What the hell are you talking about ? Your comments are superfically sensible, but on reflection, utterly insane.
> But creativity isn't necessary here. Just implement the function that's needed.
If it's the same function, then it doesn't need to be implemented at all, you just use it again.
If it's slightly different, then the common patterns need to be identified and factored out. Any time you write a new function you need to be creative, otherwise it doesn't need to be written at all. If you're writing a lot of similar functions, then you almost certainly haven't factored out the common stuff, which means that the code is longer and more bug prone than otherwise.
Software should not be comparable to production line work when replication is free. The fact that it is indicates part of the problem.
There is an equivalent to production line programming, but only by crappy coders. Good quality software is inherently creative - a lot of the creativity goes into finding ways to simplify, minimize, and increase flexibility, reuseability, generate good regression suites etc.
> The fact that Chennai's Advanced Information Systems company has achieved the astonishingly low 0.05 per kloc defect rate, and that 22 of the 28 companies with a SEI Level 5 cert are in India, demonstrates....
It demonstrates nothing except that this stupid process has got fuck all to do with producing good quality software. Where does good software come from - more than anywhere it comes from start up organisations who are at "level 1" on the Humphries scale, or from open source developers who would also register at "level 1". The larger and more tightly managed software organisations become, the worse their software. This is no accident - it's because the focus on process results in the actual product being forgotten. Level 1 works, level 5, works, levels 2,3,4 produce nothing of value.
So two bools are almost sensible. Since the Eating&&Sleeping state is illegal, it would be better to represent it with an enum anyway - the lack of enums is one of Java's less debateable bad design decisions. [Yes, I know there workarounds to this ommision but they suck].
As a Brit, I would normally side with Brits in a US v UK contest, but I am a nerd by choice, and a Brit by accident of birth, so this wasn't such a hard choice.
You don't need to be a genius to write utterly incomprehensible perl code.... Bah, this comment is too obvious. Look, I'll start again.
I would like to see a contest for programs that are readable and work even though they shouldn't, like if you create a c program which includes a bunch of macros
#define like {
#define ok }
#define um ;
#define int dude
etc
then you can write fairly serious programs in valley speak. I started doing this once, but then I got a life..
If something is moving away from us at nearly the speed of light (say.9c), then the light from that object still moves towards us at c, rather than.1c as you might expect.
Still, I agree that it is funny how people always assume that the observed limit of the known universe is the actual limit of the universe
This result doesn't change my understanding of the universe at all - I don't pretend to have one - unlike many cosmologists.
It's funny how little cosmologists confidence in already knowing everything (like age & size of universe) was shaken by the fact that their most fundamental prediction of the behaviour of the universe was completely wrong. (The fact that distance galaxies are accelerating away from us as opposed to decelarating)
You do get more than twice as much work done working 70 hours a week than 40.
He's also right about productivity. The best programmers do get 10x as much done. The exceptional programmers typically get 10x as much done as the good programmers.
Companies try to listen to this advice, but they never take it seriously. They think they can get really good programmers by paying 1.5x as much and buying a few toys. Unless they are very lucky, they just end up with average programmers like everyone else, and maybe a couple of engineer-savants (great programmers with no common sense).
The fact is that no companies will pay them 10x as much. In fact, it's hard to find a company that will pay 2x as much. Say an average developer can make $50,000 per year working 40 hours per week. If paid according to achievment, a really good programmer working 70 hours should earn around $15,000 PER WEEK. There's no way on earth that traditional companies will pay that much - the director of HR would go mental if they found out some hacker at the bottom of the management chain was earning twice as much as they were.
The only way for really good programmers to make the kind of money they are worth is by working for a start-up where they can have a decent equity slice. Then they'll work 70 hours + per week. Otherwise there's no point unless the work is *very* interesting. Contrary to popular opinion, the very best programmers do have a life. It's the 2nd tier geeks who happily work 70 hour weeks as a matter of course and spend the rest of their life playing video games.
If companies took the hyper-productive programmer thing seriously, the job market for programmers would look like that for sports stars. Until companies are prepared to pay those hyper-productive programmers enough that they could work for 6 months and then spend 6 months doing whatever the hell they like. In the mean time they just pretend to themselves that they have the best people.
> If the master wants to code in 'vi', who's to
> stop him? If the punk intern wants to do all his > work in Rational Rose, he's wasting time, and
> it's the project manager's responsibility to
> assign him to something small enough that
> he can learn good practice through practice.
That is pure poetry. It's also an accurate example, the master often does want to code in vi while the punk intern pisses about with Rational Rose, generating 1000 pissy ill thought out classes.
College is a good place to learn because you spend 4 years getting laid and drunk and puking up in flowerbeds, on the basis of which you will be better paid for the rest of your life. Sometimes you have to go and listen to pointless drival and you have less money than you would in a job, but it's more fun anyway.
