|
Without wishing to start a major language debate, I think its important to understand that C was designed to be efficient at using the underlying hardware of the machine, and the way things are declared reflects this:
char *sz declares a single variable that points to a place in memory, supposedly holding a char value. This pointer (and here's where dragons lie!) can be made to point to anywhere and to anything regardless of its declaration, you are just telling the compiler that at the place where that pointer points, YOU consider there to be a char value (whether it is or not in reality).
char sz[1024] says that at the address represented by sz is a reserved block of memory that holds (in theory) 1024 char items.
In your code you can (with some severe caveats!) use both variables in the same way and it's up to you to remember and manage whatever you think you are looking at in that location - this is both the beauty and danger of a language like C - it allows you to manipulate things in much the same way as in assembler and with as few restrictions, but it provides no protection against you doing something stupid.
However, even the early compilers would detect if you attempted to treat these two variables as exactly equivalent, but would often only warn rather than prevent it, allowing you to create absolute havoc.
Much code rot is caused by the developer incorrectly assuming that, by default, either of these variables is initialised (which is why most compilers these days will attempt to put something sensible in newly declared variables to protect the innocent). Before initialisation char *sz may contain a random address and even attempting to look at it might cause a total system crash if it accidentally points into hardware protected memory. With char sz[1024] though you can be sure that doing char x = *sz is safe because sz already holds a valid address and points into memory allocated to your program, what you don't know is what value you will actually get back from that place before the array is initialised.
Why have I bored you all with this stuff? Because so many developers these days seem to know absolutely nothing at all about how the hardware they are driving works - that's fine if you are writing in a high-level domain specific language where everything like that is hidden - no help at all if you are writing a device driver for some complex piece of hardware that is integrated into an operating system of some sort.
|
|
|
|
|
I think you may have misunderstood me
char[] sz = char[100];
would be more consistent even if it requires extra typing.
I'm not complaining about the way accessed either, just how it's declared. I don't mind that either way sz can be accessed via pointer or array index.
In this case you could dereference it with a * like you can otherwise. I'm not arguing you shouldn't be able to.
It's not a huge deal for me either. I've coded in the C and C++ for years so I'm used to it.
It's just a syntax wrinkle i don't like.
Real programmers use butterflies
|
|
|
|
|
|
You, in your last paragraph, point out my pet peeve with C#;
Everything is separated by '.' - it makes things easier and detaches the developer from what the object really is. Value in enum? Member in a class? Namespace?
OK - they worked to make all of this (seem to) work together, smoothly, but it also, by shielding the users from reality, fosters and allows the maintenance of a degree of ignorance. Essentially, C# is a dumbed-down version of C++ . Hardly a surprise as the original stated intent of MicroSloth, when they created it, was to wean users away from vb.NET .
Probably this type of pseudo-rant "points" to my stogie attitude developed with age. Or that I learned assembler before C .
Quick Anecdote: when my brother got his first computer I worked on getting him to master DOS (a while ago!). At first he started but then he stopped: they installed Windows. I asked him how he'll fix anything if there's a problem, without understanding what Windows was running on. His answer? He'll ask me.
And so it goes with C# vs C/C++ background developers.
| Ravings en masse^ |
|---|
| "The difference between genius and stupidity is that genius has its limits." - Albert Einstein | | "If you are searching for perfection in others, then you seek disappointment. If you seek perfection in yourself, then you will find failure." - Balboos HaGadol Mar 2010 |
|
|
|
|
|
Java's stupid (=political) C compatibility
C started as a simple language on top of assembly language. Lingual considerations of abstractional consistency had no priority. As error awareness. Everything is an int, boolean, pointer, array, functions.
In fact declarations have a weird kind of syntax requiring a large parsing. Or for a human a bit of back-and-forth reading.
As java was launched, it intended to improve some parts of C++/C. So in java the entire type info becomes separate from the variable.
char[] jarr;
But as novice java kept the C style for accustomed C/C++ programmmers:
char jarr[];
If one looks at java's:
char[] jarr = new char[1024];
one get's an idea why C chose:
char str[1024];
The 1024 is not part of the `char*` data structure as field, but some compile time allocation.
