relies on the competence and hard work of many, many other people, most of whom would certainly acknowledge and admire your efforts. Maybe you should return the complement or just think of them sometimes.

And another whisper: the guy who understands what “sub ax,ax” means (and implies) can probably change places with you and write some pretty good Java if he sets his mind to it. Can you do his work as well!?

Speaking as a one-time IT lecturer, I have to say that NO one language is suitable for teaching programming to beginners. All languages have their quirks and the last thing novice students need is to get bogged down in syntax before they've even had the chance to learn the basics.

Back when I was lecturing, I developed a curriculum module for entry-level students called "The Robot's Kitchen". The premise was simple: given a gridded map of a kitchen, with all the ingredients and equipment in specified locations and a recipe for chocolate cake, get a robot to make the cake. The robot can only understand simple commands, like MOVEFORWARD, TURNLEFT, TURNRIGHT, PICKUP etc. as well as IF/ELSE/ENDIF and DO/WHILE/UNTIL for repetitive task optimisation. In this way, I was able to impart the essential principles of programming - complex problem breakdown, function calls, conditional branching, looping and process flow - without the students having to learn the syntax of a specific language or writing a single line of code. Most of the students were amazed at how a half-page cake recipe turned into a twenty-page assignment, but as I explained to them, that's exactly what programming is all about - taking a complex problem and breaking it down into simple, single-action steps.

This paid dividends to the students in the second term, when I started teaching them actual programming, using Pascal to begin with, then C - NOT C++! Because I'd primed the students with programming methods in the Robot's Kitchen assignment, they had far fewer problems getting their heads around the principles of programming than students in other classes. While I don't want to sound like I'm blowing my own trumpet, I should point out that my class consistently had the highest pass rate in the faculty, and that by my second year, new students who had friends in my previous year's classes were asking specifically to be in my class at enrolment.

As to programmers not needing to know about machine micromanagement (memory allocation/deallocation, managing the stack and so on), I have to disagree. As a programmer, you are taking complete control of the machine, and if you don't understand how the machine works, you have no hope of controlling it. Even if you end up as a DBA or frontend developer, you should still know about memory management, stack size and clock cycles because you still need to optimise your code for security, speed and efficiency - and you can't do that if you don't know what makes a computer tick. All programmers, even 3GL programmers, should be able to think in binary and hexadecimal and know the powers of two at least to 65536. They should know about the limitations of stack size and the importance of memory management - the famous buffer overflow exploit is a prime example of what happens when programmers don't understand these basics. That's like needing to know your alphabet before you can read Shakespeare.

Finally, C is probably the best language to teach once your students have a good grounding - not at entry-level. Nearly all other 3GLs are based on C and use its syntax, so if you can program in C, you can easily learn any other language. I don't like C++ because it takes too many shortcuts and is too idiosyncratic. But C enforces good programming practice and forms a solid foundation for the student to build their programming career on.

"Students found it hard to write programs that did not have a graphic interface"

Are they for real? What's easier than System.out.printing to console? Have you ever tried to write GUI code from scratch? Sure, drag-and-drop GUI-building tools tend to come built-in to Java IDEs these days, but it's not a feature of the Java language. Considering how many GUI-building Java tools there are available, and how bloody hard it sometimes is to write Swing code that works, I'd rather say it's more a feature of Java to /obfuscate/ GUI operations!

"had no feeling for the relationship between the source program and what the hardware would actually do"

Well, isn't that the point of Java? "Write once, run anywhere"? I hear that writing generally portable Java code is not always possible in practice, but, as long as the porgram is doing what thought I told it to, I'd rather not care what the hardware is up to, thankyouverymuch! If you wanna get in touch with the hardware, write in C or assembly!

"(most damaging) did not understand the semantics of pointers at all, which made the use of C in systems programming very challenging"

C can't express everyting that assembly code can, but it gives the option to call assembly code from its functions. Similarly, methods in Java may be decalred 'native'; that is, they are implemented in C.

Shock-horror: Java-learned folks come to C and understand nothing of the semantic's of C's pointers (shocking, conisidering Java deliberately abstracts such things), which makes the use of C is systems programming very challenging. Great, why not teach them C then,if the point is to program systems in C?

Java is a different language for a different purpose, and Java can't be blamed for not being able to express the concepts of a language of a lower level than itself.

This article seems quite unfair. For example, it states

"they specifically laud the virtues of C, C++, Lisp and Ada"

but Java is also in the list of languages they single out for praise.

Further, the criticisms they level at Java are actually not aimed at the language itself. What they say is that "Java programming courses did not prepare our students for the first course in systems". The quote about pointers was taken out of this context.

They go on to state

"A further pitfall of the early use of Java libraries and frameworks is that it is impossible for the student to develop a sense of the run-time cost of what is written because it is extremely hard to know what any method call will eventually execute"

Isn't that a fair criticism of Java?

I'd recommend readers to skip the article and go straight to the links, which are quite interesting.

C? Yes its one of those perfect languages.

(COBOL too could be considered a perfect language of its generation.)

But C++? Hmmm no. (See my rant below. ;-)

If you're going to look at a good OO language, Objective C would get my vote. (Think a blend of small talk and C.)

We're talking practical languages here. Not just something to teach you theory.

<begin rant>

But I'd still take Java over C++. Have you ever had to debug C++ written by someone who doesn't understand languages? The issue in the real world is that you're required to turn concepts in to reality quickly and for a lowest cost possible. Java is very portable and you can find "cheap" labor to write something that "runs". Think of it as the AK-47 of the programming languages. Think of C as an M1A1 and C++ as an early M-16.

Its not to say that C++ is a *bad* language, but that its harder to support or find good support and frankly you could pretty much write the solution in C that is more maintainable and less prone to errors.