Be wary of "formal development" - don't think there's anything magical about it, some people just prefer following mindless procedures to doing anything useful. These people manage to give off the impression that they're doing something important and "professional" but rarely produce anything. Developing large projects *is* difficult, it requires experience, intelligence and most of all an insight into the problem that takes a while to obtain, but doesn't come from any "methodology". There's no subsitute for understanding.
Become a good programmer first. Write your own lisp interpreter in assembly before you learn UML. Invent your own design patterns before you read about other people's.
The argument "how will these people support themselves", we must have laws to protect the income of these people from the free market.. is a socialist argument. That's what unions, price-protection schemes, minimum-wage etc are all about. We must have government laws and enforced government control of everything... It's hilarious - you think you're fighting the capitalist/socialist argument and it's *not* related, but if anything you're arguments are the wrong way round.
I'm a software developer, trust me, we'll manage just fine... We won't get as rich as Gates, but then again we weren't going to be anyway.
Your ideas about money and property are antiquated and not clearly thought out - I know them well, I would have agreed with you entirely two years ago. This may seem ridiculous to you but in 50 years or so, people will look back and laugh at the ridiculous efforts people made to ensure false scarcity of "their" information - a classic prisoners dilemma. If everyone on the planet had access to *all* information on the planet everyone would be better off.
Then there is the "who would create it" argument, which is also bogus. People (in "advanced" world) don't work for material wealth - they work for status. Does a lawyer really work 70 hours a week because the latest model mercedes is so much more comfortable - of course not. The only reason he cares that he's driving the latest model mercedes is the status it affords him in his brain-damaged world. In an IP free world, status will be obtained by creating useful stuff and having it used by as many people as possible.
The difference between a "bit" and an "it" is disappearing. I'm a little ahead of the game here - I used to work at 3Dsystems where I could download a vrml model off the net and print it out. As technology advances, nanotech etc, almost the only type of wealth that matters will be IP. Our model of an optimal economy is stuck in a local minima at the moment. Capitalism is a good system for optimal distribution of scare resources but people are too stupid to notice when the model is no longer valid. The internet provides enough of a jolt to push us out. Various misguided selfish interests will try to keep humanity stuck, but I don't think they'll win.
Nonsense. People work longer hours due to stupidity not technology. They need to work longer because they're not doing anything useful anymore - they're sitting in meetings discussing mission statements or ISO9000 compliance tests or a million other worthless activities. Dilbert is entirely accurate. The more extreme useless people use technology to generate anti-work more efficiently than ever before, but the root cause is addiction to ritual, not technology (www.reciprocality.org). In fact it's only technology that prevents the whole edifice from disintegrating and snapping people out of their stupors through economic collapse and eventual starvation. Left to their own devices the majority of humnity will slump into an eternal ritual where every moment of their lives is entirely predictable from the greetings that their co-workers give as they walk into the office to the format, outcomes and even dialog of their favorite shows. As it is, the economic surplus provided by increasing mechanisation nicely matches the increasing stupidity and worthlessness of humanity. We're already living in a post-scarcity world (in the west at least), the wheels would keep turning just fine if only 10% of the population worked. People don't want more leisure time though, they want to keep their minds in a state of minimum utilization, so everyone spends more time at work performing pointless rituals and spends the rest of their lives watching predicatable television. Anything unpredictable must be destoyed - such as children who don't follow the rules - that's OK though, there are drugs to cure that (ritalin).
I did AI at Edinburgh, and had Mitchie as a lecturer as few times. I would have actually turned up and listened more frequently if I had known his past - as it is I spent 2 years down the pub, followed by 2 years programming in the bowels of the machine rooms - they actually gave undergraduates keys to the machine rooms full of sun workstations so you could work all night if you preferred (which all the serious guys did - you get more resources at 3.00am).
I write software for these systems - (former employee, now consultant). It was fun to run a test overnight, come to work in the morning and have 14 little resin models of the starship enterprise waiting for you. The algorithms are fun - similar requirements to computer graphics, but interestingly different (I got about 5 patents for 3DSystems, back when I was young and naive and didn't understand that software patents are evil...). Good times, nice high end SGI boxes to play with (my Linux box has only just caught up with the system I was using 7 years ago - coppermine,GeForce, SuSE 7.0, and open source OpenInventor feels very similar to R10,000 with high-impact graphics card, but that was $50,000 worth not $2000) - fun job, but California is too damn sunny for my tastes, and Dilbert was too accurate to be funny back then.
You could use it for everything from building someone's skull from an MRI scan - so the brain surgeon would know whay to do, to building models of the internals of Cruise missile warheads (single material, no moving parts though - a bit like the T1000 (terminator 2) but much slower), to Hawaii Barbie...
There are several types of machine, accuracy tends to be about 5/1000 of an inch (injket based models do about 300dpi, but downfacing surfaces are shitty) takes about 10 hours for average part.
It would be cool to use similar process on a larger scale for building office blocks.
