The D Programming Language, Version 1.0

penguinblotter writes in a journal article: "Soon, Walter Bright is scheduled to release version 1.0 of the D Programming Language. D is a systems programming language. Its focus is on combining the power and high performance of C and C++ with the programmer productivity of modern languages like Ruby and Python. Special attention is given to the needs of quality assurance, documentation, management, portability and reliability. D has appeared on Slashdot a few times before, and Walter has continued to add more and more features. Most Slashdot community comments in these articles have been offered on feature X or spec Y without reading through the extensive D newsgroup archives. It has been here over the past seven years where extremely gifted and experienced programmers hashed out discussions and arrived at excellent implementations of all the ideas discussed." Read on for the rest of penguinblotter's writeup.

For those with a C/C++ background, D offers:

• native code speed
• extremely fast compilation times
• garbage collection (although you can manage your own memory if you want)
• OOP - by reference only, easy initialization, always virtual
• cleaner template metaprogramming syntax, more powerful templates, as well
• built-in dynamic and associative arrays, array slicing
• versioning (no preprocessor madness)
• link-compatibility with C
• nested functions
• class delegates / function pointers
• module system

For those with a C#/Java background (a shorter list, but one with big wins):

• similar syntax
• No virtual machine or interpreter
• built-in unit testing and design-by-contract

These two comparison sheets can go into more depth on how D stacks up against other languages.

From D's creator:
For me, it's hard to pinpoint any particular feature or two. It's the combination of features that makes the cake, not the sugar, flour or baking powder. So,

1. My programs come together faster and have fewer bugs.
2. Once written, the programs are easier to modify.
3. I can do (1) and (2) without giving up performance.

Get your compilers and start hacking D!

• DMD (Digital Mars reference compiler, Windows & Linux, x86)
• GDC (GCC front-end)

This won't work...

[PurifyYourMind (776223)]

...it's just another version race. D may have won for now, but someone out there is already working on the E programming language. ;-)

Re:This won't work...

[0racle (667029)]

Now, after F do we get G or 10?

Re:This won't work...

[TERdON (862570)]

Now, after F do we get G or 10?

Either it'll be called 10, or H. G, has already been taken, not only once [wikipedia.org], but twice [wikipedia.org].

For your reference (kudos goes to Wikipedia [wikipedia.org]), the following single letter (sometimes including some additional nonalphabetic characters) have also been implemented:

A+ [wikipedia.org] A++ [wikipedia.org] B [wikipedia.org] C [wikipedia.org] C-- [wikipedia.org] C++ [wikipedia.org] C# [wikipedia.org] D [wikipedia.org] E [wikipedia.org] F [wikipedia.org] F# [wikipedia.org] G (now known as Deesel) [wikipedia.org] G [wikipedia.org] J [wikipedia.org] J# [wikipedia.org] J++ [wikipedia.org] K [wikipedia.org] L [wikipedia.org] M4 [wikipedia.org] Q [wikipedia.org] R [wikipedia.org] S [wikipedia.org] S2 [wikipedia.org] T [wikipedia.org] X10 [wikipedia.org]

So - that only leaves you the letters H, I, N, O, P (sic!), U, V, W, Y and Z if you don't want to have a name clash with another programming language. Technically, M and X are followed by numbers in the previous examples, so you could argue for them as well, and even A (as it has a plus behind the letter)

I'm mostly surprised that noone has thought of a (P)rogramming language. :)

Re:This won't work...

[sacrilicious (316896)]

So - that only leaves you the letters H, I, N, O, P (sic!), U, V, W, Y and Z

I hope someone creates an "O" language. That way, interface bindings for it could be called "O-interfaces", or "O-faces" for short. So when you create a nifty library for the language, publishing it could be known as "showing the world your O-face".

Re:This won't work...

[jlarocco (851450)]

Why do we need yet another programming language?

Exactly. There's already Fortran and COBOL, everything else is superfluous.

Seriously though, why don't we need another programming language? It's not like we only get a finite number of them. We're not going to run out of space or anything.

If it doesn't interest you, don't use it.