<end rant>

And everyone who has ever programmed will have their favorite language.

..of course, if you program in a language or problem space that doesn't require in-depth knowledge of the hardware or pointers or something else you probably won't need and thus develop an in-depth knowledge of said thing..

Learning that aspect, however, is, like most things, usually easily done, as and when you need to. If that time is now, for learning from this professor, perhaps his teaching leaves something to be desired?

My programming went from BASIC on various zx81/zx spectrum / BBC micros and similar. A step up from either playing games or typing in page after page of hex digits in a computer magazine, for a machine code game - no cover CD or tape.

Computer magazine's had competitions like "write a program to solve rubik's cube" - contrast that with more modern reg / inq buffoon journalism "I tried to install linux to see whether it's mainstream - I had to select my printer from a list - this is too hard for average users, who don't know what a printer even is. Me, your gifted writer was glad he'd saved the box for his...however my HP 5550 wasn't there, I opted for the Hewlett Packard 5550 and luckily it seems to work..."

At the time, aged 10 to 13, books on machine code flew over my head a bit and the internet didn't exist. Mum dad and "upgraded" comprehensive school maths teachers were more clueless than I was, which ironically is saying a lot - albeit that effect probably explains the myth of the teenage computer whizz. The sad thing for me is, via my own kids, noting how this clueless-about-computing-in-any-respect teacher syndrome has barely changed over 2 decades. Academia is hugely flawed no doubt, but it has nothing to do with java.

A few years later I started at a firm using Cobol...and then got access to some brains that could fill in the gaps such that machine code, and thus pretty much any high-level language and how it related to the hardware made sense. Years later OO languages like C++ became more popular and that was supposed to stymie all existing coders as we'd have to learn something else. Nonsense of course.

Today, finding programming information [even in-depth specs for those old 8-bit microcomputers] can be found with ease over the internet...not to mention one or two more kids, my own included, might have a parent / uncle / older brother that has a clue too...perhaps the professor just has a class full of those people who see computing as nothing more than a lucrative career option rather than a class of genuine wannabee programmers?

"Joke: Assembler is the only true language! All hail assembly!"

I think you may be confusing symbolic machine language with assembler language.

Certainly the IBM System 360 Assembler (and it's presumable Nth generation descendants) was a devil of a lot more complicated than mere machine language. On the one hand, you had to worry about making sure the hardware environment was suitable (all those lovely base registers -- anybody else remember those?) and on the other hand you had a macro facility that was an invention of Satan himself.

[Let me annoy the moderator by discoursing at some greater length on the topic, how to teach programming.]

It seems to me that it's a mistake to think that teaching any one language is "teaching programming." All that does is churn out graduates with a single skill and very little understanding, especially if the one-and-only language is a very high level one. There's no particular virtue in suffering, but exposure to the hardware guts of a computer and to a selection of languages of varying levels of abstraction is really The Thing To Do. (If I use caps like that, can I use the Man from Mars icon?)

The point of doing so is to teach not gruesome details of syntax but rather, and much more importantly, underlying principles. The successful student would be able to relate the rather simple structure of classical procedural languages (Algol, Fortran, Cobol) to all sorts of wild and wooly derivatives: OO programming environments, Lisp, Ada, Python, Java, you name it.

With the right selection of languages taught, the student is then able to get his head around whatever turns up next a lot faster than the poor doofus who thought learning one language was "learning programming."

I've been out of touch with the field for so long I can't begin to even speculate what selection of progamming languages would be optimal for teaching, but I think readers will understand my point.

[Hey, what happened to the Man From Mars icon???? Guess PH will have to do in lieu.]

I am not a programmer, I am a lowly systems admin. But can anyone tell me what high-level mathematics has to do with computer science? Boolean logic, yes. Binary, of course. Anything else? (Not at the physics level of little chunks of electricity and circuits)

Java is the best language to teach students - not only is it good for learning concepts gradually, but in the end you've been trained in a language that has very solid commercial prospects. You can use it for web applications, GUI applications, command-line applications, database stored procedures, business logic, mobile applications, and even Blu-ray interactive stuff.

Languages like C still have their place, of course, but it's a much smaller place - low-level stuff like drivers, kernels etc. I'd include it in a later stage of programming tuition.

Personally, one of the things I hated about C++ is that if I wrote a program using Borland C++, and later had to switch to MS-C++, all the dependencies on the Borland libraries didn't work any more. Oh, and forget about switching to a different platform like Unix. Anything beyond some fairly basic standard libraries is different in each brand of compiler. The shift nowadays is for the language to include a substantial library of classes/routines which is always present whichever compiler or platform you use. So, whether it's Java or C# or PHP, you're not so tied down to one brand. Java is tops in that regard.

It's good for programmers to have at least some knowledge of what goes on behind the scenes when they make use of a library, but on the other hand it's nice to be able to just call a function to read an HTTP connection without having to learn all the details about how to handle proxies, redirect status codes, and all the other nitty gritty.

There are 2 sorts of computer staff.

1. Professionals who understand the theory of what they are doing and are capable of independently learning new skills, and who are capable of making a wide variety of decisions on IT and related fields.

2. Technicians/trades people who understand a language or two and a few specific tools. They are generally better at using those specific tools than the professionals, but their limited range of knowledge limits the range of tasks they can safely do.

Our industry has not recognized these 2 categories yet. But this is similar to accounting, engineering, and medicine.

Yes a professional might use Java. But also a professional will know the theory of what is going on behind the scenes, which you don't learn or think about when you learn Java. Of course you could learn just the C part of Java, and then you could learn behind the scenes, but then you would be learning C not Java.