It was fun to find VRML models on the net, run a little conversion program and print them out. Another neat trick is to get yourself scanned and make a personal voodoo doll, but the software to fix the model is kinda tricky.
You may get away with 20% when you're brilliant at it. For now expect to be 90% manager, ie forget about doing any useful coding for the moment. The best thing about this move is that if it doesn't work out - it'll make you a better engineer - you'll have more sympathy for good managers, and more power against bad ones.
Read "the 5 minute manager", "7 habits..", "mythical man month", "principles of software engineering management" (Tom Gilb), "the prince", "the art of war", "time management for dummies" and neither last nor least: http://www.reciprocality.org/Reciprocality/r0/inde x.html
As someone mentioned earlier - the hardest thing about this is that it requires an orthogonal mindset. People who are good at focusing intently on a single task for months and getting deeply stuck into a problem ("ADD" people, strangely enough - the best programmers), are *absolutely crap* at keeping track of 100s of trivial tasks. The necessary skills can be learnt by a tech-head, but take it seriously. It's at least as difficult for a tech-head as say - getting a physics degree.
Well, we'll see where the two systems are three years from now. Solaris *is* a good OS now, it's just none too quick (TCP/IP stack aside). I'm still bitter about ancient history. Solaris was bloody horrible until around 2.6, by which time I had abandoned Sun and moved to SGI. Sun shat on any customers who didn't want to move to slow, buggy Solaris as opposed to nice old BSD based SunOS. And if you had been trying Java on and off since before 1.0 was released, you would be pretty sick of it by now too. 10 years ago, Sun was leap years ahead of what you could find elsewhere - dbxtool came for free with the system - they were just fucking great machines. Now, they're great for an enterprise server, but having a personal sun workstation just isn't the intellectual status symbol it used to be. If they hadn't let their marketing department take over their software direction, life would be better and I would still be using Suns.
Yeah exactly. Sun make good hardware, but their software has been going backwards ever since they switched from SunOS 4.1.2 to Solaris, and made the c compiler an (expensive) optional extra. What difference does their endorsement of Gnome make to anything.
If I were a Gnome supporter I would be more worried than jubliant about getting sun's endorsement - look at the wondrous advancement in CDE and motif since sun got on board. And don't even get me started about java...
KDE2 on the other hand actually provides a browser that seriously competes with IE5.5. I wouldn't hold my breath looking for that in the GTK/mozilla world. I'm no great fan of MS, but they were right about one thing: netscape engineers *are* weenies. It'll be a long time before mozilla overcomes that background, and in the mean time, there will be no decent Gnome browser.
I think you've missed my point slightly.
I agree that GUIs are good for computer->human
interaction. Computers are good at generating images.
> However, language was designed to work with certain hardware limitations: humans can't display imagery as complex as a computer in real time.
That's my exact point. What I'm saying is that the most powerful effective mechanism for humans to tell computers what to do is through language.
"Do I have to draw you a picture"
Saying that in order to use a computer really effectively, you need to be a programmer (to a certain extent) - it's not like saying you should be able to fix a car if you're going to own one - it's more like saying you should learn to drive.
Gui's generally present the user with a small set of choices to make at each point. Your example of choosing a color is a perfect example of what GUIs are good for. The power of language is that the range of choices grows exponentially with the length of the expression.
In the starkest terms: excessive use of GUIs reduce your Turing machine into a finite state machine. The only reason this is considered remotely acceptable is that the bulk of humanity thrives on pointless repetitive ritual. They would rather spend their days repeating almost exactly the same thing than have to stop and think occasionally. The emphasis on GUIs allows people to stay comfortably in a semi-conscious trance relying on their mechanistic mental facilities to get them through the day - using language effectively requires a capacity for inductive thought which many people prefer to leave dormant.
The job of programmers is to bring the power of the computer within reach of the user. Not to tame it by pretending it's a glorified bloody toaster, or to waste their time by demanding they know a bunch of esoteric details (like printcap files) which were introduced for the benefit of the programmer. The best example I know of where an application got it right is autocad. Where I was using autocad the users were all draughtsmen. The best user/programmer there had left school at 15 - he learnt what he needed to know on the fly. You're quite wrong thinking that autocad worked because it's users were engineers. It worked because humans have brains, and with a little encouragement they'll use them.
I agree that GUIs are much more intuitive for some things. I think most computer->human interaction should be GUI based. But I also think the focus on them as the primary mechanism for Human-Computer interaction has held us back tremendously. When we stop pretending computers are something else, things will improve.
This might sound ridiculous now, but it will be blatantly obvious in 20 years. (I'm an arrogant son of a bitch, to be sure).
This daft window metaphore crap is largely responsible for reducing the most powerful invention of the previous century to just another time-wasting useless idiot-box where humanity's natural sloth and aversion for thought can reign supreme.