Re:

[0racle (667029)]

Did you say the same thing when Python or Ruby were created? People do things in different ways, different languages end up with their strengths and weaknesses that way. So until there is a perfect language that is all things to everyone, yes there is always a need for another language.

Re:This won't work...

[TheRaven64 (641858)]

UTF-32 does, indeed, do that. It is quite a good way of working internally. You can do things with UTF-32 much more efficiently than with UTF-16 or UTF-8 (e.g. searching for a character just requires you to compare each 32-bit value to the target, without having to check it isn't a special character that is the first in an escape sequence). Most modern processors come with a vector unit that can handle vectors of 32-bit integers, so if you have to handle large quantities of text you can speed certain things up even more by running streaming calculation on the vector unit.

Re:

[LunarCrisis (966179)]

Though I agree with you for the most part (finding the nth character in a UTF-8 string is unnecessarily long), this is wrong:

searching for a character just requires you to compare each 32-bit value to the target, without having to check it isn't a special character that is the first in an escape sequence

UTF-8 was designed so that no complete set of bytes representing a character occurs as a substring of any other. This makes the search problem into a simple search for a string inside another string. The searching routine doesn't even need to know whether or not it's a UTF-8 string or not, just as long as it doesn't mangle the last bit.

Re:Because the ones we have suck?

[TheRaven64 (641858)]

There is no such thing as an interpreted language. There are languages, and there are interpreters and compilers. Lisp, for example, can be both interpreted and compiled (Scheme is usually interpreted and Common Lisp is usually compiled, but there are exceptions). Tcc can both compile and interpret C code. Java can be compiled by something like gcj or interpreted by a JVM. If you compile Java you lose some of the features; the JVM bytecode format is designed so that it is easy for automated tools to reason about. At load-time, the JVM will parse the class files and check that they do not violate the Java security model; this is theoretically possible with compiled code, but much harder.

The Squeak runtime for Smalltalk is written in Smalltalk. There is a smallish subset of Smalltalk used to write the basic functionality, which is compiled to native code. This then supports the whole language. The same model is, I believe, used for JNode, an operating system written in Java...

Re:

[jyoull (512280)]

There is no such thing as an interpreted language. There are languages, and there are interpreters and compilers.

Are you really well acquainted with gcj? I'm sorry, but I don't get how the end result or even the stuff going into it (and the required inputs, like making some explicit calls that would never be required in Java) can be called Java anymore.

the point I didn't make well was that when a language has been designed to execute inside a containing environment (the JRE or whatever facsimile thereof) yo

Re:Because the ones we have suck?

[thephotoman (791574)]

Wait...C++ without the bullshit? That takes away all of C++!

D is surprisingly good.

[FishWithAHammer (957772)]

I'm looking at using it via GDC for my next project. For people who use C/C++ regularly, this is something you ought to look into.

It's not a toy language. If you're a C++ programmer, you'll be almost immediately functional in the language. And you can call C and C++ libraries seamlessly. It's pretty sweet.

Re:D is surprisingly good.

[matrixise (643691)]

from http://digitalmars.com/d/interfaceToC.html [digitalmars.com]

D does not provide an interface to C++. Since D, however, interfaces directly to C, it can interface directly to C++ code if it is declared as having C linkage.

D class objects are incompatible with C++ class objects.

Re:D is surprisingly good.

[eluusive (642298)]

Please see: http://dsource.org/projects/bcd [dsource.org]

Re:

[Brandybuck (704397)]

And you can call C and C++ libraries seamlessly.

Really? Not according to their FAQ. C yes. C++ no. Otherwise I would be in the process of switching over as we speak.

Re:

[Geoffreyerffoeg (729040)]

If you don't know pointers, then you don't know how the machine works. I would never use a doctor that didn't know how my body worked.

Yes, but last time I went to the doctor he didn't grab a hand pump and start pushing my blood around.

And when was the last time C let you manually push items onto the stack?

Weird writeup:

[Per Abrahamsen (1397)]

From the compared to C/C++ list:

        * native code speed

As opposed to C/C++.

