That's only partially true. If the language would only have checked exceptions and you would be coding against interfaces (often still the case with dynamically-loaded code), then checked exceptions would still work, whether the code implementing said interfaces was dynamically loaded or not.

Now the problem here is that there are also unchecked exceptions which the code might throw, but this is really unrelated to the dynamic loading. One thing that could break things though, is if you not only dynamically load the code, but also execute said code via reflection. Then indeed checked exceptions might not do anything...

Right, and so many companies actually did with Java. Only with one 'minor' difference: they didn't call it "Java SE".

The problem with MS was not that it was extending Java, but that it was luring programmers into thinking they were coding for the Java SE standard. If you were at a MS platform and only checked your code with Visual Studio, then you as a programmer wouldn't know. Until your users tried to run your Java app on their Linux or Solaris boxes, and found out it didn't work at all.

A very simple solution for MS would have been to provide flags in their environment for "standard compliance Java development" and "non-standard compliance Java development". If my memory serves me well, they had such flags for C++ in their compiler.

The only thing that has remained the same in about 7 years are the posts about "Java is dead", meanwhile Java is still the most widely used language and in those 7 years additions have been made to the language (generics, annotations, type safe enums, etc) and soon we'll have some extra goodies like closures and automatic resource management. Maybe those are not revolutionary, but enough IMHO to keep the language with the times.

Meanwhile, there is a lot of innovation going on in the platform, especially with Java EE and how it uses annotations for things other languages might use keywords for (e.g. annotations to make methods transactional).

Far from it actually... of course Java has had the absolute low level concurrency primitives from the very beginning (Threads, synchronized blocks, wait/notify). More than half a decade ago, the java.concurrent library was added to the platform, which added tons of goodies for concurrent/parallel programming like concurrent maps, blocking queues, thread pools and executor services, cyclic barriers, programmatic locks with timeouts (which actually performed better than the build-in locks based on the synchronized keyword) etc.

Now Java 7 will be extended with the join/fork framework, which is essentially a thread pool and support code for (recursively) computational intensive operations and supports advanced features like work stealing. The join/fork framework has been specifically designed to scale into the many, many multi-core range. Not just quad, hex or oct cores but well beyond that.

Parallel array is another topic on the agenda, which allows you to express in an almost declarative style operations on arrays, which the library will then execute for you in parallel. To really make this work elegantly, closures are needed, which were on and off the radar for the Java SE 7 release. Because of that, parallel array has somehow stalled. Now that closures are back, so might parallel array be, but I haven't heard anything about it for a while to be honest.

This blog post has a nice summary about some of the added concurrency items in Java 7: http://www.baptiste-wicht.com/2010/04/java-7-more-concurrency/

Newsweek has an interesting series of articles on that this week. Turns out there actually is a limit to the Chinese success story. I've read the article "Smart, Young, and Broke" with much interest, which is about a Chinese software developer who's in the same boat as the UK graduates. Graduated with excellent results, yet unable to find a decent job.

Another factor is that the outsourcing of IT jobs assume these IT workers don't want to consume any of the IT goods themselves. I'm not sure this is going to be true indefinitely. If you're working in IT, and you produce for €1 that product people in 'the west' are selling for €100, what do you do when you want that product yourself?

You can't possible buy it for the price it's been sold for, since that's about a 100x more than you get payed. But wait... you know it's actually manufactured for €1, so why not convince people to sell it locally for say €2? Then you can buy it, and those people in the West can also still buy it for €100 and everybody is happy, right?

But then some clever kid will realize that especially software can travel just as easy the other way around, so if you can buy it locally for €2 and those crazy Western people are willing to pay €100, why not sell it to them for say €20? This clever kid will instantly make €18, which would be a lot of money for them, and the Western buyer would still feel he has gotten the product dirt cheap! Win-Win, right?

Eventually this will not work of course. Producing low and selling high will only work as long as the kids producing stuff don't also want to consume, don't want to improve their standards of living. In manufacturing this is maybe possible to uphold, but in IT we're talking about highly educated persons, who have the Internet at their fingertips. They also DO want that iPhone and they DO want this 40" hi-def LCD and they are tired of those crappy low-quality VCDs and prefer those shiny new Blu-rays.

