OSDev.org

I like a simple imperative PL like C for system-level programming, and something like perl for much high-level stuff.  I HATE operator overloading, operators should do simple, well-defined things that you can understand just by looking at them, and if you need something more complex you make a function call.  I like mostly-weak typing, although I do like to have checks that prevent me from using a pointer as an integer or vice versa without a cast.

My wishlist for C-like languages is as follows:

*  I want explicit endian attributes on integer types
*  I want all the built-in integer types to be explicitly sized, preferrably so that I can say how many bits I want in an integer, and all other types derived from them using typedefs.
*  I wish libc functionality was split up into smaller components, such that string handling is separate from system interfaces, etc.

I think that most important thing about a language is standardization. Many troubles in the past would never happen if not everybody was trying to invent something new. I think that most used languages are c/c++ (+asm) and all should steak to them. Making always something new and better is just adding more confusion to the thing.

You all here write English, right? But how many of us are realy native speakers of it? (btw sorry for my english:-)

This is my opinion

I think that we need a variety of languages. Different languages are good for solving different kinds of problems. You can solve a problem (that is elegantly solved with a simple PROLOG program) with a bigger amount of complex C++ code. If everyone were using the same language that would lead into more complexity because people could no longer solve their software problems in an elegant way.

IMHO the reason why you are demanding for standardization is the inherent complexity of software engineering. The main problem in software engineering is managing complexity. I think this complexity can only be reduced by using many different programming languages and more dynamic languages.

Ask yourself why people need to invent something new or to reinvent something old? Because present solutions are not reusable enough. This is not the result of using many different languages. It is the result of bad software design and complex languages together with existing solutions which are too complex (not reusable).

- Esa

What Ezanahka said.

If you're writing your own OS as a hobby (especially if you're the only coder), you'll want to keep things simple. You might want to stick with C, **OR** you might do as I'm doing and work on a language that's even simpler to parse and compile. I'm working on the prototype for a compiler (http://www.karig.net/0010.html) that understands only ONE WORD (a word used to poke machine-code bytes into a code buffer) and four commands (define new word, compile CALL to word, execute word now, and push number onto parameter stack).

I'm doing this so that I can start coding in my OS before I even really HAVE an OS. Also, I'd like my OS to have a kernel like that in Chuck Moore's colorForth (http://www.colorforth.com/cf.html).

Sure, we can shoot ideas back and forth.

[email protected]

Hi again

You are talking about complexity, but I think adding on that some more languages adds some more complexity. And not always the reason for inventing some new language is 'not reusability' of the old one. Maybe it is a luck of knowladge about the language it self.

Bad software design could also come from not knowing the language right. Concetrating on one language and learning it well in my opinion is better then knowing many not so well. The question could be what is that language. I think that the time has choosen c/c++, not me.

Goran

True... adding more languages to the mix can add complexity especially if you have to combine source code written in many languages into one object binary.

But then again not adding those languages adds different kind of complexity. Having a variety of languages creates the problem of knowing and choosing the right one for the task. However if the right one is chosen the implementation of the program can be much simpler.

True... not knowing how to use your tools does usually lead into bad results. I agree that it is better to know one language well than to know many languages superficially.

However much more important still is having the necessary know-how which does not depend upon any one language. Good knowledge of data models, structures, algorithms and machine architecture is necessary. A person who knows these basics can learn and use various languages much more efficiently.

True... C/C++ is basically an industry standard. However the problem with C/C++ is that they are complex in themselves. C/C++ uses a mixture of prefix, postfix, infix, etc. notations. It is a syntactical mess. I dislike the idea that when writing software I have to concentrate on how to use the tool right instead of concentrating on the problem I'm trying to solve. C/C++ is syntactically complex and therefore hard to learn to use correctly. Why choose such a language? Simply because others do so as well?

- Esa