I believe students should learn the ideas behind the languages, for example by getting an intro into one language of each group. As most languages can be learned in 2 weeks (there are exceptions like C and C++) it shouldn't be a problem.

I'd start with the classical "spaghetti-code" languages like old BASIC or some assembler. It gives the student understanding of the basic way of how computers work.

Then I'd teach stack-based languages. (like Forth) The shock is rather small and gives much understanding for the next group.

Algol-like languages. (Pascal, C) Unfortunately those are now preety much the onĺy thing in use.

Symbolic languages: Since they are so small I'd teach both LISP and Prolog

Object Orientated languages (Smalltalk, Ruby, etc) to show people how object orientated languages work.

Pseudo OOP-languages: Like C++, to show how you are not supposed to do object orientation.

.

It's definitely a bad idea to throw students into C or C++. In my opinion C++ is dangerous to your mental health.

Hear, Hear! Teach programming not programming languages is the gist of what you and the article quoted by Phil Manchester are saying, I think.

By the way, my old Computer Machinery lecturer (may he rest in peace) used knitting patterns from the Womans Weekly for the same purpose that you used the kitchen and cake with your students. (I think I just gave my age away. In those days women's magazines dealt with knitting and crocheting, not with nooking and crotches.) Eventually, as an assignment we had to describe the knitting language grammar in BNF. Ah, the good old days! But what I would have given for the PC I'm pecking this on!!!

I have taught programming basics both in pascal and more recently java. The main reason we deprecated pascal at our department is that the version we had available at that point (an older gpc) did not do good run-time checking. Given that pascal's strong type checking and good run-time checking in the previous incarnation (HP pascal under HP-UX (especially bounds checking (buffer overflow errors anyone))) are extremely helpful to the novice programmer, one of the main reasons for using pascal was lost. A novice might deal with an error message along the line:

Array index out of bounds at line ....

but not with

segmentation fault

which C would produce (if you were lucky). The problem we face teaching imperative programming in Java is that many students treat the problem of progamming as a "cut and paste" excercise. Too often they just look up a suitable class/method and cludge together a program which does sort an array as required, but they have not understood ANYTHING about the algorithms or data structures behind sorting. This is unacceptable. In practice, it is an important skill to use available classes etc, but a programmer, and certainly a computer scientist must be able to develop NEW algorithms as well. This requires a thorough understanding of existing algorithms and data structures. I have actually just developed a new algorithm which performs an image filtering in O(N) rather than O(N^2), and it is hugely satisfying to see computing time drop from 4 - 10 minutes to less than a second on a 3 megapixel image. This may be developed in an "ivory tower" environment, but it will be used (hopefully) in real-world applications. Without a thorough grounding in the basics of programming, including complexity analysis, I could not have developed this algorithm.

I think the auhtors of the paper raise a valid point: we need something which allows us to focus on teaching people the basics of algorithms and data structures. I could imagine using first C and then going on to C++/C#/Java. But only if we can get C to give our students sensible error messages, etc. Of course industry needs people who know their way around the vast number of available classes and methods of, e.g., Java, and it is very nicely portable, but industry also needs people who can develop new algorithms.

" most applications are run on the web " ... well a lot anyway

however most of these applications are not written in Java!.

Outside the "gated community" of the Fortune 500 web applications are written in Perl, PHP, C, C++ in fact were you to name any programming language of the last 20 years there is a 50/50 chance I could point you to a site implmented in that language.

As for the list you give:-

Google -- Mostly C/C++

Ebay -- Java

Amazon -- almost everything that can run on a windows box

Facebook -- PHP

Myspace -- Cold Fusion

Wikipedia --PHP

So one "Java" site out of 6. I suppose you could argue "ColdFusion" is really Java.

Anyway what did these acedemics ever do for us?

Computer Science gets less and less relevent as the technoligy improves. Its as if music was taught as "Resonance Sciences" and aspiring violinists had to do a paper in "String Science" before they were given a violin.

Currently on "Joel on Software" there is an artical advocating that there should be a Bacholer of Arts course in programming and application design where students would spend most of there time programming in the same way that film school students spend most of thier time filming.

Robert Dewar and Edmond Schonberg of New York University are clearly fools. Everyone is taught as they learn Java that it is ia severely limited language in some respects. As people that claim they are academics they are doing nothing to further the cause of students who have moved across disciplines to Java and have found it any easy route into learning development skills.

Such elitist nonsense is typical these days. I am a science grad who is retraining in development- Java has been an easy starting point for me. On my own I have gone on to fill in the holes in my geometry, algebra and low level language skils.

These people are clearly thinking only of furthering their own cause by promoting their own languages of choice. Unfortunately other people listen to them.

people will defend to the hilt the langauges that they were educated with. Personally, I started coding when I was about 8 in BASIC, and only made the leap to C++ about 4 years later. I've done a game coding course and a course in Java since, but I was always a C++ advocate, until recently. It was only very recently that I started using C#, but I really love it. Anyway, that's by the by, since the real issue is NOT if Java is a better language than others, but should it be used as a teaching tool? Personally, Java has an important part to play, however much you like or dislike the language itself, and I have to say that Java has my favourite IDEs over those available to other languages.

Yes, it might be portable, and you might be able to get "cheap" labor to code it, but the truth is that a lot of corporate companies have legacy code in C++ or C, and companies that code in the games sector will more than likely have a very high percentage of code in C++. This would not in itself be a problem, but the techniques that are involved in C++ are easily transferrable to Java. My own experience of this is that I learnt to use Java reasonably well within two weeks or so, and part of this is because I already understood fundamentals of programming at a lower level. I don't think that it would be so smooth the other way around.