Simultaneously, you have opposite forces working in the West. If you want to sell something for €100, then people you want to buy that product also have to make a €100 at the least. With manufacturing this worked, since the lowly payed jobs were outsourced and the local population was trained for higher payed/highly educated jobs. With IT outsourcing this seems to be the other way around; the highly educated jobs are outsourced and the local population is supposed to take on lower payed jobs?

So the Eastern guy producing the stuff we outsource is going to demand more, thereby increasing the average salary there, while the Western guy is going to have to demand less, thereby decreasing the average income there. Soon, "produce for €1 sell for €100" may not work anymore, since Honghui is going to demand €25 for his work and John will have no more than €30 to spare.

I don't quite agree with your ranting against CS. For starters, I don't really see you mentioning the fact that CS typically has different tracks.

At my university we had two main tracks, applied computer science and theoretical computer science, with the first being further sub-dived in "computer systems" and "software engineering" and the theoretical track being sub-dived in "algorithmic" and "foundational computer science".

In the bachelor part of the education (3 years), you get a mix of subjects from all tracks. The computer systems track will give you courses like computer architecture ,which allows you to read this particular nice book: http://www.amazon.com/Computer-Architecture-Quantitative-Approach-4th/dp/0123704901 and where you write your own CPU emulator. It will also give you the subject operating systems, which gives you time to read this book http://www.amazon.co.uk/Operating-System-Concepts-Abraham-Silberschatz/dp/0470233990/ref=sr_1_1?ie=UTF8&s=books&qid=1278151357&sr=8-1 or this one http://www.amazon.co.uk/Modern-Operating-Systems-International-Version/dp/0138134596/ref=sr_1_3?ie=UTF8&s=books&qid=1278151357&sr=8-3 and typically gives you assignments where you write some kernel module like an IO scheduler or memory manager.

The software engineering track will give you subjects about requirements engineering, software engineering (obviously) and teach you diverse stuff like UML diagrams, development cycles, design patterns, etc.

The algorithms track on its turn invited me to look at a diverse range of algorithms (obviously again), but also to datastructures (how does something like a hashmap works internally, what kinds of trees do we have, what variations on linked lists are there, etc).

The foundational track then let me look at stuff like turing machines, grammars, finite state machines, theory of computation etc. This is the stuff few 'programmers' would study by themselves if not told they should.

Finally, knowledge of several tracks was combined for the subject compiler construction, where you had to write in C a Pascal to MIPS compiler. For this course you needed to have (C) programming skills, enough skills to understand a language you might not know yet (Pascal), understand how a machine works at the low level (registers, assembly, etc) and have some idea about context free grammars.

Now all of this is in the bachelor, meaning all the subjects are basically introductions to their respective fields. You're not a scientist yet if you have completed them. In the Master phase, you choose a specific track to specialize in but you can still take subjects from the other tracks if you want. In my case I choose the computer systems track and learned some additional stuff about grids, parallel computing, software architecture, etc. Now the thing is, you can't really say that CS educates you to become a scientist or not or that CS skills have no practical value if you don't take into consideration the track chosen by the student. Obviously an applied computer science track has more practical value for the average company than the theoretical track, but it depends on what you want to do really.

Most of all I don't agree with your point that CS somehow tried to cram knowledge in the heads of dumb students. Far from it... the way I experienced CS was a period of my life where I was simply allotted time and opportunity to directly dedicate on bettering myself. Classes weren't there to teach me stuff, but to *support* me in learning. Basically what the CS program does is compiling a list of books for you to choose from and a set of assignments to challenge you, and the rest really is up to you. If you don't give it your best, you don't progress much if at all. People who don't realize this will drop out or will be sent away. That last thing is maybe a little controversial, but it does uphold a certain level of quality.

Next to that, CS gives you a base level of knowledge, but at least at my University we were also clearly invited to study and practice other materials outside of the curriculum. If your goal is to become the ultimate über programmer (which is indeed not very scientific), participate in open source projects, take summer jobs that let you do some practical programming. Likewise, there are similar opportunities for those wishing to pursue a scientific career, like participating as assistants in ongoing research.