        * extremely fast compilation times

Point granted (compared with most C++ compilers).

        * garbage collection (although you can manage your own memory if you want)

Point granted, even though C and C++ arguably have optional garbage collection as well (if you link to the right library).

        * OOP - by reference only, easy initialization, always virtual

Only value semantic? Meyer had to accept that value semantic was useful, and add it to Eiffel eventually, and C# added it over Java.

And no way to specify that a function will always be the one specific. Good luck doing any kind of reasoning there.

Bragging about missing features, that are essential to many tasks.

        * cleaner template metaprogramming syntax, more powerful templates, as well

*More* powerful templates? The usual complaint is that C++ templates are too powerful (a Turing-equivalent compile time language).

        * built-in dynamic and associative arrays, array slicing

Not exactly a recommendation that the core language apparently is so weak that these can't be put into libraries.

        * versioning (no preprocessor madness)

I'm guessing he meant variants here, the preprocessor is often used for variants, rarely for versioning.

        * link-compatibility with C

Which C and C++ of course lacks?

        * nested functions

Point.

        * class delegates / function pointers

Obviously both C and C++ have function pointers.

        * module system

More preprecessor replacement here.

The C#/Java list:

        * similar syntax

But totally different from C++?

        * No virtual machine or interpreter

You can compile Java and C# to native code as well, so this is just another case of bragging about a missing feature.

        * built-in unit testing and design-by-contract

I'm a C++ programmer, and this is by far the most interesting aspect of the D language (and of Eiffel before that). Don't know why it should be in the Java/C# list.

Re:

[91degrees (207121)]

Couple of points:

* native code speed

I think this is a response to criticisms from C programmers about most modern languages, rather than a benefit over C.

Not exactly a recommendation that the core language apparently is so weak that these can't be put into libraries.

Some of this is useful enough to be built in. stl and the like are pretty handy but sometimes it feels a bit of a kludge. Plus, built-in allows better optimistations for specific cases.

Obviously both C and C++ have function pointe

Re:

[donglekey (124433)]

This strikes me as the thoughts of someone who hasn't given D a good look. I have been using it recently and it is phenomenal. It is a breath of fresh air. It is not that it has many of these features over C/C++, it is that it cleans out the enormous amounts of headache inducing things about these languages while retaining what Java/C# lose. Native code speeds. No VM. Trivial Integration with C.

java native code compilation

[Nutty_Irishman (729030)]

All other considerations aside, runtime speed really should be a justification as a bonus to Java. Java isn't that much slower if you actually take the time to compile it to native code first. Using something like a JIT compiler http://en.wikipedia.org/wiki/Just-in-time_compilat ion [wikipedia.org] can greatly increase the speed of your code and put it close in line with C++. I would certainly consider D if both 1 and 2 were better than Java.

Re:java native code compilation

[Decaff (42676)]

Java isn't that much slower if you actually take the time to compile it to native code first. Using something like a JIT compiler http://en.wikipedia.org/wiki/Just-in-time_compilat [wikipedia.org] ion can greatly increase the speed of your code and put it close in line with C++.

This is a bit of an old myth. Almost all Java is run as native code these days, even on VMs, and is mostly pretty close to C++ speed. Benchmarks that show Java as significantly slower than C++ usually result from not allowing the VM enough time to perform native code translation of time-critical code. Java has moved away from JIT compilation (as against the later optimisation of HotSpot) because it led to long start-up times - you had to wait for code to be compiled to native before it ran. Now Java usually starts up as interpreted, with the translation to native code happening later on, in the background.

Where C, C++ and D win out over Java in terms of performance is when you need programs that have to start up fast, run fast, but only for short periods (a few seconds).

Currently learning D

[kihjin (866070)]

Note: I've been programming in C/C++ for four years.

I took it upon myself to learn D not more than a few weeks ago. A classmate introduced me to the language last spring.