Using Java in education is a shortcut, it means that you can skip all those difficult sections on pointers, stacks and heaps, and concentrate on the other difficult concepts such as the basics of O-O coding. Having known people who learnt Java as a first language, I know that a lot of them didn't get O-O, and would write procedural code inside a few objects.

Personally, I think that everyone should learn C first. Not because I think it's a better language, but because without a knowledge of C, you can't possibly understand why a language is needed, or where it fits into the heirarchy or C-style languages, you need to learn concepts in a backwards fashion (such as how parameters are passed internally, and why that int isn't changing outside your function), and also because I think you need a knowledge of procedural before you move to O-O, so that you know exactly what the difference is. After you learn C, and the principals of O-O, it should be reasonably easy to pick up other languages (especially if you start with C++). A lot of languages have a lot of different features, but C++ has pretty much all the features that other languages have in some form or other (pointers, templates/generics, namespaces, almost all forms of inheritance, ability to slip into assembler, include files) and once you have a good understanding of the aspects of C++ you will have a grounding that cannot be matched. Of course, with the advent of C++0x, which supports garbage collection and many other important new features, in the next few years, the face of programming could easily change again.

Do I think that pointers are a good thing? Not necessarily, but the theory behind them makes for intelligent and well-versed programming.

Folks, this isn't about which language is best for production development; it's about which is best for illustrating the core principles of Computer Science. Java is great for certain kinds of high-level development, largely because of the scope of the libraries and development platforms that come with it.

If you actually want to understand what's under the hood, though - which is what Computer Scientists are supposed to do - you'd be better off trying to implement a JVM or JIT in C or C++. And if you _really_ want to understand what's happening, you need to implement a kernel, which can only be done in C with small bits of assembler.

(CS degree, former assembler & C programmer, now using C++ *and* Java *and* PHP, for different jobs)

Like many people born at the beginning of the 70s', my first exposure to computers was through Basic. But *not* the brain damaged basic from microsoft which couldn't even get its arithmetics right, rather a CBASIC dialect from DRI. In many respects, Java is simply a grand-son from that particular kind of basics, with regard to the pseudo-code / VM part of the thing. I had much fun with it, doing some relatively complex things, but I have to say that I really got a feeling of programming only years later. At that time I had already decided to pursue another career path than becoming an IT professional, but I was still hanging with nerdy friends (all were beginning their CS degree). So I self taught myself C and x86 asm just to be able to do their assignements as a challenge. There, I met pointers and my life changed forever. Once you can imagine your pointers dancing in front of you, you can program anything in any language : you're sure to beat the cr*p out of any high-level programmer, in terms of performance of course, but also in terms of internal logic ; your programs are sure to be of a *much* better quality.

I understand the rant of the two professors : java hides the pointers, and it's the worst trick you can play on a wannabe programmer. If you can't tell a value from a reference, you can only produce substandard code. Maybe java is right for seasoned professionals, to help them meet their goal quickly, but seasoned professional intuitively knows what the language is really doing from the look of it. So they feel if a method is actually using a value or a pointer, and pick the right one accordingly. A beginner can't make a good choice while he doesn't knows there is a difference. C is an extremely good language to teach pointers because if you mess with them, they bite you hard.

Today ? I'm still not part of the IT crowd, but I like to occasionaly toy with programming. While I can do decent C, and a few other languages, I've reverted to my old Basic friend since I've met Gambas under Linux. This language despite its quirks can produce amazing results (think of it as java semantics with a basic syntax). The interpreter/VM is not polished yet, but its main author is working hard on it, and it already works extremely well on a 32 bits linux-intel box (next step : 64 bits cleanup).

"no feeling for the relationship between the source program and what the hardware would actually do"

Too true. There are way to many people around today who have no clue what's happening underneath, and who leave cr@p everywhere because the underlying implementation will clean up their garbage for them so they don't have to care.

It's not the language that's the problem, it's the sloppy practices it encourages by hiding the programmers from the consequences. That's why we end up with monstrostities like Windows Vista that won't run in less than 1GB of RAM. My home file server has 32MB (and no Java) and even that seems extravagant... Now where did I leave that Fortran compiler?

Students should learn to program in pseudo-code, hand-compile, and run the results, before they are ever "taught" a high-level language...

This is the most sensible thing that has been said and is also the method by which I was taught. I would venture to say that language is completely irrelevant when learning to program. What makes a good programmer is the ability to solve problems. You can take any language you like but when it comes down to it all you ever do is move things from one place to another and use some basic mathematical functions, this requires sod all so called programming skill and can be done in any language. You will constantly hear, at least I have, that the best programmers have physics degrees and the worst computer science ones. The former discipline lends itself well to encouraging problem solving. As far as I've ever seen all the latter teaches you is how add up using two digits.

In general programmers are an elitist bunch with their heads so far stuck up their own backside that they can't see past their own pet language. They have perpetuated the myth that what they do requires immense skill so much so that they actually believe it themselves. Utter twaddle, any fool can program in any language.

Courses in problem solving is what's needed not a specific language. After all, the Java of today will be the COBOL of tomorrow and programs will still be adding, subtracting and moving bits about.

The best thing about this article is that I can use it to annoy the Java programmers in our department, by emailing them the link and pretending that I agree with the "report".

Such is office life - I'm sure you all have ways of livening up your day too.

Of course it's total bollocks though, because actually they are very skillful guys who are good at their jobs and write nice software.

As arguments go, it's about as useful as the Mac vs. Windows or Vi vs. Emacs debates - i.e. not at all. You might as well print something that says scientists have proved photography is better than painting.

Hi,

I have done a computer science degree and computer systems engineering degree... and High level languages are just too removed from hardware...