Currently I'm the lead developer of a team of 9. We're building sophisticated enterprise software where we're dealing with some 500k LOC, a rich portal where customers can subscribe, log-in, see their data, etc and a highly clustered back-end capable of doing thousands of transactions per second. This product makes me responsible for the software architecture, the right design patterns being used at the right places, performance at very low levels of the system, setting up the development process for our team, working with complex and sometimes conflicting requirements, etc etc.

A lot of stuff that I studied during university applies to my job... daily!

No, I don't write my own HashMap implementation each day, but when one day some particular HashMap was misbehaving I found the problems in *minutes* since I understand how these things work. The guy originally tasked with solving this particular issue was staring at the problem for the whole day already and could only slam his fist repeatedly on the table crying: "It's broken I tell ya, it's broken!", but he simply didn't understand the (utterly simply) theory behind it.

I also don't construct new foundational CS concepts, but I do apply the stuff I learn practically. A while back some other guy was wrestling with a huge number of if/else statements and many variables involved. After studying his code for a while, I proposed him to use a finite state machine, since it seemed to apply perfectly and would reduce the clutter and complexity of his code immensely. However, he didn't knew what a FSM was. He quickly learned thereafter and did apply it, but because he had never learned about such things in advance he would not have thought of it himself. Googling would also not really helped him here, since he had no idea of what to Google for.

These are just two simple examples, but there have been many, many more of such occasions.

To sum up, CS is absolutely very much worth the effort. Just don't think you can come in dumb, just sit in class every day, come out smart and don't have to learn a thing anymore for the rest of your live. That's not how things work of course. But if you come in reasonably prepared, work hard, and consider the education as a foundation to continue learning, you'll reap the rewards for sure.

Yeah, all that cycling around instead of riding cars is really bad for the environment. And those windmills they historically used to keep the land (polders) dry... oh man, that really must have dealt some blows to the environment...

Well, I wasn't really asking anything to be honest. It's true that the default collections in Java can not store primitives directly, and thus by extension their generic versions can't either.

Java and primitives have a rather debatable relationship with each other. Some say Java would be cleaner if there weren't any primitives and only their Object versions. Like bit-shifting operators and native arrays, they were included in the Java language for the class of algorithms where simple and direct memory manipulation offers huge performance gains. But Java is not really used a lot for these kinds of niches (C and C++ shine here) so most of the time they feel a little out of place and they often bleed into business code where they have no place really.

Well, in Java the primitive wrapper types are final (can't be sub classed) and immutable, so that takes care of a whole slew of problems. However, incrementing a primitive wrapper type may seem heavy weight, since it would always require unboxing of the wrapper, evaluating the increment expression and then creating a new instance (and discarding the previous instance). In the end though, all that matters is what instructions are eventually generated for the CPU. Because of the strong run-time optimizations being done by the VM, the actual code that gets generated may in fact not at all do this and may actually be optimized into a simple register allocated value that gets incremented, instead of the entire Object manipulation with all the complicated stuff around that.

I agree that the current state of generics in Java won't give you any advantage even if it was possible to declare say a List<int> But just to be sure, let's not confuse that with generics in general (for Object types), which actually do make your code more type safe.

For the love of god, wake up man. Not every web application is Joe's Burgers. Not every web application is essentially a web site with some dynamics. Some web applications are simply enterprise applications where the front-end happens to be rendered via HTML accessible via a web browser. They do heavy computations, processing, reporting, run simulations and yes, they DO NEED concurrency, and yes they DO NEED transactions.

If a web application is "too advanced for PHP" then mark my words "it's too advanced for PHP". The model of fetching parameters from the request at the top of the page, feed them into an SQL query, and iterate over the result set while spitting out some HTML is fine for those 4 page web sites, and maybe even the 20 page ones, but it does not fit every possible application out there.

Are you the same kind of person who walks up to a civil engineer designing that bridge over a large river telling him al his or her schooling and techniques are wrong, and the design needs to be simplified, and that it all can be done in an afternoon instead of 3 years, since that 16 year old kid just threw an old shelf over a ditch and he had a bridge?

And I'm not sure how many times I have to explain this, but there is nothing complicated about USING some of this 'advanced' stuff. Yes, it's quite complicated to implement this under the hood, but guess what, the guys who bring us Java EE implementations have already figured out the hard stuff. I can simply use it, and with 'simply' I really mean 'simply.