In general one should accept C as it is: a compiler strong, thin layered, fun language.
For language features it is pedagogical unsuited w.r.t. type systems, actual data, error prevention and so on. Taking any language, like a simple Pascal, might give less confusion with pointer versus array like in parameter passing. Then going back to C one sees that arrays are passed as pointers i.o. the pointed-to data as in Pascal.
|
|
|
|
|
When I was learning C, it bothered and confused me. Now, I recognized it as the type and it doesn't bother me. Is that age, tolerance, or just exposure to too many languages???
|
|
|
|
|
Yeah I've coded in C and C++ for a long time, to the point where it doesn't confuse me. I just don't like the inconsistency. I understand why it is from a syntax perspective.
In c#
int[] array = new int[10];
But in C it doesn't make sense to use new, and int[10] array; is ugly.
Still, it just bugs me, because they're both type modifiers.
Real programmers use butterflies
|
|
|
|
|
You are thinking about it the wrong way, I believe.
There is an implicit Algebra: * = 1/&
so, char *p, LITERALLY says (*p) is a char.
therefore &(*p) => Address of that char!
but knowing the algebra & and * cancel each other out (as do & and [)
So, p is a pointer to a specific char. And the beauty of this, is that how you declare it usually implies how you are going to use it. A fixed array is declared as char a[10]; Which again, says (to me), I want an array of 10 characters, a POINTS to the first character, and a[1] == *(a+1).
I always found this part of C very expressive and self-explanatory.
Of course, I learned PDP/11 Macro-11 Assembly BEFORE I learned C. So most of this was mapping to register level access of R0 vs (R0) [Memory direct and indirect addressing].
Mov 65, R0 ; Stores the value in the register
Mov 65, (R0) ; Stores 65 at the memory location of R0
Mov 65, (R0++) ; Stores 65 at the memory location and increments the memory location AFTER to the next
; Note its been 35 years, the syntax is illustrative of the features.
If you know how the variable is declared, you should instantly understand the various ways you can dereference it.
|
|
|
|
|
We seem to be talking about two different things.
I'm simply talking about type modifiers and how they are applied to types. In C the pointer modifier comes after the type, and the array modifier comes after the variable or member declaration. It's just a slight syntactic (purely cosmetic) inconsistency that has irked me for the years that I've used C and C++.
Real programmers use butterflies
|
|
|
|
|
Okay, I apologize.
Now, consider this, how do I declare 3-5 arrays of different sizes?
char a[2],b[5],c[6],d[7];
char [2]a, ...
char[2] a;
char[5] b;
char...
also, neat trick, 2[c] == c[2];
// because ANSI declared n[c] == *(c+n), and therefore c[n] == *(n+c)
I think I would be confused if it was any other way than it is now.
The cool thing is you could use the preprocessor to make it work like you want (and be threatened by every living programmer for HORRIBLE form, LOL). [I used to use the preprocessor... A Lot... LOL]
|
|
|
|
|
wait, are you declaring char [2]a backward to illustrate a point?
just to be clear literally I'd ask for is a single change from this:
char sz[100];
To this
char[] sz[100];
Even if it requires extra typing it would be more consistent.
Everything else would be the same.
And I'd never abuse the preprocessor that way with the exception of using it to help code complicated compile time computations and make writing them easier. Like the Spirit parsing framework.
Real programmers use butterflies
|
|
|
|
|
Ah,
therein lies the dilemma:
What does
char[] a[100], b;
declare b as?
Worse, when I read "a", it says: "a[100]" is an ELEMENT (or pointer) to an array,
Because I see it as:
char []a[100];
char *a[100];
and
char[] a(100);
The * modifier and the [] modifiers are applied to the variable, not the type.
hence
char *a,**b,c[100],d;
// And yes the [2]a was just "putting it in front, as requested"
But I feel we must "adapt" to nuances of every language. I primarily code in Delphi (Pascal based) for the last 20+ years... So I am having to THINK HARDER about C and C# syntax.