So what happens is that coding for performance and elegance goes out the window... where as knowing low level languages requires a good interaction between hardware and software...

When I taught in university I always made sure my students psuedo coded everything... and then used the psuedo code as comments to code... That way in theory they could code in any language.

However that said, what is really lacking in the graduates I have employed in the last 4 years, is a real foundation in language agnostic problem solving! Being able to look at a problem and figure out 2 or 3 approaches to solving the problem.

Changing programming languages should be easy... However solving problems is where the focus should be on.

The second thing I was taught to write in at college (after BASIC of course) was Z80 machine code. That's right, the institution concerned was too tight to shell for an assembler, so we sat in front of the memory map and keyed the stuff in. The 0-9 and A-F keys on the keyboard were well worn and the remainder virtually untouched. If we wanted to write in Assembly language, we did it on a bit of paper, hand-optimised it and then translated it into hex ourselves.

With the benefit of hindsight it was a bit like building a house without using any power tools. Perfectly possible, but you wouldn't want to make a habit of it.

Kids of today <mutter, mutter>.

I accuse Maksim of misuse of the "Joke" icon.....

Cool your egos chaps, use whatever language, tools, or os for that matter, you like. If you and those that pay your wages are happy with the work you do then who gives a flying... Why is it that IT people get so fundamentalist about thier favourite language/framework/os? Different tools for different jobs chaps, diversity is a good thing.

excuse me whilst i strap a bomb to myself and recite my mantra "there is one but language but ...

I sort-of agree with the comments of the pofs.

Personally, I don't like Java, but that's just me - I don't have a problem with anyone using it.

The problem with Java (and indeed any very high level, highly abstracted language like it) is that it's very easy to loose sight of what's actually going on behind the scenes. Which, of course, if this is the only language you are using, you NEVER appreciate this.

If your shiny new program is running in a multi-GHz, Multi-GB PC then this doesn't really matter much. The problem comes about when the hardware you are writing for is not so capable. Embedded systems with limited resources need a much better appreciation of what's ACTUALLY happening, rather than what a cursory glance at the high level source code would suggest is happening.

If you don't realise that your single Java do_everything() method is actually allocating shed-loads of memory and churning through huge amounts of data, then you are going to come unstuck! Unfortunately, there are a surprising number of people who DON'T appreciate this. I see code all the time (written in C) that is desperately inefficient and resource-hungry. On an embedded system (which is what I usually work with), this matters. It matters a LOT. And this is code written in C; a language which SHOULD automatically give you an appreciation of what is really going on under the covers. Just imagine how bad (by 'bad', I mean inefficient and resource-hungry) the code would be if it was written in Java.

And it's not just embedded systems. Why do you think today's desktop applications are massively larger than the ones we had 10 years ago, and despite it not actually doing a great deal more than the old software ever did, and running on hardware that is hugely faster and more capable, actually runs SLOWER?! The reason is, of course, because the code is massively less efficient!

Comments by some people here on the lines of "why use something cruddy like C when you can just click your fingers and Java will do it for you" obviously don't need to worry about running their code on low-powered resource-limited systems. Indeed, some of them almost certainly do not even appreciate the problem. It is THESE people that the profs are referring to - softies that have been brought up on a diet of very high level coding that simply do not understand the full implications of what they are coding. As a result, these people are useless when it matters.

This article was always going to produce a fanboy comment war.

Steve Roper's comment above hits the nail firmly on the head - it's not about whether Java is a good language or nay, it's about whether it's a good FIRST language. It's not, just as a Hummer isn't probably the best vehicle to learn to drive in.

The toolkit approach starts at way too high a level of abstraction. The problem isn't that you can't learn the techniques this way, it's that you start to assume that the toolkit IS the language and thus make completely invalid assumptions as to its capabilities and flaws. In a similar vein, I think that students should have to write and run programs outside of an IDE for at least a little while to realise what goes on under the hood.

Am I showing my age by using that title? If you remember it then you will know the author made some very good point, he favoured FORTRAN over PASCAL but his point stand the test of time. Some bubblewrapped kid can't compete with a dirty handed code monkey. Like him I would not trust a Java Programer to go back to his code 6 months after they last looked at it and expect them to fix/modify it, or worse still have them fix/modify some-one else's code. That would be a job for a real programmer not one of those quiche eaters.

@Trix,

You know the computer games you play. You would be surprised to find many if not all of them have some rotational matrix calculations in there somewhere.

@Martin Gregorie

I want what you're smoking, maybe then I would start to like my employers. Anywho, you'll find an awful lot of java web applications do have a graphical frontend. I'm not saying that Java isn't used as a backend, but I've never seen it (yes, humour intended).

Pseudo-code is by far the best to teach first.

I self taught Basic, then a small amount of assembly. School taught us pseudo as the teachers knew no languages, Best education I've had to date.

Elec degree at uni taught us pseudo, C and assembly. Because that's what we'd use as gingerbeers (Okay, HDLs too).

In current job a firm is being contracted to write fairly simple embedded system (I wasn't allowed, despite the saving it could have made the company, because I would have actually enjoyed it). They tried to insist on writting in C/C++. ???? Why?

I mostly work in C, but there is no doubt that this project should be done in asm. It's just not big enough to need high-level.

So, how about teaching the students a short bit on asm. a larger bit on C and some on a language like Java (about as much as you teach on C). There is nothing wrong with any of these languages (well, okay lots wrong, but meh ;) so let the students learn the differences and where each should be used. Of course, start with theory of structure, modules and pseudo code.

A project on a language of their choice should be given (ie, write tetris, choose any of the taught, or maybe untaught, languages to do so). I think this is what is called an education, unlike what most places seem to give now.