"A picture is worth a thousand words" - it's a myth. Draw me a picture of "love". The trouble with GUI interfaces is that they are predisposed towards the computer transfering information to the human. They are not an efficient mechanism for humans to transfer information to the computer. There is a good mechanism for this - one that has been used for millenia and which are brains have even evolved to use effectively. And that mechanism is language. This is why text based interfaces will always rule over this GUI shit. The "integrated" desktop is doomed to forever strive for the level of power, speed, simplicity and componentisation already provided by the shell tools.
Why use a metaphor for a work-style whose time has past (WIMP - "desktop") when the reality of the computer is orders of magnitude more powerful and more flexible. Even as programmers deny this, and preach the religion of GUIs they implicitly acknowledge that to do anything powerful, you need to use a language - a programming language.
"Oooh, but that's too complicated for my users..."
Am I saying that every computer user should be a programmer ? Yes I am, but only to the extent that it makes sense in the domain. For instance, any real power-user of autocad will write small lISP programs. How could a commerical company make a fully functional, mathematically pure programming language as the basis of an interface for a program intended for draftsmen ? Because they understood that stupidity is like work - it expands to fill the space allocated to it. Expect intelligence and you will receive it. Autocad dominated their market like this. This approach would work in other markets if programmers had the humility to admit that their "users" were as smart as they were, and deserved the same power that programmers reserve for themselves.
If I want to do GUI programming (which *is* good for computer->human interaction), I'll use fltk, a far more powerful and better designed toolkit than either Gnome or KDE. Plus it's cross platform, and LGPL. If I want to knit things together, I'll use a bloody pipe.
I have been wanting this for so long. I met the OpenInventor team about 5 years ago - smart cookies. I've been begging SGI for exactly this every year for the past 4 years.
They did an incredible job - it's one of the best designed C++ libraries I've ever used. It's not suitable for games but for putting good 3d interactive graphics in a professional program without spending *years* learning the details of OpenGL it is wonderful.
This is the one move that SGI could have made to prevent Direct3D winning the 3D api wars. Scenegraph based applications are an order of magnitude easier to program.
Thanks SGI - I might even buy another SGI now (I know SGIs came with OpenInventor - but it wouldn't work gcc before)
gtk is the obvious choice for anyone coming from a motif background. Nothing else is sufficiently awkward to program with - it just wouldn't feel natural.
is almost ugly and verbose enough to make a motif or X programmer feel comfortable. Plus GTK has got it's own home-grown C based OO model, just like X, so you can get all the complication and grief of OO techniques without having to deal with any of the syntactic sugar like operator overloading or templates which can (but normally doesn't) make C++ readable. Instead you can have a C++ wrapper that never quite works right stuffed on top of a subtly incompatable object model. It really is the perfect choice for motif converts.
Disclaimer: it might have improved, it is 6 months since I used it.
Reliability was dreadful, lots of functions don't behave as logic dictates they should, several things plain broken. It takes ages to find these problems and work around them. Your prototypes look good, but then several months down the line you find peculiar bugs popping up - for instance pop-up menus that occasionally just decide to stay visible for ever.
Complexity - huge amount to learn
String label = (String)JOptionPane.showInputDialog(frame,
"Input label for ",
"Info",
JOptionPane.QUESTION_MESSAGE,
null,
null,
field.substring(i));
what are all those args for - I can't remember just now, but you need to know it.
Verbosity - it took me 5000 lines of fltk to replace 25k of swing.
Performance - this was the worst. Memory footprint was horrendous, CPU useage was bad. Stick a breakpoint in and examine the call stack sometime - it'll be at least 18 levels deep. This is a sign of shitty design, no matter what anyone tells you.
If you're writing GUI apps in C++, fltk is the way to go. I've used X, motif, MFC, GTK, Qt, Java Swing, and god knows what else and the only pleasureable experience was with FLTK. It's very fast, very light, very simple, just lets you get on with it.
Qt was pretty good - close second, but it's not LGPL and you need to pay for the windows version.
Fltk is just cleaner. MFC was the worst, closely followed by swing.
It's not that simple. You get most of that energy back. The stuff closing in behind you can push you forwards, (and it's not that simple either).
You could use a similar calculation and prove that tuna couldn't swim that fast without an onboard nuclear reactor. But they do, and you don't need a geiger counter in a sushi restaurant.
Slashdot Mirror
What the hell are you talking about ? Your comments are superfically sensible, but on reflection, utterly insane.
....
> But creativity isn't necessary here. Just implement the function that's needed.
If it's the same function, then it doesn't need to be implemented at all, you just use it again.
If it's slightly different, then the common patterns need to be identified and factored out. Any time you write a new function you need to be creative, otherwise it doesn't need to be written at all. If you're writing a lot of similar functions, then you almost certainly haven't factored out the common stuff, which means that the code is longer and more bug prone than otherwise.
Software should not be comparable to production line work when replication is free. The fact that it is indicates part of the problem.
There is an equivalent to production line programming, but only by crappy coders. Good quality software is inherently creative - a lot of the creativity goes into finding ways to simplify, minimize, and increase flexibility, reuseability, generate good regression suites etc.