But okay, Mr Einstein, let me play along a little. Tell me how 'simplifying' my design magically gives me transactional semantics for non trivial operations.

Good one. Take an app, ANY app. Slap a web interface on top of it, and any part of the app that was CPU bound suddenly becomes IO bound. Really?

Dependency injection is about separation of concerns mister, and has very little to nothing to do with typing.

Which is precisely how it is done in Java, via the very stable and mature Java Transaction API, that one can either use programmatically or declaratively via enterprise bean. Your point again?

Maybe you're referring to some home grown 'transaction decorator' as may typically attempted to be made in PHP by programmers who have a basic understanding of stuff, yet overestimate their abilities and think they can just build transparent transactions themselves. I mean, what could possible go wrong, right? If you're referring to the default declarative transaction support in Java EE, then really this is rarely a problem. If you don't know your stuff and simply do too much work in a transactional method, then you're simply doing too much work. That would have been a problem regardless of using a transactional method or not. Like so many other things, it's a professional tool that's utterly simple to use, but you do have to know what you're doing. Just as you should not blindly make each and every method asynchronous or each and every method synchronized etc.

I think it actually proves you are someone who jumps a little too fast to a conclusion. For the record, I started programming when I was 6 in C64 BASIC, quickly moved on to assembly, then C on an Amiga and later Apple, C++ on an SGI Indy, some smalltalk in between, took up a job using C++/MFC on Win32, in University I worked on a rules engine using Prolog, did some work in Scheme, took up some other C++ jobs, did a bunch of Java work... a while back I did some iPhone development using Objective-C, hacked together some Trac plug-ins in Python, ... and of course in between there's stuff like javascript, HTML, SQL even some XSLT.

Well, you had your opportunity of being downright rude ('too stupid to learn'), let me now take the opportunity to say that this proves you're one of these PHP developers who is apparently too stupid to understand that not everything on this world has to do with the difference between dynamic and static typing or even strong and weak typing. An MVC framework, concurrency, transactions, O/R mapping, etc has nothing to do with this difference. But of course, only a stupid PHP zealot would ever think that.

This is the same old argument over and over again. Facebook and a couple of other large sites building on PHP don't work the way they do because of PHP, but despite being build on PHP. There's a lot of engineering effort put into those sites, with a not unimportant share of that devoted to things having nothing to do with what PHP handles. With such engineering effort being poured in, Facebook could have been developed in assembly or COBOL, neither of which would have been prove that they are inherently suited for such environments.

Also, don't forget that Facebook is migrating to their C++ based hiphop toolchain and have build their own internal framework, a framework that resembles the Java EE stack in a couple of ways and is NOT available to the general public. The kind of PHP that Facebook does is NOT your typical 'phpfreakz' kind of PHP that is most commonly practiced.

I've also personally talked to the lead developer of Hyves -- the Dutch version of Facebook with some 12 million users -- and he said 'PHP is a little shit language' (PHP is een kut taaltje). I understand that out of context this may not mean much, but he did said that. Additionally, Hyves, Tweakers (a Dutch news site) and a couple of other companies I know that use PHP professionally, some of which I know people working there, almost all seem to agree on that it's very difficult to find really skilled PHP programmers. There is an abundance of programmers that are of the Joe's Burgers variety and have hacked on some PHP script, but that's a completely different category really and not at all what a professional organization is looking for.

So, even though a select few companies are using PHP very professionally, the norm in the PHP market is that it's used by people who barely know the difference between a while loop and an if statement, let alone that they know anything at all about data structures, algorithms, or efficiently structuring their code.

And that's precisely where it all goes wrong. We don't need no education anymore, we don't need no knowledge about CS theory, we don't need no transactions, we don't need no unicode, we don't need no MVC... PHP is so easy... a baby can now program any web site. .... Until it all comes falling down.

It's the typical Visual Basic kind of mindset. A mindset about which you can read everyday on thedailywtf. Sure, idiots can produce crap in any language, but for some reason both PHP and VB seem to produce a disproportionate amount of that.