While I'm still learning D, it has some notable features:

• auto keyword for inferred type declaration
• lazy keyword for evaluation
• delegates are like function pointers, but cooler. Literal statements can be passed as variables, and aren't evaluated until the delegate is called.
• scope(exit|failure|success), specify a block of exit code
• in/out/inout function keywords, offer readable code for determining what a parameter in a function is designed for.
• get/set methods automatically become a property (accessed like a public variable)
• foreach, foreach_reverse, container iteration
• with statement, C++'s using on a object-level

Of course one may argue that none of this is necessary and could be made independent of the language itself. My belief is that would increase the complexity of coding in D.

If you're interested in D you should visit http://www.dsource.org/ [dsource.org]. There are some interesting projects such as Derelict [dsource.org] (collection of C game bindings) and Bud [dsource.org] (make and SCons replacement).

It looks like a step down

[Anonymous Coward]

After working so hard to get from C to C++, I don't see why I would settle for D. My next programming language is going to be B- or better.

Re:

[4D6963 (933028)]

After working so hard to get from C to C++, I don't see why I would settle for D

I'd gladly have my girlfriend to go from C to D, but I admit there's nothing I would love more than DD.

Edit : yeah I know, quit telling me that I must be a liar, I don't have a girlfriend, WHO CARES?!?
Edit #2 : What? You can't edit your comments on Slashdot? So how am I doing that, jerk?

Do we really need another D infomercial?

[Anonymous Brave Guy (457657)]

I'm sure D is a lovely language, but it just seems like another incremental change over C++, like Java and C# before it, and like both of those languages what it's lost and the opportunities it misses are as telling as the little tweaks it makes to improve things.

No-one has yet been successful, IMHO, in developing a really good industrial programming language. Those that make it tend to be pragmatic, practical tools like C and Perl and FORTRAN and COBOL. To be sure, each of these has many widely-acknowledged weaknesses, but the overall balance between those weaknesses and what you could get done using the language was right.

I can increasingly see why some well-known programming language designers shy away from feature comparison ticklists. I think it's because as soon as you go down this route, you bias the comparison so much that it's meaningless. For example, consider the first checklist [prowiki.org] cited in the Slashdot write-up. (I note in passing that this is a wiki, and may change before you read this.) Here are some of the "yes or no" (almost) categories:

Templates or executable macros

No difference in expressive power is acknowledged between LISP macros and C++ templates.

Thread synchronization primitives

With no reference to how expressive they are, and how powerful the idioms supported by them? This one is really telling, IMHO, because I don't believe the future lies in classical thread sync and locking primitives. The whole approach is just too prone to deadlocks and race conditions to withstand the heavily parallel future that multicore chips are starting to bring into the mainstream. When you have ideas like pure function programming languages, operating in a world without side-effects where explicit locking isn't necessary, or interesting ideas for inter-thread communication such as those in Erlang, another variation on built-in pthreads just isn't worth much.

Enumeration

So again, we acknowledge no difference between simple and low-level enumerations such as those in C, and concepts such as disjunctive types and pattern-matching that are very powerful, remarkably elegant, and mainstream in certain families of programming language. Again, this is just papering over a gap, where other languages operate on an entirely different level.

Long double floating point (80bit)

This is just desperation. Pretty much no-one uses 80-bit floating point arithmetic IME (and yes, I do work in the field). The portability hazards and lack of true support from almost every mainstream architecture make them almost irrelevant, except perhaps for a few very small niches.

Lexical Closures

Another telling omission: the power of all those neat functional programming features is dramatically reduced when you can't construct functions on-the-fly.

On top of all of this, the feature lists invariably gloss over some less concrete things that are nevertheless very important to systems programming languages. How portable is D? How many production-quality implementations are available? Is the language standardised or under the control of a single, commercial body? How much backing is there behind the language in the commercial development space; do others write libraries specifically for this language, or is it reduced to using C-style interfaces at the lowest levels anyway, and what impact does this have on the usefulness of features like DBC, exceptions, and so on? Does the language have an active hobbyist/volunteer community supporting it?

I could go on, but I don't want this post to disappear into the oblivion any more than it already will. Although I'm deliberately focussing on criticising in this post, as I often do with D, I keep an open mind and will happily engage in debate with others, or even be proved wrong by people who have found D to have compelling advantages. So go ahead, D advocates, start your counter-arguments here...