> The fact that Chennai's Advanced Information Systems company has achieved the astonishingly low 0.05 per kloc defect rate, and that 22 of the 28 companies with a SEI Level 5 cert are in India, demonstrates
It demonstrates nothing except that this stupid process has got fuck all to do with producing good quality software. Where does good software come from - more than anywhere it comes from start up organisations who are at "level 1" on the Humphries scale, or from open source developers who would also register at "level 1". The larger and more tightly managed software organisations become, the worse their software. This is no accident - it's because the focus on process results in the actual product being forgotten. Level 1 works, level 5, works, levels 2,3,4 produce nothing of value.
There are three states.
Eating, Sleeping, Neither
So two bools are almost sensible. Since the Eating&&Sleeping state is illegal, it would be better to represent it with an enum anyway - the lack of enums is one of Java's less debateable bad design decisions. [Yes, I know there workarounds to this ommision but they suck].
As a Brit, I would normally side with Brits in a US v UK contest, but I am a nerd by choice, and a Brit by accident of birth, so this wasn't such a hard choice.
You don't need to be a genius to write utterly incomprehensible perl code.... Bah, this comment is too obvious. Look, I'll start again.
I would like to see a contest for programs that are readable and work even though they shouldn't, like if you create a c program which includes a bunch of macros
#define like {
#define ok }
#define um ;
#define int dude
etc
then you can write fairly serious programs in valley speak. I started doing this once, but then I got a life..
If something is moving away from us at nearly the speed of light (say .9c), then the light from that object still moves towards us at c, rather than .1c as you might expect.
Still, I agree that it is funny how people always assume that the observed limit of the known universe is the actual limit of the universe
This result doesn't change my understanding of the universe at all - I don't pretend to have one - unlike many cosmologists.
It's funny how little cosmologists confidence in already knowing everything (like age & size of universe) was shaken by the fact that their most fundamental prediction of the behaviour of the universe was completely wrong. (The fact that distance galaxies are accelerating away from us as opposed to decelarating)
You do get more than twice as much work done working 70 hours a week than 40.
He's also right about productivity. The best programmers do get 10x as much done. The exceptional programmers typically get 10x as much done as the good programmers.
Companies try to listen to this advice, but they never take it seriously. They think they can get really good programmers by paying 1.5x as much and buying a few toys. Unless they are very lucky, they just end up with average programmers like everyone else, and maybe a couple of engineer-savants (great programmers with no common sense).
The fact is that no companies will pay them 10x as much. In fact, it's hard to find a company that will pay 2x as much. Say an average developer can make $50,000 per year working 40 hours per week. If paid according to achievment, a really good programmer working 70 hours should earn around $15,000 PER WEEK. There's no way on earth that traditional companies will pay that much - the director of HR would go mental if they found out some hacker at the bottom of the management chain was earning twice as much as they were.
The only way for really good programmers to make the kind of money they are worth is by working for a start-up where they can have a decent equity slice. Then they'll work 70 hours + per week. Otherwise there's no point unless the work is *very* interesting. Contrary to popular opinion, the very best programmers do have a life. It's the 2nd tier geeks who happily work 70 hour weeks as a matter of course and spend the rest of their life playing video games.
If companies took the hyper-productive programmer thing seriously, the job market for programmers would look like that for sports stars. Until companies are prepared to pay those hyper-productive programmers enough that they could work for 6 months and then spend 6 months doing whatever the hell they like. In the mean time they just pretend to themselves that they have the best people.
> If the master wants to code in 'vi', who's to
> stop him? If the punk intern wants to do all his > work in Rational Rose, he's wasting time, and
> it's the project manager's responsibility to
> assign him to something small enough that
> he can learn good practice through practice.
That is pure poetry. It's also an accurate example, the master often does want to code in vi while the punk intern pisses about with Rational Rose, generating 1000 pissy ill thought out classes.
College is a good place to learn because you spend 4 years getting laid and drunk and puking up in flowerbeds, on the basis of which you will be better paid for the rest of your life. Sometimes you have to go and listen to pointless drival and you have less money than you would in a job, but it's more fun anyway.
Be wary of "formal development" - don't think there's anything magical about it, some people just prefer following mindless procedures to doing anything useful. These people manage to give off the impression that they're doing something important and "professional" but rarely produce anything. Developing large projects *is* difficult, it requires experience, intelligence and most of all an insight into the problem that takes a while to obtain, but doesn't come from any "methodology". There's no subsitute for understanding.
Become a good programmer first. Write your own lisp interpreter in assembly before you learn UML. Invent your own design patterns before you read about other people's.
The argument "how will these people support themselves", we must have laws to protect the income of these people from the free market.. is a socialist argument. That's what unions, price-protection schemes, minimum-wage etc are all about. We must have government laws and enforced government control of everything... It's hilarious - you think you're fighting the capitalist/socialist argument and it's *not* related, but if anything you're arguments are the wrong way round.