To the above comment, solving problems comes down to extracting fingures and trying to think. Kids of today sit back and expect everyone else to do the thinking (Yep, I'm nearly one of these kids, and I do work with these kids in my free time). They are always given the answer so don't ever expect them to work it out.

Can we have an icon with a teacher/lecturer being spanked please?

"Students found it hard to write programs that did not have a graphic interface"

If someone could find a way of writing a Java program without the need of a graphic interface, students could learn how to program properly. Hang on... I already do program in java and NEVER use the GUI interface.

Its no different to using Visual C++ to build a form based program.

Edsger Dijkstra described those exposed to Basic as "mentally mutilated beyond hope of regeneration"'.

As far as Beginners All Purpose Instruction Code goes, I thought that it was supposed to be simple and easy to learn as a first step on the ladder to programming? Should every 13 year old computer studies student wanting to write their first "Hello World" program jump straight into C ?

I started on LOGO, then BASIC then Pascal and finally C and C++ before learning Java. so does that make me mentally mutilated beyond hope of regeneration?

Probably

The Paris tag is to show what a real mentally mutilated person beyond hope of regeneration really looks like.

Although I wouldn't kick her out of bed!

"I have regarded it as the highest goal of programming language design to enable good ideas to be elegantly expressed.".... http://www.cs.ucsb.edu/~ravenben/papers/coreos/Hoa81.pdf

Amen to that Biggy, Brother. ITs Holy Grail.

"You do realize that there's slightly more to programming languages than syntax, don't you?" ..... Would ConText be the Prime Core Discipline for Real Virtual Drive, AC? If Programming Languages do not Relate to Real Life Conditions they will not Drive IT Forward with Shared IDers* Programs....... NeuReal Agendas in New World Order Programming .....always AI Work in Progress, Intellectual Property Exercise ..... Presently executed Abysmally, probably definitely because it is not exercised at all outside of NEUKlearer HyperRadioProActivity Rings of Steganographic Stealth.

* IntelAIgent Designers

"Naturally, I don't post with my real name." ..Morely Dotes Posted Thursday 10th January 2008 22:48 GMT

It is real enough for work here, Morely Dotes. And such CyberIDEntities as are Created for InterNetional Space Stationery Archives .... Virtual Reality Libraries for Realising Application .... can easily MetaMorph/Phlip Phlop In and Out of Imagined Spaces made Real. Today's Reality is only the Result of Yesterday's Thoughts Shared and Acted upon. Does C stand for C.ore? And C++, C.ore Creation/Immaculate Source. Or would that be a Special Application of Programming 42 Provide Virgin Root for New Growth for a Pretty Good Privacy Perly Gates Python snakeoil of ITs Own .... biding in the Open for Added Security with a Taste and a Swallow for all Languages.

AI Unifying Script for Virtual Machinery, Magical Mystery Turing in Uncharted and Unchartered Waters/Environments.... Private Pirate PlayGrounds?

Hmmm. I suppose UN Charted and UN Chartered Virtual Environments would be a logical Tangent for AI Political Control of CyberIntelAIgent HyperRadioProActivity although that would Create a QuITe Enigmatic Moral Dilemma for IT would surely be asked .."Who is in Control of Whom and Servering to What?" Suck IT and See will Answer that Riddle, Definitively.

To ask the question is to Imagine that IT be True? :-) Indeed, In Deed, it does.

Language is Siloed Hardware which Creates everything Artificial [the things that blight/enhance the Natural Environment .... which we also plant and infest with Seeds and their attendant Weeds. Although such Natural Growth is badly Abused and Used.

Fair point, but surely this comes down to programming ability not the language.

e.g.

the String vs StringBuffer methods in java.

if you append a String to a String then a new String object is created for every append., StringBuffer only uses the one object thus saving (potentially) shed loads of memory.

You can write a poorly running, resource heavy, inefficient program in any language.

(well the ones i have experience with, I know, I've done it!)

P.S.

Java was originally designed to be run on embedded systems (back in the old Oak days). So saying that Java is a web based only language is just plain wrong!

There are many situations where maths is an essential part of the solution e.g. minimal spanning trees, numerical analysis, linear programming and optimisation, goal seeking, signal analysis, Fourier transforms, pattern matching, statistics.... the list goes on and on. Although it is possible to use canned methods it's best if the developer at least understands the underlying mathematics.

But the single most important reason for being a reasonable mathematician is because it makes you think in a certain way about logic, proof and method. This way of thinking is excellent for problem solving in the IT field. I have always found that although non-mathematicians can do a good job, mathematicians can often do it a little bit better.

(And I won't bore you here about the guy who pulled my bacon out of the fire on a tricky printer driver job using advanced trig and calculus. But it happened.)

The actual article is on a "U.S. Air Force" affiliate website. This explains the bewildering choice of ADA as the "best first language" to learn. If you are programming for the military it's the only language they'll use (I've heard). I've never heard of ADA being used in any of the companies I've worked for.

The Stroustup quotes come from "The Power of Ten – Rules for Developing Safety Critical Code", so it seems that these boys are only interested in teaching how to develop safety critical code for the Air Force in ADA, presumably using consultants from ADACore. Their assertion that you shouldn't learn java because it results in an inability to program without a user interface is also a bit inexact. Do you think they actually meant visual basic 6?

Well they've completely missed the point of Java spectacularly; to not be dependent on hardware/platform specific details. If they wanted to give students a grasp of more technical issues, they should get them to write java byte code to execute on the JVM.

They also might want to know about System.out.println("Hello world"); which makes it easy to write non-ui programs.