Re:

[stevenj (9583)]

Long double floating point (80bit) — This is just desperation. Pretty much no-one uses 80-bit floating point arithmetic IME (and yes, I do work in the field). The portability hazards and lack of true support from almost every mainstream architecture make them almost irrelevant, except perhaps for a few very small niches.

Um, considering that the vast majority of the world's floating-point hardware is x86 and supports extended precision, saying that it lacks "true support from almost every mains

Re:

[Anonymous Brave Guy (457657)]

Um, considering that the vast majority of the world's floating-point hardware is x86 and supports extended precision, saying that it lacks "true support from almost every mainstream architecture" is comical.

Not really. Take a look at the performance, addressing modes, and so on. The support for 80-bit on Intel boxes is not the same as the support for 64-bit, at least in practical terms.

Also, while the vast majority of the world's FP hardware may indeed run on x86 (I don't know), I would suggest that a

Re:

[Anonymous Brave Guy (457657)]

Ironically, I consider most of the C++ to D examples there to be flaws in D.

Supplying predefined comparison operators is all very well, but what if a class doesn't support the concept of equality? Alternatively, suppose it supports only equality and not ordering, or vice versa? How do I do that in a natural way, with a single comparison function to define?

The whole concept of relying on scoped variables completely misses a major advantage of RAII, which is that in the common usage, you can't forget anyt

Re:Do we really need another D infomercial?

[metamatic (202216)]

So the only way to represent the concept of a class with an ordering but no equality is to have a run-time failure every time someone tries to compare them for equality?

What do you think a programming language should do when I try to compare two things that can't be compared for equality?

Lazy Questions

[Quantam (870027)]

Looking at the comparison lists, D looks pretty nice. It has a lot of features that I'd consider switching languages for (from C++), but any such language would have to have a few particular properties (due to the kinds of things I program):
1. Must be able to disable garbage collection and manage allocation explicitly
2. Must be able to allocate classes on the stack
3. Must minimize use of exceptions in the standard library (in other words, exceptions must only be used for exceptional cases)

Java fails all of them, if I recall correctly (I don't know that much about Java, actually). C# fails 2 and 3. It looks like you can disable garbage collection in D, but in the comparison list I didn't see mention of 2 or 3. Does anybody know, off the top of their head?

Re:

[TheRaven64 (641858)]

2. Must be able to allocate classes on the stack

This is a really, really strange requirement. Why would you need to be able to do this in a language? In a high-level language, memory allocation is abstracted away from you so you don't need to worry about it. A decent JVM, for example, creates pools for frequently allocated and de-allocated objects so that they have a much lower cost than heap-allocation of C++ objects.

Objective-C, for another example, follows the Smalltalk model of putting as little in the language and as much in the libraries as

The feature-checklist school of software design...

[stevenj (9583)]

...is obviously the birthplace of D. Heavens, it's the union of C++ (already a nightmarishly huge language), C99, and C#, along with dozens of other features halfway borrowed from functional languages and other inspirations. More features must be better, never mind that the language becomes insanely complicated! (Next they'll be throwing in the Fortran 2000 standard.)

My feeling is that languages shouldn't try to satisfy all possible needs. Rather, we should have small and clean languages, use the right tool for each job, and combine code libraries from different languages when needed. (I regularly use 3-6 languages in a single project and my life is much happier for it.)

(Legacy support is critically important too, but it is vastly better to provide legacy support by providing ways to call older languages, especially the lingua franca of C, rather than demanding that the new language be a superset of the old. I still call numerical libraries written in pre-1970 Fortran, but that doesn't mean I have to write my code in a Fortran derivative.)

Keeping It Simple

[rumblin'rabbit (711865)]

Amen. Here's a quote from the article:

Special attention is given to the needs of quality assurance, documentation, management, portability and reliability.

So where is the word simplicity in all of this? Anybody that has learned to use C++ really well has, to my mind, earned the equivalent of a Master's degree. This makes C++ a brilliant failure. So to learn D will we require the equivalent of a PhD?