That adaptation in C is pretty low, and coding standards would say the Decoration stays with the variable, not the type! That rectifies about 80% of it to me.
Proof:
char* a,b,c;
char *a,b,c;
IMO...
|
|
|
|
|
You make a good point. At which my response is I don't like that C puts the type modifiers with the variable in general.
C# doesn't, and i appreciate it for that, but I guess I've been spoiled as this last year i was doing mostly C# stuff.
Real programmers use butterflies
|
|
|
|
|
As I understand it declarations are written as if operators e.g. you would write "name[123]" to access the 123'rd element in the array named "name" so that's how it is declared. Maybe it should be "[123]name".
|
|
|
|
|
You seem to be talking about accessing the array, not declaring it. I'm talking about declaring it.
Edit: Or if you are, then I think get your point? My response would be that pointers aren't declared like pointer operations.
Real programmers use butterflies
|
|
|
|
|
They aren't? I would declare a pointer using the rules of operator precedence always starting with the identifier e.g. "char* name" and utilize it as an operator as "*name" so except for the white space and the casting operator "char" which is no longer needed they are the same. Cheerios
|
|
|
|
|
I think we're misunderstanding each other. My OP maybe wasn't as clear as it should have been.
I've always declared pointers with the * next to the type. You CANNOT declare arrays that way. Hence my complaint.
Real programmers use butterflies
|
|
|
|
|
Greetings My point is that if you accept the rules of operator precedence and associativity which I am assuming you do than you must accept the rules of declaration since they are identical. Otherwise you must argue against both and insist e.g. that array elements be referenced as "[123]name" or pointer targets as "name*". Or perhaps you would argue that the rules of declaration be different from the rules of operation but that would complicate things. Kind Regards Cheerios
|
|
|
|
|
Oh, I get you. I guess using operator precedence on type modifiers kind of threw me, but i can see why you look at it that way. 
Real programmers use butterflies
|
|
|
|
|
Greetings and Kind Regards Just in passing I wish to mention in case you do not already know about it is that the C text by Harbison & Steele is what the K&R text wished it was. H&S is a beautiful text and is where I learned that declarations and operations follow the same rules. Why K&R didn't explain this is as simply is difficult to understand. Best Wishes Cheerios
|
|
|
|
|
You could use 123[name] as an alternative - that matches the original PDP addressing mode syntax better as well, I suspect!
Java, Basic, who cares - it's all a bunch of tree-hugging hippy cr*p
|
|
|
|
|
honey the codewitch wrote: I think it's inconsistent, and I think the array specifier should have been declared with the type since it's essentially a type modifier like * and &
One thing it's not is inconsistent - the pointer modifier belongs with the variable, not the type. For example, the following fragment declares a pointer to integer variable and an integer variable.
int *pa, a;
See this Godbolt...
One option, were you using C++...
template<class T, size_t N>
using Array = T[N];
template<class T>
using Ptr = T*;
Array<int, 10> test_array;
Ptr<int> test_pointer;
Java, Basic, who cares - it's all a bunch of tree-hugging hippy cr*p
|
|
|
|
|
Stuart Dootson wrote: One thing it's not is inconsistent - the pointer modifier belongs with the variable, not the type.
Yet it's a type modifier. A pointer to an int is a different type than an int.
Real programmers use butterflies
|
|
|
|
|
Quote: Does it bother anyone else that you declare a pointer like:
char* sz;
It only looks odd if you use it like that. If you use it like the way it was meant to be used:
char *sz;
It makes sense. The way you write it is not consistent and lends itself to errors.
char* szA, szB, szC;
char *szD, *szE, *szF;
Putting the "*" next to the typename is logically inconsistent - you still have to put the "*" in the correct place for other pointer types:
void (*fptr) (void);
So instead of doing it one way for some variables and the correct way for others, just do it the correct way for all.
|
|
|
|
|
I suppose then that I do not like that type modifiers are not declared with the type.
Real programmers use butterflies
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
