New child language - (Function member address (structure))

[Aramil of Elixia]

why not just use python. its a scripting language and google uses it which is good enough for me. and scripting languages are programming languages

[chrisname]

It's undefined behaviour (bad) to read a member of a union after writing to a different one.

[hamsterman]

union z_Data{

bool bit[64];
long l;

}data;

Takes 64 bytes because bool takes 1 byte. You should probably just do l & (1 << i) to get the i'th bit. Or use a bitset. Though why would you need this?

I'd like to know some advices and suggestions : ...

I did suggest to just do integers and also not to borrow too much C++ syntax. Also, these look like array declarations.
Did you look at BNF? If you have, you should be able to write a grammar for this yourself...

[LB]

hamsterman wrote:

Takes 64 bytes because bool takes 1 byte.

On all compilers I've used, sizeof(bool) is 4. Or 8, if you're compiling for 64-bit.

[chrisname]

My guess is it's 4 or 8 bytes because CPU time is more expensive than memory, and CPUs are slower when operating on integers bigger or smaller than their native size.

@Aramil of Elixia
Just because Google uses something, doesn't make it good. Google also uses Java and Go, after all.

[Cheraphy]

L B wrote:

On all compilers I've used, sizeof(bool) is 4. Or 8, if you're compiling for 64-bit.

That always bothered me. Why does bool have to be anything larger than a single bit?

[chrisname]

It's because it's faster for the CPU to work on data aligned to its native integer size (so 4 bytes aligned data is faster on a 32-bit CPU).

[S G H]

On all compilers I've used, sizeof(bool) is 4. Or 8, if you're compiling for 64-bit.

I always found it to be 1 byte... Weird. But I suggest using chars, they have fixed size hopefully.

That always bothered me. Why does bool have to be anything larger than a single bit?

Because the 'sizeof' operator returns bytes, not bits, so sizeof'ing a bool will result in a null size or wrong anyways.

[LB]

As stated, it is for efficiency. It is much faster to deal with the native integer size than a single bit. If you need to have a lot of booleans, use the std::vector<bool> specialization - it uses compact storage at the cost of efficiency.

If sizeof(bool) is 1 for you, either your compiler isn't optimizing for speed or you're compiling for an 8-bit system.^{[citation needed]}

[helios]

sizeof(bool) is usually 1. sizeof(A) where A is a type that contains only a bool may or may not be 1.
The compiler doesn't pad absolutely everything, or the code it generates would be really inefficient in space. Imagine creating a buffer of chars that takes up 4 times as much space as necessary.

[S G H]

L B wrote:

If sizeof(bool) is 1 for you, either your compiler isn't optimizing for speed or <Insert Random Joke Here>

Well that's all I remind from experience, I'm used to test everything in debug mode - But yeah, I believe that for a CPU it's faster to handle a CPU-register-sized variable than a even-less-sized variable.
Again, because of this fact, and because you can't use efficiently the bool's data, I often use the char/unsigned char's if I need to store more bools and handle bits.

[Imadatobanisa]

Firstly, thanks everyone who replied me and give me helpful information, and hamsterman, who helped me solving my 'union' question!!!

Second, now, I'm confused. A programming language interpreter (a parser language) always needs a very complex structure (In my opinion at least it contains a primary loader class and many child data structures - In short it's called "Tree structure") So IMO That is a very huge step - I think.
Recently I've tried making an interpreter structure... But it's quite complex and looks awful...So It really gives me a headache... Finally now I've decided to delete it immediately - I don't want to see it anymore, and after that I have to start again. :( Important categories : Variables (Basic and structures), structures (class, struct), functions, const definition, commands, code tracking (goto), array, multiple dimensional arrays, functions inside structures, handmade functions, attributes, returning value, parameter list, multiple structure accesses (sirparent-> parent-> child -> subchild...), value data, data pointer etc........??????....

I would like to discuss and know how to build an interpreter structure. Which way is best, simplest, shortest, smartest? Which way is most comfortable, convenient? To build an interpreter properly, I'll need to know helpful comments, suggestions, opinions, and advices... It's really a big problem and this is (maybe) the most important step. And more importantly, I've already defined my own grammar completely. I've just written an example and now I'm totally ready to start the interpreter construction. But I don't know where the right start point is...
(Any help would be greatly appreciated <:))

[helios]

My first suggestion would be to trim down your list of features by two thirds or more.

The simplest possible language (to implement) is a stack machine:

//*(1) = 10
push 1
//stack: 1
push 10
//stack: 1 10
store
//stack: empty

//*(1) = *(1) + 2
push 1
//stack: 1
push 1
//stack: 1 1
load
//stack: 1 10
push 2
//stack: 1 10 2
add
//stack: 1 12
store
//stack: empty 

Basically, it's a more verbose form of reverse Polish notation. Once you can evaluate simple arithmetical expressions, expand the language with jumps and branches:

//*(1) = 0
push 1
push 0
store

loop:
//*(1) = *(1) + 1
push 1
push 1
load
push 1
add
store

//if (*(1) != 10) goto loop;
push 1
load
push 10
cmp
jump_if_eq break
jump loop

break:

[Imadatobanisa]

@ helios (10057)

I like assembly...but maybe It takes a plenty of time...
I'm taking about C++... So ? New assembly?... This is really a gigantic step, IMO...
Edit : ("New child language" not "New programming language") <:)

[helios]

Who are you going to trust? Your misguided intuition, or the advice of more experienced people?
Implementing an Assembler is two or three orders of magnitude simpler than implementing a C++... well, a C++ anything, really.

[Imadatobanisa]

@helios

You said making a programming language is easier and simpler than making an interpreter (script) language?
I know Assembly -> everything (language etc)...
But I have no idea. Should I choose the assembly interface - it can make a new programming language (very high-performance), or a programming language interface....???

[Imadatobanisa]

Actually, I'd like to know is there any way which can make an interpreter with a programming language. And also I want to know how to run the code immediately after the code checking process.
Therefore, I can't wait anymore...!! The construction process should start now...

Oh that... 333... My lucky number... <:D

[helios]

You said making a programming language is easier and simpler than making an interpreter (script) language?

No. First of all, a scripting language is a programming language. Second, I'm not suggesting any particular form of execution (although interpretation is easy to implement).
What I'm saying is that implementing the language I described above is ridiculously simple compared to this:

Variables (Basic and structures), structures (class, struct), functions, const definition, commands, code tracking (goto), array, multiple dimensional arrays, functions inside structures, handmade functions, attributes, returning value, parameter list, multiple structure accesses (sirparent-> parent-> child -> subchild...), value data, data pointer

Should I choose the assembly interface - it can make a new programming language (very high-performance), or a programming language interface....???

I have no idea what you're asking.

[Imadatobanisa]

Should I choose the assembly interface - it can make a new programming language (very high-performance), or a programming language interface....???

Sorry, by that time I was confused... [:D
Um... I'm now constructing the script programming language. Certainly it's based on C++, and the start point will look like this :

(This is the simplest structure I think... :))

class lngData
{
vector <int> nData;
vector <float> fData;
vector <double> dData;
vector <long> lData;
vector <short> sData;
vector <char> cData;
unsigned char nAttr;
};

class Function
{
vector <lngData> param;
unsigned char nAttr;
};


class lngParser{public : 
char ReadFile(const char *FileName,const char *Source);

vector <lngData> data;
vector <Function> func;
};

And, I want to define a two-dimensional (or multidimensional) vector. How to do this?

Any suggestion?

[Imadatobanisa]

How to calculate ∆¹²³? And, I want to apply bitwise operators for float and double values... How to do this?...