I'm a software developer, trust me, we'll manage just fine... We won't get as rich as Gates, but then again we weren't going to be anyway.
Your ideas about money and property are antiquated and not clearly thought out - I know them well, I would have agreed with you entirely two years ago. This may seem ridiculous to you but in 50 years or so, people will look back and laugh at the ridiculous efforts people made to ensure false scarcity of "their" information - a classic prisoners dilemma. If everyone on the planet had access to *all* information on the planet everyone would be better off.
Then there is the "who would create it" argument, which is also bogus. People (in "advanced" world) don't work for material wealth - they work for status. Does a lawyer really work 70 hours a week because the latest model mercedes is so much more comfortable - of course not. The only reason he cares that he's driving the latest model mercedes is the status it affords him in his brain-damaged world. In an IP free world, status will be obtained by creating useful stuff and having it used by as many people as possible.
The difference between a "bit" and an "it" is disappearing. I'm a little ahead of the game here - I used to work at 3Dsystems where I could download a vrml model off the net and print it out. As technology advances, nanotech etc, almost the only type of wealth that matters will be IP. Our model of an optimal economy is stuck in a local minima at the moment. Capitalism is a good system for optimal distribution of scare resources but people are too stupid to notice when the model is no longer valid. The internet provides enough of a jolt to push us out. Various misguided selfish interests will try to keep humanity stuck, but I don't think they'll win.
Um, your boss sounds very stupid. I have to wonder who would be dumb enough to work for someone like that :)
Nonsense. People work longer hours due to stupidity not technology. They need to work longer because they're not doing anything useful anymore - they're sitting in meetings discussing mission statements or ISO9000 compliance tests or a million other worthless activities. Dilbert is entirely accurate. The more extreme useless people use technology to generate anti-work more efficiently than ever before, but the root cause is addiction to ritual, not technology (www.reciprocality.org). In fact it's only technology that prevents the whole edifice from disintegrating and snapping people out of their stupors through economic collapse and eventual starvation. Left to their own devices the majority of humnity will slump into an eternal ritual where every moment of their lives is entirely predictable from the greetings that their co-workers give as they walk into the office to the format, outcomes and even dialog of their favorite shows. As it is, the economic surplus provided by increasing mechanisation nicely matches the increasing stupidity and worthlessness of humanity. We're already living in a post-scarcity world (in the west at least), the wheels would keep turning just fine if only 10% of the population worked. People don't want more leisure time though, they want to keep their minds in a state of minimum utilization, so everyone spends more time at work performing pointless rituals and spends the rest of their lives watching predicatable television. Anything unpredictable must be destoyed - such as children who don't follow the rules - that's OK though, there are drugs to cure that (ritalin).
And here endeth the rant, for today anyway.
The most popular language (in terms of number of programmers) is VB
The most popular language (in terms of active lines of code still in use) is cobol
The most popular language (in terms of useful lines of code still in use) is C
The most popular language (in terms of useful lines of code being written) is probably C++
The most popular language (in terms of brand new projects) is Java
I did AI at Edinburgh, and had Mitchie as a lecturer as few times. I would have actually turned up and listened more frequently if I had known his past - as it is I spent 2 years down the pub, followed by 2 years programming in the bowels of the machine rooms - they actually gave undergraduates keys to the machine rooms full of sun workstations so you could work all night if you preferred (which all the serious guys did - you get more resources at 3.00am).
I write software for these systems - (former employee, now consultant). It was fun to run a test overnight, come to work in the morning and have 14 little resin models of the starship enterprise waiting for you. The algorithms are fun - similar requirements to computer graphics, but interestingly different (I got about 5 patents for 3DSystems, back when I was young and naive and didn't understand that software patents are evil...). Good times, nice high end SGI boxes to play with (my Linux box has only just caught up with the system I was using 7 years ago - coppermine,GeForce, SuSE 7.0, and open source OpenInventor feels very similar to R10,000 with high-impact graphics card, but that was $50,000 worth not $2000) - fun job, but California is too damn sunny for my tastes, and Dilbert was too accurate to be funny back then.
You could use it for everything from building someone's skull from an MRI scan - so the brain surgeon would know whay to do, to building models of the internals of Cruise missile warheads (single material, no moving parts though - a bit like the T1000 (terminator 2) but much slower), to Hawaii Barbie...
There are several types of machine, accuracy tends to be about 5/1000 of an inch (injket based models do about 300dpi, but downfacing surfaces are shitty) takes about 10 hours for average part.
It would be cool to use similar process on a larger scale for building office blocks.
It was fun to find VRML models on the net, run a little conversion program and print them out. Another neat trick is to get yourself scanned and make a personal voodoo doll, but the software to fix the model is kinda tricky.