Maybe they should go back to doing something they're good at; plagiarising other people's material for their own courses (something I've seen done in academia)

High level maths has everything to do with programming. Absolutely everything. Not just the basic language, nor the operations performed but the whole theory of how and why programs work, how efficient they are how they interact with the hardware or the virtual machine is all mathematics.

The code you see may not look that way, but the best programmers are the ones who can understand the maths of what they are doing.

The Cambridge computer science course uses lots of languages for different purposes, but the one that it is used to introduce new students to the idea of computer *science* is ML. This certainly inculcates some useful habits of thought.

http://www.cl.cam.ac.uk/admissions/undergraduate/myths/

"We start by teaching a language that few if any will have even heard of let alone used – so all our students start off at roughly the same level, and we build from there. "

"Teaching yourself to program can lead to your picking up bad habits that will hinder your progress later. In particular, you should avoid languages like C and C++. "

Any sensible uni course teaches a variety of technology and language; mine used Java as the main teaching language (that is, for things like OO concepts, software architecture/frameworks and various concepts) but augmented it with courses using languages like C and a few others for the specifics that Java couldn't cover (like lower level code/assembly, pointers, compiler writing, etc).

It seems they're making a criticism of the language (derived from its real-world usage) when it really should be a criticism of the use of it in teaching, or teaching methods too wedded to a specific language or task area.

Well they've completely missed the point of Java spectacularly; to not be dependent on hardware/platform specific details. If they wanted to give students a grasp of more technical issues, they should get them to write java byte code to execute on the JVM.

In the real world a lot of programming is in a web environment where platform details are explicitly not important.

Scientists in Physics spend a large percentage of their time fixing pointer errors instead of writing code that solves their problems. What an earth has the exact details of an x86 processor got to do with the mass of a photon ;-)?

They also might want to know about System.out.println("Hello world"); which makes it easy to write non-ui programs.

Maybe they should go back to doing something they're good at; spending 'charity' donations plagiarising other people's material for their own courses (something I've seen done in academia) and why I did an applied degree instead of something abstract like computer science.

I was getting a bit worried because some of amanfromMars' posts on other topics were starting to make sense.

Good to see he's back on top form with this one - but what happened to the little Mars icon? I smell a conspiracy.

I remember starting programming by learning Basic on the Amstrad CPC machines. That gave me the fundamentals of programming introducing for loops, while, if statements and so on. When I was at college, they taught Pascal and I loved the language as it was a strongly typed language.

We also learnt Z80 assembler. The way we did that was we were given a problem and we had to make pseudo code in how to solve it. Then we had to convert it to assembler. We had to predict what would happen to all the registers and the stack. We also had I think it was a HP Z80 boards that would input with the built in keyboard. We had to convert the code into the opcodes and enter it in Hexadecimal. The other thing we had to work out was how long it would take. We also had hooked an oscilloscope to watch the pins going up and down when we used the step by step function. I really enjoyed programming from then on because I learnt exactly what my code was doing on the legs of the Z80 microchip (no innuendo intended). Those were the days.

Unfortunately these days, when applying for jobs, they don't care that I used MSDOS or have used assembly language. It has never really helped me to get a job.

: MYOPINION 10 do ." Forth Rocks" LOOP ;

You have spectacularly missed the point of the article, and the following comments.

Programming in byte code would nullify the exercise of learning a high-level language. The argument is, that Java is too high-level, and of course, byte code is too low level. A better way of illustrating what you have said is, teach them to write platform independent code in C (which is pretty straight-forward) so that they know platform-discrepancies and what to look out for, and then show them a platform-independent language such as Java, which circumvents the problems they may have faced writing platform independent code.

In the real world, I would venture that a lot more code is platform-dependent apps and systems code.

"Scientists in Physics spend a large percentage of their time fixing pointer errors instead of writing code that solves their problems. What an earth has the exact details of an x86 processor got to do with the mass of a photon ;-)?"

This is a ridiculous statement, and completely off-topic. The article is based on programming taught in CS courses, nothing whatsoever to do with physics scientists who use coding to make their life easier. They should be using a high-level language, designed for science, and the comment is irrelevant to the discussion.

println has been mentioned plenty already.

It's not plagiarism, it's referencing.

Lastly, why post the post you already posted again? How about just putting the new bit in the middle into a new comment instead.

It's a bit like Javascript isn't it? ;)

Java is promoted by a HW vendor, what do you expect. They argue it's platform independent, but they make sure the platform it runs the fastest on is theirs (which they having trouble with anyway, J6 is sluggish on T2 processors compared with Xeon or Itanium).

So the slower & more bloated the better, it means more hardware.

The next step, I think, is ship some hardware 'JVM add-on PCIe x 256 cards' to boost the slug. Which is going to be the JV(irtually)V(irtual)M, I see the irony comin' up.

Come on it's not only bad teaching wise, it's bad on all counts, except marketing.

Anybody who spouts off about language a being better than language b, needs a bit of a reality check. It's really all about the best tool for the job, and there are times were C++ wins over a managed language like Java and vice versa.

After all, you wouldn't write a real-time rigid-body simulation to u on specialist hardware in Java, would you? And at the same time, you would probably not want to write a web app, or multi-threaded server code in C++.

And as for those of you claiming that a certain programming language makes it easier to make mistakes, I suggest you find another career outside of IT. Nobody's interested in your excuses for your incompetence, least of all me.

having worked at universities for 8 years now, i can safely say that this is the kind of crap academics spout ad infinitum. they are always poo-pooing stuff which they aren't involved in. 6 months/years later, when they are involved, they extoll the virtues of it.

what riles me is that these people are the ones teaching others.

