Linux on the desktop is still nowhere, so you can't seriously be suggesting that is important to Sun.
Sun cares enough about Linux on the desktop to push the Java Desktop System. (Of course, some people think that this is a push to make Java a new de facto application programming language....)
Laying aside the question of why I'd be writing code that's not complete (what does that even mean?), I'm very much not interested in writing code that I can't prove actually meets my customer's needs.
With that in mind, I cannot see how proving that my code works by writing and running automated tests is a waste. Even further, if I do need to change the code later, because I have automated tests, I can anchor both ends of any refactoring with those tests.
Now it's reasonable to talk about writing good tests and not wasting time on bad tests, but that's a different argument altogether.
I'd rather tell the customer what such stories would cost and let him decide if they're really necessary. If your project requires such sneakiness, you may have larger problems than stupid customer requests.
I think it's rather that overdesigning and overgeneralizing (or in XP terms, writing code you don't need) is a waste of developer time and customer money better spent solving the customer's most important problem. Increasing programmer joy of programming is a nice side benefit.
Perhaps a better approach is to talk to the customer (whoever that is; usually a representative of who needs the software or who's paying for the software) regularly, balancing what the customer really wants with when and how the developers can produce it.
Provided you or trusted community members have the ability to keep things in line and exercise that ability frequently, you likely won't need to dole out mass moderation powers for a long time. It's possible to stay on top of things if you only have a few dozen posts every day.
I'd prefer to say that a static, weakly typed system requires you to do typecasting. That's my gripe with C's type system (as well as that of C++ and Java).
If it worked well, I'd be fine with it. I don't find it to work well, thus I don't care for static typing as done in these languages. (I don't care to do extra work to tell the compiler what it's capable of figuring out on it its own.)
Now a nice type inferencing system, that'd be worth using.
Dictatorships tend to be more efficient than democracies (especially representative democracies), but democracies tend to be more pleasant for everyone outside of the ruling class.
It's much, much easier to come up with an idea than it is to see that idea to completion. A finished product is more valuable than an idea. Consequently, anything that helps you finish the project is more valuable -- that includes coders and artists.
It's nice that people have ideas. I'm sure some of them are really good. They're just not as useful or valuable as people who are willing and able to turn an idea into a finished product.
We're not going in numerical order anymore; we're going in order of importance. Apocalypse 12 should come out in the next few weeks. That's everything objects.
It's NCI, the Native Call Interface, and it's immensely easier than Perl XS. (Admittedly, XS isn't bad for simple things, but it's nasty for tricky things.) You do have to know Parrot assembly fairly well to get things done, but parameter handling is all done for you.
I've been working on SDL bindings for Parrot. They're reasonably straightforward and easy -- and you don't have to have a compiler to make them work! See examples/sdl/* for examples.
Either Dan or I will have an article or two on ONLamp.com about NCI in the near future. If you're really curious, send me an e-mail and I'll tell you more.
Hmm, that's really odd. I just grabbed 0.1.0 from the CPAN and it has a README and a lib/Parrot/BuildUtil.pm. The configurator also works just fine here.
They're wasting my bandwidth, my CPU cycles, my disk space, my time, and, occasionally, they're forging my domain on spam sent to other people. How is that not hurting me?
Slashdot Mirror
Not always true; several dynamic languages allow you to declare or to change inheritance at runtime.
Perhaps you meant that inheritance is a property of the class, in a class-based language.
Parrot already runs on Windows, with nightly smoke tests.
Sun cares enough about Linux on the desktop to push the Java Desktop System. (Of course, some people think that this is a push to make Java a new de facto application programming language....)
Laying aside the question of why I'd be writing code that's not complete (what does that even mean?), I'm very much not interested in writing code that I can't prove actually meets my customer's needs.
With that in mind, I cannot see how proving that my code works by writing and running automated tests is a waste. Even further, if I do need to change the code later, because I have automated tests, I can anchor both ends of any refactoring with those tests.
Now it's reasonable to talk about writing good tests and not wasting time on bad tests, but that's a different argument altogether.
I'd rather tell the customer what such stories would cost and let him decide if they're really necessary. If your project requires such sneakiness, you may have larger problems than stupid customer requests.
I think it's rather that overdesigning and overgeneralizing (or in XP terms, writing code you don't need) is a waste of developer time and customer money better spent solving the customer's most important problem. Increasing programmer joy of programming is a nice side benefit.
Perhaps a better approach is to talk to the customer (whoever that is; usually a representative of who needs the software or who's paying for the software) regularly, balancing what the customer really wants with when and how the developers can produce it.
(Disclaimer: I wrote a book about that.)
Provided you or trusted community members have the ability to keep things in line and exercise that ability frequently, you likely won't need to dole out mass moderation powers for a long time. It's possible to stay on top of things if you only have a few dozen posts every day.
A lot of those authors are easy to track down. The problem is in getting 70+ year old corpses to sign legal forms.
Any law that requires raising the dead for the public good is bad law.
I'm not sure that relying on spammers to follow voluntary rules is workable. (Actually, I'm sure that it's not workable.)
Last I knew, it was every Wednesday night. Do you have more recent information?
I'd prefer to say that a static, weakly typed system requires you to do typecasting. That's my gripe with C's type system (as well as that of C++ and Java).
If it worked well, I'd be fine with it. I don't find it to work well, thus I don't care for static typing as done in these languages. (I don't care to do extra work to tell the compiler what it's capable of figuring out on it its own.)
Now a nice type inferencing system, that'd be worth using.
Yes, that's true. I meant "efficient" in the short term. Transitions between dictators tend to be messy.
Dictatorships tend to be more efficient than democracies (especially representative democracies), but democracies tend to be more pleasant for everyone outside of the ruling class.
I'm not sure that's entirely correct. Have you seen the Smalltalk browser? Somehow, they manage to do those things well enough in a dynamic language.
Here's SDL. It's an interesting idea. Go do it!
Perhaps we should tattoo your first paragraph on everyone who writes such an article.
It might be less painful to write "You have permission to implement your good ideas", but I do prefer your phrasing.
It's much, much easier to come up with an idea than it is to see that idea to completion. A finished product is more valuable than an idea. Consequently, anything that helps you finish the project is more valuable -- that includes coders and artists.
It's nice that people have ideas. I'm sure some of them are really good. They're just not as useful or valuable as people who are willing and able to turn an idea into a finished product.
We're not going in numerical order anymore; we're going in order of importance. Apocalypse 12 should come out in the next few weeks. That's everything objects.
It's NCI, the Native Call Interface, and it's immensely easier than Perl XS. (Admittedly, XS isn't bad for simple things, but it's nasty for tricky things.) You do have to know Parrot assembly fairly well to get things done, but parameter handling is all done for you.
I've been working on SDL bindings for Parrot. They're reasonably straightforward and easy -- and you don't have to have a compiler to make them work! See examples/sdl/* for examples.
Either Dan or I will have an article or two on ONLamp.com about NCI in the near future. If you're really curious, send me an e-mail and I'll tell you more.
Hmm, that's really odd. I just grabbed 0.1.0 from the CPAN and it has a README and a lib/Parrot/BuildUtil.pm. The configurator also works just fine here.
Rob had probably already accepted and scheduled this story for today.
You should have said straight out that you don't believe that I have the right to decide who can use my property. It would have saved time.
They're wasting my bandwidth, my CPU cycles, my disk space, my time, and, occasionally, they're forging my domain on spam sent to other people. How is that not hurting me?
That's an incomplete analysis; it ignores the quality of a product. In my mind, that's more important than popularity.