Good designers make things both simpler and more powerful. They improve the product as much through s

Python and D

[MightyMooquack (241719)]

One area I see D being useful in is integration with Python. Writing to the raw Python/C API is cumbersome. (Managing reference counts is tedious.) Boost.Python is difficult to build and slow to compile. I've written a library for D called Pyd [dsource.org], whose purpose is not entirely unlike Boost.Python's.

Pyd is easy to use. It provides its own extension to Python's distutils. Usually, you just need to make sure the D compiler is on your PATH, write a setup.py file, and run python setup.py build.

"Hello world" in Pyd looks something like this (and I apologize for the lack of indentation):

import pyd.pyd;
import std.stdio;

void hello_func() {
writefln("Hello, world!");
}

extern (C) void PydMain() {
def!(hello_func);
module_init();
}

EVERY Language Sucks

[JoeCommodore (567479)]

I mean, like, I try to write this program, and like, the language does ONLY what I tell it to, not what I really want. That really sucks.

Can't someone write up some stuff that understands proper english? Geez.

Re:

[istartedi (132515)]

%ecc posting.english -o test
warning 1: ``like" is superfluous teen speak, construct discarded.
warning 1: ``and like" is superfluous teen speak, construct discarded.
warning 3: ``english" assumed to be ``English".
warning 3: ``Geez" undefined. Linking against null action function.
% ./test
The poster's meaning is that he tries to write a program.
We have established that ``That" really sucks, and we are
assuming that ``That" is the poster's act of writing a program.
Can't someone write up some stuff that understan

object oriented

[elmartinos (228710)]

From TFA:

// Object Oriented Programming
CmdLin cl = new CmdLin(argc, argv);

Why do they always insist on this bad syntax? Why can't they write
CmdLin cl = CmdLin.new(argc, argv);
This way you can think of CmdLin as an object, new is just a method that returns a new instance. Voila, one keyword removed, and much cleaner syntax. This would not even offend the Java people because they can have the same syntax with a static method.

Comparison to Objective-C?

[Jimithing DMB (29796)]

I noticed that a comparison to Objective-C is quite conspicuously absent from the list of languages compared to D. Why is it missing? Granted D seems to be a much greater change to C than Objective-C is but I can't help but thinking that one of the main attractions to D seems to be its heap-based garbage-collected object system. You can already get the object runtime with Objective-C. If you use GNU you can even have Boehm GC (which is apparently the GC that D uses). If you use Apple you will have to wait for Leopard to get GC. Another new Objective-C feature is the ability to use full C++ objects as instance variables in your Objective-C classes and do the right thing with initializing (calling the default no-argument constructor upon alloc).

On top of that, Objective-C actually includes tons of reflection information. Although Objective-C has protocols which are roughly equivalent to Java/C# interfaces they are almost completely unnecessary. In Objective-C one can query at runtime whether a method is implemented or not and if so call it. So whereas in Java you'd do this:

if(anObject instanceof MyInterface) ((MyInterface)anObject)->doSomething();

in Objective-C you can do this:

if([anObject respondsToSelector:@selector(doSomething)]) [anObject doSomething];

The difference being that in the Java case you have to declare MyInterface as containing the one doSomething() method and inform java that your object extends MyInterface whereas in Objective-C you merely need to provide a doSomething method on your object.

Basically that means that in Objective-C every single method effectively becomes an interface. You would not believe how useful this is once you realize it. Note that at runtime there is ZERO difference. In both the Java and Objective-C cases the object is being checked to see if it implements something. Same with C++ if you use dynamic_cast<>()

Granted every language has its niche and I'm sure D will find its. Objective-C's niche is definitely GUI programming. The ample reflection information allows for easy implementations of archiving (serialization) and most importantly key-value coding and the related action methods pattern. It's a pretty damn cool thing when your RAD tool simply outputs archived objects that refer to methods to be called upon certain actions simply by name.