I did okay in school and CS went well too, thank you. I'm currently the lead developer of a 12 person strong development team I've worked with for about 7 years. We're building an enterprise application consisting of several 100k LOC and every day have to deal with an infuriating amount of feature requests that have to balanced out with an ever increasing demand for yet more performance and the overarching wish for our software to be as stable as rock and a perfect accuracy of the numbers we produce for our customers.

I think I have some real world experience

It's a famous fallacy, often uttered by PHP enthusiasts. "Yeah, we may not know about transactions, but we can deliver software fast!". I'm not sure if you have been reading my posts above, but I've explained a couple of times that it boils down to a single annotation in Java EE. There are no months of planning, there are no months of 'considering stuff'. I apply one freaking annotation to my class and it's transactional. This takes me mere seconds! Precisely because the platform already offers so many services and so much structure, I don't have to spent months on designing my own transactional system.

What you are ranting about is not directly about Java, but about 'thinking before your code'. I understand it's difficult for the typical PHP programmer to think a little ahead and that they are keen to hide the gaping holes in their knowledge by crying that the only good software is software that hasn't been thought through a little. But again, this kind of mindset only leads to the kind of stories you find so often on thedailywtf.

How about planning a little and still delivering? What about having a solid software architecture where year after year you can easily fix bugs and implement new features? I understand the typical PHP mindset is about quickly implementing a Joe's Burgers web site and then moving on, but some of us are in the serious type of business were we run for years or even decades on some software platform. You'll be glad then that people put a little thought in the architecture of the system instead of letting a 16 year old kid screaming "we must not think, we must code and push that code into production!!!" do the foundation.

I'm sorry, but I think you're making the exact same mistake, but the other way around; thinking that because Joe's Burgers 4 web page website doesn't need transactions and asynchronicity, no application needs it. That's equally false. Quite a lot of (web) applications actually DO need that, and especially parallel and asynchronous execution is going to be a lot more important now that e.g. the top tier of commodity hardware can give you no less than 32 cores, while cheap entry level servers already feature 8 cores. In general you're not going to keep those cores busy just by means of letting each request go to a core. At some point you have to start dividing the work that needs to be done for each request. Java is ready for this, PHP hasn't even started with the absolute basics.

For starters, there is a LOT more server side software running in this world than just all those crappy CRUD-style websites. More importantly perhaps, you somehow equate 'advanced' stuff like transactions with complexity. But nothing could be further from the truth. Do you know what it takes in Java to make the methods of a class transactional? 1 annotation. You annotate the class with @Stateless, and its methods are transactional. Now you tell me, where is the complexity there?

Furthermore, even simple web apps that manipulate data via a GUI could easily benefit from transactions. If I operate a simple web app that lets my customer say order stuff, I won't really be happy if I have an order in my DB, but nobody to bill. Simple does not necessarily needs to be the same as simplistic. I can build a very simple CRUD style app, but still take advantage of parallel execution and transactions in my persistence methods. If a simple page needs to execute 3 distinct queries in order to get the data it needs to display, why not execute those queries in parallel? It's not the complexity you have to be afraid of, just apply the @Asynchronous annotation to a method and it will execute asynchronously, giving you a very simple yet super powerful way to exploit those cores in your machine.

Excuse me??? Difficult to install Tomcat??? It's a freaking unzip of the archive you downloaded and it's fully operational. Same goes for Glassfish and JBoss AS. You download something, unzip it, and it's there. Or you use your package manager and execute a simple sudo apt-get install tomcat. If that's already difficult for you, you probably shouldn't be in programming really.

My Java apps are not going wrong at all. They're humming along nicely thank you. Now have you ever tried to maintain a PHP app of any non trivial size? I mean, have you really had to? If you had, you'd known about the typical unreadable spaghetti mess that most PHP code so quickly deteriorates to. People may occasionally curse at Java's request life-cycle, backing beans and enterprise beans, but at least my view code, validators, converters, view logic and business logic are neatly in a default location. With PHP, the sky is the limit. It's basically an assembly kind of web language. There is no pre-defined structure, no guide lines, no nothing. Want to dump that critical business logic between the tr and td tag in your view code? PHP happily lets you and worse of all, your peers and the whole community even seem to push you in doing that, or at the very least turn a blind eye or just simply don't care.

If that's the kind of community you want to live in, have fun! But it's not my kind of fun.