You may get away with 20% when you're brilliant at it. For now expect to be 90% manager, ie forget about doing any useful coding for the moment. The best thing about this move is that if it doesn't work out - it'll make you a better engineer - you'll have more sympathy for good managers, and more power against bad ones.
e x.html
Read "the 5 minute manager", "7 habits..", "mythical man month", "principles of software engineering management" (Tom Gilb), "the prince", "the art of war", "time management for dummies" and neither last nor least: http://www.reciprocality.org/Reciprocality/r0/ind
As someone mentioned earlier - the hardest thing about this is that it requires an orthogonal mindset. People who are good at focusing intently on a single task for months and getting deeply stuck into a problem ("ADD" people, strangely enough - the best programmers), are *absolutely crap* at keeping track of 100s of trivial tasks. The necessary skills can be learnt by a tech-head, but take it seriously. It's at least as difficult for a tech-head as say - getting a physics degree.
Well, we'll see where the two systems are three years from now. Solaris *is* a good OS now, it's just none too quick (TCP/IP stack aside). I'm still bitter about ancient history. Solaris was bloody horrible until around 2.6, by which time I had abandoned Sun and moved to SGI. Sun shat on any customers who didn't want to move to slow, buggy Solaris as opposed to nice old BSD based SunOS. And if you had been trying Java on and off since before 1.0 was released, you would be pretty sick of it by now too. 10 years ago, Sun was leap years ahead of what you could find elsewhere - dbxtool came for free with the system - they were just fucking great machines. Now, they're great for an enterprise server, but having a personal sun workstation just isn't the intellectual status symbol it used to be. If they hadn't let their marketing department take over their software direction, life would be better and I would still be using Suns.
There's no fury like an engineer scorned.
Yeah exactly. Sun make good hardware, but their software has been going backwards ever since they switched from SunOS 4.1.2 to Solaris, and made the c compiler an (expensive) optional extra. What difference does their endorsement of Gnome make to anything.
If I were a Gnome supporter I would be more worried than jubliant about getting sun's endorsement - look at the wondrous advancement in CDE and motif since sun got on board. And don't even get me started about java...
KDE2 on the other hand actually provides a browser that seriously competes with IE5.5. I wouldn't hold my breath looking for that in the GTK/mozilla world. I'm no great fan of MS, but they were right about one thing: netscape engineers *are* weenies. It'll be a long time before mozilla overcomes that background, and in the mean time, there will be no decent Gnome browser.
I really like your idea of AST. Your comments seem very accurate and relevent. Could you email so we can continue this offline.
Joss
I think you've missed my point slightly.
I agree that GUIs are good for computer->human
interaction. Computers are good at generating images.
> However, language was designed to work with certain hardware limitations: humans can't display imagery as complex as a computer in real time.
That's my exact point. What I'm saying is that the most powerful effective mechanism for humans to tell computers what to do is through language.
"Do I have to draw you a picture"
Saying that in order to use a computer really effectively, you need to be a programmer (to a certain extent) - it's not like saying you should be able to fix a car if you're going to own one - it's more like saying you should learn to drive.
Gui's generally present the user with a small set of choices to make at each point. Your example of choosing a color is a perfect example of what GUIs are good for. The power of language is that the range of choices grows exponentially with the length of the expression.
In the starkest terms: excessive use of GUIs reduce your Turing machine into a finite state machine. The only reason this is considered remotely acceptable is that the bulk of humanity thrives on pointless repetitive ritual. They would rather spend their days repeating almost exactly the same thing than have to stop and think occasionally. The emphasis on GUIs allows people to stay comfortably in a semi-conscious trance relying on their mechanistic mental facilities to get them through the day - using language effectively requires a capacity for inductive thought which many people prefer to leave dormant.
The job of programmers is to bring the power of the computer within reach of the user. Not to tame it by pretending it's a glorified bloody toaster, or to waste their time by demanding they know a bunch of esoteric details (like printcap files) which were introduced for the benefit of the programmer. The best example I know of where an application got it right is autocad. Where I was using autocad the users were all draughtsmen. The best user/programmer there had left school at 15 - he learnt what he needed to know on the fly. You're quite wrong thinking that autocad worked because it's users were engineers. It worked because humans have brains, and with a little encouragement they'll use them.
I agree that GUIs are much more intuitive for some things. I think most computer->human interaction should be GUI based. But I also think the focus on them as the primary mechanism for Human-Computer interaction has held us back tremendously. When we stop pretending computers are something else, things will improve.
This might sound ridiculous now, but it will be blatantly obvious in 20 years. (I'm an arrogant son of a bitch, to be sure).
This daft window metaphore crap is largely responsible for reducing the most powerful invention of the previous century to just another time-wasting useless idiot-box where humanity's natural sloth and aversion for thought can reign supreme.
"A picture is worth a thousand words" - it's a myth. Draw me a picture of "love". The trouble with GUI interfaces is that they are predisposed towards the computer transfering information to the human. They are not an efficient mechanism for humans to transfer information to the computer. There is a good mechanism for this - one that has been used for millenia and which are brains have even evolved to use effectively. And that mechanism is language. This is why text based interfaces will always rule over this GUI shit. The "integrated" desktop is doomed to forever strive for the level of power, speed, simplicity and componentisation already provided by the shell tools.
