Files holding data

[neandrake]

I have to write a program that can access a file, read it, and hold it inside a dlinked list for database access. I was wondering though, the EOF character inside a file indicates that it's the last character of the file, right? So what if you write more data to the file after the EOF character? Where does the data go, if anywhere? Is it accessible? Thanks,

christoph

[sean]

Picture the following:

 char months[13] = "JFMAMJ\0JASOND";

It's stored in memory as simply 13 bytes in a row associated with eachother just like any other string. But it's generally accepted that a string ends at the null ('\0') character, so passing something like this to a function that counts the words in a string would be the equivalent of only passing the first 7 elements of the array, because the function is programmed to stop at a null character.

[neandrake]

But would it be possible (using higher access functions) to write past it?

[hk_mp5kpdw]

From the FAQ, EOF is not a character. A read function may return a value that equates to EOF when an error has been encountered or when there is no more input to process.

[anonytmouse]

>> But would it be possible (using higher access functions) to write past it? <<

Yes, of course, the file can be expanded. Why don't you try it? There is no actual EOF character at the end of files on modern systems. The size of the file is stored instead. The C/C++ file functions will return an EOF value when the end of the file is reached. However, this value is added by the C/C++ I/O system and does not physically exist in the file.

[neandrake]

>>The size of the file is stored instead.

Where is this information kept? I tried opening small files in a hex editor, but I didn't see any information about the size of the file. Or what you are talking about is file specific? Like, special files registered to software keep that as part of a format?

[anonytmouse]

>> >>The size of the file is stored instead.
>> Where is this information kept?

It is not stored in the file. It is stored by the file system for each file. Just like the name of each file is stored, its last modification date, which directory it is in, etc.

Let's look at an imaginary file system. Each file entry may look something like this:

struct fileEntry
{
    char fileName[MAXPATH];
    char parentDirectory[MAXPATH]
    int flags; /* Read only, archive flags,  etc */
    time_t creationDate;
    time_t lastModDate;
    unsigned long long fileSize;
    unsigned long long positionOnDisk;
    sid securityDescriptor; /* Access restrictions, owner, etc */
};

Each file would have an entry. Most of this information can be retrieved with various functions. The only information actually stored in the file is the data you write to it.

[neandrake]

Thanks for the insight, it's clearer now. So it wouldn't be possible to change the size of the file that is stored in the FS without the data being cut off?

[anonytmouse]

>> So it wouldn't be possible to change the size of the file that is stored in the FS without the data being cut off? <<

I'm not sure what you are asking. Obviously, if the file is made shorter data will be lost, while if the file is made longer data will not be lost. There is no function in the standard C/C++ libraries to shorten a file (without starting from scratch).

[Hunter2]

I imagine data-retrieving utilities work in the reverse of this (by looking beyond the bounds of the file system's set filesizes). Of course, that's also why they're not always successful, since you'll get stuff overwritten/corrupted when you don't have any protective markers saying "This here memory's part of a file".