ANONYMOUS wrote:
> Hello,
>
> I went to see Amitava today to ask him about file pointers vs file descriptors, I understand what a file descriptor is but I was told by Amitava that a file pointer is just a pointer that points at the file descriptor. But other sources on google and the reading provided says something like
>
> "When we do any kind of I/O in C, we do so through a piece of data that you get in the form of a FILE* type. This FILE* holds all the information needed to communicate with the I/O subsystem about which file you have open, where you are in the file, and so on."
> (from
https://beej.us/guide/bgc/html/split/file-inputoutput.html#file-inputoutput one of the readings)
>
> which seems to indicate that it is more than just a pointer to the file descriptor and stores other information. Im not sure if I misunderstood him or my quesiton wasnt clear enough. If it is the case that its just a pointer to the descriptor why is fopen vs open used in some cases.
>
> The question below is the reason i ask, although I did not show Amitava the quesiton when asking him.
>
> "With respect to support for file input and output, provided by both the C11 standard library and an operating system, explain the relationship between file pointers and file descriptors.
> Explain why, or why not, it is possible to mix the use of file pointers and file descriptors within the same C11 program."
>
>From my understanding, file descriptors are low-level system calls USUALLY handled by the operating system that returns a single int which represents the currently opened file. Remember that C is a program that is used to contact directly with the kernel so we will have the option to fine tune as much as we want.
File pointers are a higher level more abstract and user friendly system call. It remembers until exactly where you read/wrote on the file and continues where it left off. It uses buffers and is usually more efficient.
> Another thing is that with the principle of referential locality, is it just the idea that files typically execute sequentially? Because in the final workshop he mentions that its 2 things but doesn't really mention another thing.
Principle of referential locality means that memory references (processes in RAM we currently use) usually cluster around each other and there is rarely a jump where we go from A to Z then D, it goes usually A B C D and so on (unless there are function calls). Imagine a list where you iterate from first to last index, if you're at index 2, you would usually use index 3 then 4 and so on. With this, it means that we don't need to load ALL the list, maybe just the first 10 so that we save space for other processes to take place in the RAM. The real challenge here is deciding how much do we constrain this, too much then we won't have as much space for other processes. Too little and page thrasing happens (when the thing you used earlier and will need again keeps getting swapped in and out over and over again).
I'm not too sure about file descriptors and pointers and Joshua and Amitava would know better than me but I hope this helps!