Why use a metaphor for a work-style whose time has past (WIMP - "desktop") when the reality of the computer is orders of magnitude more powerful and more flexible. Even as programmers deny this, and preach the religion of GUIs they implicitly acknowledge that to do anything powerful, you need to use a language - a programming language.
"Oooh, but that's too complicated for my users..."
Am I saying that every computer user should be a programmer ? Yes I am, but only to the extent that it makes sense in the domain. For instance, any real power-user of autocad will write small lISP programs. How could a commerical company make a fully functional, mathematically pure programming language as the basis of an interface for a program intended for draftsmen ? Because they understood that stupidity is like work - it expands to fill the space allocated to it. Expect intelligence and you will receive it. Autocad dominated their market like this. This approach would work in other markets if programmers had the humility to admit that their "users" were as smart as they were, and deserved the same power that programmers reserve for themselves.
If I want to do GUI programming (which *is* good for computer->human interaction), I'll use fltk, a far more powerful and better designed toolkit than either Gnome or KDE. Plus it's cross platform, and LGPL. If I want to knit things together, I'll use a bloody pipe.
I have been wanting this for so long. I met the OpenInventor team about 5 years ago - smart cookies. I've been begging SGI for exactly this every year for the past 4 years.
They did an incredible job - it's one of the best designed C++ libraries I've ever used. It's not suitable for games but for putting good 3d interactive graphics in a professional program without spending *years* learning the details of OpenGL it is wonderful.
This is the one move that SGI could have made to prevent Direct3D winning the 3D api wars. Scenegraph based applications are an order of magnitude easier to program.
Thanks SGI - I might even buy another SGI now (I know SGIs came with OpenInventor - but it wouldn't work gcc before)
gtk is the obvious choice for anyone coming from a motif background. Nothing else is sufficiently awkward to program with - it just wouldn't feel natural.
I mean code like this:
attributes.y = widget->allocation.y;
attributes.width = widget->allocation.width;
attributes.height = widget->allocation.height;
attributes.wclass = GDK_INPUT_OUTPUT;
attributes.window_type = GDK_WINDOW_CHILD;
attributes.event_mask = gtk_widget_get_events (widget) |
GDK_EXPOSURE_MASK | GDK_BUTTON_PRESS_MASK |
GDK_BUTTON_RELEASE_MASK | GDK_POINTER_MOTION_MASK |
GDK_POINTER_MOTION_HINT_MASK;
attributes.visual = gtk_widget_get_visual (widget);
attributes.colormap = gtk_widget_get_colormap (widget);
attributes_mask = GDK_WA_X | GDK_WA_Y | GDK_WA_VISUAL | GDK_WA_COLORMAP;
widget->window = gdk_window_new (widget->parent->window, &attributes, attributes_mask);
is almost ugly and verbose enough to make a motif or X programmer feel comfortable. Plus GTK has got it's own home-grown C based OO model, just like X, so you can get all the complication and grief of OO techniques without having to deal with any of the syntactic sugar like operator overloading or templates which can (but normally doesn't) make C++ readable. Instead you can have a C++ wrapper that never quite works right stuffed on top of a subtly incompatable object model. It really is the perfect choice for motif converts.
Disclaimer: it might have improved, it is 6 months since I used it.
Reliability was dreadful, lots of functions don't behave as logic dictates they should, several things plain broken. It takes ages to find these problems and work around them. Your prototypes look good, but then several months down the line you find peculiar bugs popping up - for instance pop-up menus that occasionally just decide to stay visible for ever.
Complexity - huge amount to learn
String label = (String)JOptionPane.showInputDialog(frame,
"Input label for ",
"Info",
JOptionPane.QUESTION_MESSAGE,
null,
null,
field.substring(i));
what are all those args for - I can't remember just now, but you need to know it.
Verbosity - it took me 5000 lines of fltk to replace 25k of swing.
Performance - this was the worst. Memory footprint was horrendous, CPU useage was bad. Stick a breakpoint in and examine the call stack sometime - it'll be at least 18 levels deep. This is a sign of shitty design, no matter what anyone tells you.
But it's still better than MFC.
If you're writing GUI apps in C++, fltk is the way to go. I've used X, motif, MFC, GTK, Qt, Java Swing, and god knows what else and the only pleasureable experience was with FLTK. It's very fast, very light, very simple, just lets you get on with it.
Qt was pretty good - close second, but it's not LGPL and you need to pay for the windows version.
Fltk is just cleaner. MFC was the worst, closely followed by swing.
It's not that simple. You get most
of that energy back. The stuff closing
in behind you can push you forwards,
(and it's not that simple either).
You could use a similar calculation and
prove that tuna couldn't swim that
fast without an onboard nuclear reactor.
But they do, and you don't need a geiger
counter in a sushi restaurant.