Missing the point

[pammon (831694)]

The largest problem with C++ is its complexity, which causes the language to fragment into partially intersecting sublanguages representing each programmer's comfort zone. D does not address this problem; in fact, it makes it much worse. Example? C++ has seventy some keywords, but D has more than ninety. Another example? See if you can guess what this means in D:

0XA1_2_C35_4_5_6_5P6_Li

Give up? Why, it's an imaginary real. Yes, you read that right. Oh, and I specified it in hexadecimal, just to prove I can. And I threw in some underscores because you're allowed to do that too. Have you guessed what an imaginary real is yet? This is supposed to be the "simpler C/C++ replacement."

A good language strives to find a few constructs that solve multiple problems. A bad language takes the kitchen sink approach. C and C++ are not good languages, but wide deployment and support make them useful anyways. D does not have that, but the designers do have the luxury of learning from the mistakes of the past. So what the hell is their excuse?

OOP languages are slow

[bluefoxlucid (723572)]

CPUs put a lot of stock in branch prediction; due to the nature of OOP languages like C++, Objective-C (I like this one), and D, this doesn't work. The way virtuals and class inheritence works, functions are necessarily dealt with as pointers; the function is pointed to by pointing to a master class object, basically. Here's a C reconstruction:

struct myClass_members {
struct myClass_members *(*alloc)(); // constructor
void (*destroy)(struct myClass *); // destructor
int (*my_member)(); // member function
};

struct myClass { // Data
struct myClass_members *call; // Pointer to a list of members as above
int my_value; // an integer value
};

What you do is initialize a constant myClass_members (called myClass_Object here) with a bunch of pointers to static functions in one source file; then call myClass_Object->alloc to create a new one (we'll call it my_inst). Then do my_inst->call->my_member() to call the member, and similarly my_inst->call->destroy(my_inst) to deallocate the class.

Basically, OOP languages like C++ and Java use this methodology, but it's obscured through friendly syntax. What we can expose from the above is:

1. When class members change, structs change. This causes binary incompatibility between different versions of libraries. The exception is Objective-C, which looks up members based on their name in a hash table generated at run-time
2. The addresses of branches (specifically CALL to call a function) are indirect (yes, in Obj-C too); this means that you can replace the class with another class that has the same structure but different functions being pointed at. It also means that the CPU can't do branch prediction, which hurts pipelining and intelligent CPU caching, causing pretty big slow-downs.

The whole "native execution speed" thing is bunk. Script languages are executed on a native bytecode interpreter, or JIT'd to native. The amount of work that goes into the execution is what you care about; as well as the utilization of the CPU's most powerful facilities. You can only justify OOP languages by saying that either A) the majority of the work doesn't involve making calls to other class members, and thus won't be hurt by this; or B) CPU speed doesn't matter. I hate argument (B); (A) I can accept, barely enough to tip my hat to you for having good software engineering sense.

Re:

[Nasarius (593729)]

Try using .NET for systems programming. Or projects (I've made several thousand dollars from one so far) that must be portable to NT4, 2K/XP without requiring your clients to install extra junk on every computer, etc. Contrary to what Microsoft may want you to believe, .NET is not the solution to all the world's problems.

Re:

[TheGavster (774657)]

What, exactly, is the benefit of the .Net VM? There is only one full implementation of .Net (the MS one), and it runs on a single platform (Windows on x86). You might as well build native x86 code linked against Windows libraries for all the portability you have. And even if you're going to bother implementing the VM across a bunch of platforms, why not implement a standard library across a bunch of platforms and link native executables against that?

Re:

[Heembo (916647)]

In .NET it's called a Common Runtime Library, running MSLI code. (That sentence is analogous to "the Java Virtual Machine runs Bytecode"

The big win in .NET is that there is built-in administrator controllable security (even pub/priv key security) between the CRL (or vm to you) and the internal .NET framework. In fact, there are several administrative controllable hooks built into the .NET framework that we just do not see in Java and Ruby and the others. This is the feature that separates .NET from the r

Re:Another 'Toy' Programming Language

[nacturation (646836)]

I'll stick to languages that have at least a professional following, in use by real software engineers (that is, engineers who can legally call themselves engineers), not the run of the mill code monkeys.