"If your shiny new program is running in a multi-GHz, Multi-GB PC then this doesn't really matter much"

It's this attitude that lies at the heart of this debate. In a world that is trying to reduce power consumption and go-green all resources become important. A faster processor consumes more power and dissipates more heat, all that memory consumes power to remember the 1000s of extraneous bytes of code and data that innefficient resource use requires. Of course, if we used all those cycles efficiently those tasks we still complain are slow would be so much faster and thus use less power to complete but companies like Intel wouldn't be very happy as they wouldn't have such a ready market for their next multi-cored, faster processor 'cause we wouldn't need it!

I have been arguing for many many years now that garbage collection is the devil's idea as it instantly increases a running programs footprint with 'dead' objects lying around waiting to be reclaimed and removes the programmer from understanding and controlling the lifecycle of their data.

I admit to being one of the old breed of Computer Science graduate (not to mention having a childhood writing code to run on 6502s, 68000s and building hardware with my father that needed low-level bit-twiddling control code) who learnt how all these small black plastic bits with pins could be joined together to allow a game such as Elite to run using so few resources on a BBC micro. I'd like to see a modern 'programmer' create a game of such power with so few resources available now.

Rant over! I'm pleased to see that these academics and so many readers here have the same view as me - there is hope for this industry yet.

that's the way it started, and should have stayed. Java descrepit.

Ah, Forth. A language which combines reverse Polish notation with the ability to define new keywords in the language itself. And it was invented by an astronomer to control telescopes. What more could a Real Programmer ask for!

Bear in mind that I'm speaking as a physicist rather than a computer scientist, but long ago (the 1960s) I was taking a programming course. ALGOL, nothing but ALGOL. I noted that I'd rather have FORTRAN. Steam started coming out of the professor's ears.

Academe has a strong anti-utilitarian bias. Always has, always will. "Are we a trade school, then?" they will say. And as a result, they turn out people unfit for actually DOING things.

I think oranges are better than apples because they have more citrus.... really now - ever heard the term Fitness for Purpose? Academia sometimes amazes me...

[And another whisper: the guy who understands what “sub ax,ax” means (and implies)]

It means you'd better hope you're in a 16 bit environment otherwise Murphys law says you're gonna have one difficult to trace bug next time you do anything with eax.

Personally I prefer bitwise operands like xor or and to do the same trick.

Back to the article, Java has its uses in teaching. It's a tool like any other language, and like any tool it can be used for the right reason and the wrong reason. For example I can use a hammer to put a screw in a baton on the wall, but it's an awful lot more hard work than using the right tool, and the result will be somewhat less elegant.

Java is good when you're teaching logic and methods etc - the student can focus on the task at hand rather than wondering why their 4 hours of work just threw a SIGSEV. It's crap for teaching ppl about the link between software and hardware as it purposely abstracts that.

Pick the right tool for the job at hand and you'll get a better result.

And whilst we're randomly ragging on various languages - C++ is an abomination unto the Lord and it's use should be punished with fire. *Lots* of fire. You may as well add PEEK and POKE to Javascript as add OO to C. Anyway, you can approximate classes with judicious use of structs, function pointers and void pointers (oh no he didn't just say that? He did?!)

Of course, Basic screwed up your IT employment opportunities !! Unfortunately, that was saved by your knowledge of COBOL. These days, to get a half-way decent COBOL programmer, one, frequently, has to fetch a Voodoo practitioner to "re-call" them !!

Alas, excessively long hours, loads of black coffee and tons of ciggies have taken their toll !!

*** Possible Rant Warning ***

Personally, I think that a lot of comments here have been way off the mark.

To my mind, as a Computer Science graduate, there is a very big difference between CS and Computing.

I would expect a CS student to learn to understand how the CPU, Microprocessor or Peripheral Device they are coding for works **at the electronics level**. In contrast, I would expect a Computing student to learn to become a code monkey - someone who can code GUI / Web applications but relies on all the other support code (OS, Server, Comms Protocols, Support Libraries etc) to be written by others.

Now, for Computing students who don't need to know the fine detail, Java is probably fine.

On the other hand, a CS student NEEDS to know about and understand things like Pointers, Memory Management, optimisation of Algorithms and Data Structures, Threads and Concurrency (and how to do them properly). Java is not a good teaching tool for these concepts.

Because Java shields the programmer from the 'low level' concepts it is not good for anyone who needs to know and understand the low level - eg. when coding for things like Operating Systems, Embedded Systems, Safety Critical Systems, SCADA, Military Systems, Command and Control etc.

Java has its place - but its place is in courses churning out Code Monkeys, not on serious CS courses.

It's a shame Borland's Object Pascal is all but dead these days. {Turbo,Borland} Pascal 7/8 were fantastic environments for teaching the principles of programming in.

If you've mastered Object Pascal, you can pick up most other languages fairly easily (including, of course, the lecturers' beloved Ada).

"In the real world a lot of programming is in a web environment where platform details are explicitly not important."

...and a HUGE amount of programming is not in a web environment and platform details are very much important :-)

Hows the weather up there in your ivory tower?

"...and a HUGE amount of programming is not in a web environment and platform details are very much important :-)"

... So that makes the case of the web environment a tiny insignificant one then?

"Programming in byte code would nullify the exercise of learning a high-level language. The argument is, that Java is too high-level, and of course, byte code is too low level. A better way of illustrating what you have said is, teach them to write platform independent code in C (which is pretty straight-forward) so that they know platform-discrepancies and what to look out for, and then show them a platform-independent language such as Java, which circumvents the problems they may have faced writing platform independent code.

"

Well I might well be wrong but I've certainly heard of people regarding this as a useful exercise. Never tried it myself though. I agree with write platform independent code in C.

"In the real world, I would venture that a lot more code is platform-dependent apps and systems cod