Howdy! I have some questions about the socket programming structures.


Now. Here is a "simple" program from Beej`s guide to network programming. As u see, there is a struct pointer. Where is that pointing to!? I haven`t made a structure. Now, maybe im asking a bit silly questions but, i havent even made a structure. Is that structure somewhere in the header?
Can someone show me how it works and give me some basic explenation on how this is working?

The structure should be defined in the header, yes. I suppose netdb.h is that header (winsock.h on Windows) If you look closely at "Beej's Guide", you'll see he shows the structure of the structure


The memory allocation is (or should be) done in the gethostbyname function.
Now, I'm not a C expert, but I'm pretty sure that he's example is flawed. There is no memory freeing what so ever... Of course it will be done by the system when the program ends, but this is not the proper way to write a program if you ask me. I suppose he wanted his guide to be easier on beginners, but teaching incomplete techniques may be worst than not teaching them at all...
If anybody has a different opinion, please correct me. The link for this part of the guide is: Guide

Best regards,
Lucian

__________________
Please read these Guidelines before posting on the forum

"A person who never made a mistake never tried anything new."
Einstein

The function gethostbyname() (and others: gethostbyaddr(), getservbyname(), getservbyport()) typically returns a pointer to a static structure so you don't need to allocate or de-allocate memory. Unfortunately, for some applications, that means that these functions are not re-entrant, and, therefore are not thread-safe. In fact, the POSIX specification explicitly states that gethostbyname() need not be re-entrant.

Some implementations (gcc for Linux, for example) have functions gethostbyname_r() and gethostbyaddr_r() that are re-entrant (you pass a pointer to enough memory to hold everthing). As far as I know (and I could be wrong) there are no standards for the _r functions --- look on the man pages for your system to see what's up with them. I have not seen any of the _r functions in any headers for Windows compilers that I have (including cygwin/gcc).

Regards,

Dave

Thanks alot

But im having trouble with all the use of pointers in socket programming, and it would help me if someone could transform this into "human language"


Now am i right on this one:

inet_ntoa is a function to return your ip address.
starts with pointing at a structure called in_addr wich is defined inside a header file. And then points at eh, im not sure how this -> works.

You can try to work from the inside out. First of all, what is h?


So h is a pointer to a struct hostent.

Somewhere in the header files (netdb.h, I think) there is the definition of hostent. It looks something like this:


Now, the notation h->haddr is shorthand for (*h).h_addr (that is, it is the member of the struct pointed to by h). You had better get used to it, because the -> is ubiquituous.

So h->h_addr is a pointer to the beginning of h_addr_list (h_addr_list is a pointer to an array of addresses).

Now, we want to pass the string corresponding to this address to the function inet_ntoa. Now the argument to inet_ntoa() is an in_addr structure (note: not a pointer to struct, but a struct). To make a long story short, the first, and only, member of the struct we send to inet_ntoa() has the type in_addr_t, which is a pointer to the address that we want inet_ntoa to resolve for us.

Here's how it breaks down:

This is a pointer to the address

This makes it look like a pointer to an in_addr structure:

Now, this makes it an in_addr struct:

That becomes the argument of inet_ntoa.

Result (using the exact code you posted):

Regards,

Dave

Postscript: There are two ways that I could have approached the problem (there are, obviously more than two ways to approach the problem, but only two ways that I would have used).

Top-down: Start with inet_ntoa(). See what kind of argument it requires, and figure out what it will do with it. Try to figure out how to get that from what it has to work with.

Bottom-up: Start with h. See what is being done with it (->, casting result to another type of pointer, etc.)

Sometimes when I am actually writing code, I think that I am, maybe, the last Bottom-up guy in a Top-down world, but that's how I usually try to see things. Or rather: Top-down for design. Bottom-up for implementation and, especially, for analysis.

You see, when you start from the top, there are lots of branches. Like a good detective story: false clues and red herrings that cause you to backtrack. Good for your reading enjoyment, and probably good for creative design, but kind of frustrating when you really need to get to the bottom of things (and you already know what the bottom has to be).

On the other hand, starting at the bottom, you have something definite to start with. Regardless of how many ways to get there, you are there. Then go up to the next level, etc.

Thanks alot. This helped me much. And it made me realize that i have to learn more about structure pointers and linked list ( i think its called? -> )

Again. Thanks for nice replys

Thanks, I didn't even think about that!
To be honest, I don't like it, but what do I know

Best regards,
Lucian

__________________
Please read these Guidelines before posting on the forum

"A person who never made a mistake never tried anything new."
Einstein

what use do we have from argc and argv?

The int argc tells us how many things were entered on the command line.
if argc == 1, then only the program name was entered; if argc > 1, then other arguments are available. The variable argv is an array of pointers to char. Each pointer to char is treated as a "string" in C (there is no such thing as a "string" data type).

A "C-style string" is a sequence of chars terminated by a zero byte. This construct is a fundamental part of C language library function implementations. But don't let the terminology confuse you; there is no such thing as a "string" data type in the C language.

All standard C library "string" functions operate on this type of construct. This includes strcpy(), strcat(), strlen(), etc. Also things like fgets(), and printf() with %s format specifier. "String" arguments to these functions are "pointer to char".

Did I mention previously that if you want to get anywhere in C you just gotta glom onto the concept of pointers and look at lots of examples of their uses? Oh, yeah, I guess I did.

Note that in C++ there is a "string" class that is part of the Standard C++ library. It is quite a different animal, and has lots of interesting and useful characteristics not available in C.

Note, also, that in the Microsoft Foundation Class, there is a Cstring variable type. This is a special data type, not (not) the same thing as I am referring to here as a "C-style string". One more time: there is no such thing as a "string" data type in the C language.


Let's stick to C for now.

For example, the program is invoked with the host name as argv[1].

You can try this:


I named the program source file "testargs.c"

Then, on my Linux system:


Try it with other things on the command line:


Regards,

Dave

Well, i know when you start the program then argv[0], will automaticly be the location and name of file. Something like this: C:\dev-cpp\test.cpp ( Yes i have windows ) But here is one "strange" thing im wondering about.

I have looked more on the code examples to beej`s guide, and getting closer to understanding alot in the code. But here is a strange thing:

how is the host info gathered from this chunk of code? can someone plz explain abit about how this works? the he=gethostbyname(argv[1])) == NULL part really confuses me. What is argv[1] doing there!?