I'm dying to know... do you mean Fortran or Lisp?
 

Re:GC, No Vm or performance hit

[Nasarius (593729)]

garbage collection ... No virtual machine How do they square that particular circle?

It's really not that difficult. Hans Boehm wrote a garbage collector [hp.com] for C/C++ years ago, which happens to be the same one that the Digital Mars implementation of D uses.

Re:

[StrawberryFrog (67065)]

If I read that FAQ right, it is possible that "integer or other random data be misinterpreted as a pointer by the collector" since given the nature of C - no VM, the difference between a pointer and an int is at best a gentleman's agreement - anything in memory *could* be a pointer. Well, I suppose it works if he says so. But it certainly isn't pretty.

Re:

[rbarreira (836272)]

Here is something [hp.com] that I found about this. Not very good news for conservative garbage collection, I say...

Re:GC, No Vm or performance hit

[Anonymous Brave Guy (457657)]

garbage collection ... No virtual machine ... How do they square that particular circle?

The same way as countless other programming languages have in the past, I imagine. Why do you think garbage collection requires running your code under a VM?

Just In Time Compilation in C# or Java has "Native code speed", in fact it goes one better - since the compilation happens at a later time, more processor or other specific optimisations can be made.

Of course, you're overlooking all the overhead of monitoring the code long enough to determine which on-the-fly optimisations are worth performing, and of compiling the code itself, neither of which is trivial.

GC has a lot to do with the perceived slowness.

True, though of course it's not without overheads. Almost all of the Big Claims(TM) made by GC advocates in these discussions come with a catch: state-of-the-art GC method number 17 has a lower amortised cost of memory recovery than explicitly freeing it C-style!*

* But only if your system contains 10x as much memory as the program will ever need anyway.

This is traditionally followed by a wisecrack about how memory is cheap, followed by three enlightened posters pointing out the stupidity of that argument for multiple reasons. :-)

Isn't it disingenuous to tout both "native code speed" and "garbage collection"?

That depends a lot on context. If you really have a system where the overheads of GC are trivial but all the advantages are present, it seems a fair claim. It's just not likely to be universally true, and representing it as such would indeed be disingenuous.

Re:

[rbarreira (836272)]

Hey hey, calm down. Are you sitting now? Ok, read:

Programming languages are to intended to be used by humans.

Re:I have a problem with GC in a systems language.

[Animats (122034)]

I wasn't happy about that either. Garbage collection in a language with destructors leads to wierd semantics, which is why Microsoft's "Managed C++" is a nightmare. I corresponded a bit with Walter Bright in the early days of D, but didn't press the issue.

What seems to work in practice is reference counting. GC gets most of the academic attention, but Perl and Python are both basically reference counted, and the result seems to be that programmers in those languages can ignore memory allocation. Java programmers have to pay a bit more attention, worrying about when GC will run and when finalizers will be called. Reference counting is deterministic; the same thing will happen every time, so timing is repeatable. That's not true of GC.

There are two basic problems with reference counts - overhead and cycles. Overhead can be dealt with by hoisting reference count updates out of loops at compile time, so that you're not frantically updating reference counts within an inner loop. Hoisting (along with common subexpression elimination), by the way, is also the answer to subscript checking overhead.

Cycles are a more serious problem. Conceptually, the answer is strong and weak pointers (in the Perl sense, not the Java sense), which allows the programmer to express things like trees. (Links towards the leaves should be strong pointers; back pointers towards the head should be weak pointers.)

In practice, cycles aren't a serious problem, because they're generated by design errors and tend to happen in normal program operation, so they show up early in testing as memory leaks. Dangling pointers, on the other hand, tend to show up in error cases, which is why they survive testing to become delivered bugs.

Ideally, you'd like to detect cycles at the moment they're created, at least for debug purposes. This is quite possible, although there's substantial overhead.

Attempts to retrofit reference counting to C++ via templates have been made, but they are never airtight. To get anything done, raw pointers have to leak out, which makes the reference counting scheme very brittle.