Thread: Reading From a text file

Originally Posted by anduril462

Also, your fclose will cause a segfault if you can't open the file (since myfile will be NULL).

No. Passing a valid pointer whose value is NULL is different than passing NULL. For cygwin, neither one actually crash:

Code:

#include<stdio.h>
int main( void )
{
    FILE *fp = NULL;
    int x = fclose( fp );
    
    printf( "x is %d\nready to crash? Hit enter.\n", x );
    getchar();
    
    fclose( NULL );
    printf( "If you see this, there was no segfault.\n" );
    return 0;
}

Quzah.

I find it intersting Cygwin doesn't segfault on either one (since Cygwin is supposed to provide Linux-like functionality) while it seg faulted on both the first and second call on my CentOS machine.

I just spent a ton of time reading the standard and formulating well thought out response to your comment, only to realize I was the <expletive deleted> who used NULL instead of null (perhaps a little less wine before posting next time). I did learn the following about fopen and fclose along the way, however...

According to section 7.19.5.3, paragraph 8 of the standard:

The fopen function returns a pointer to the object controlling the stream. If the open operation fails, fopen returns a null pointer.

It may or may not be NULL, but it's guaranteed to be a null pointer, and will be of type FILE *.

As for fclose, section 7.19.5.1, paragraph 2 says:

A successful call to the fclose function causes the stream pointed to by stream to be flushed and the associated file to be closed. Any unwritten buffered data for the stream are delivered to the host environment to be written to the file; any unread buffered data are discarded. Whether or not the call succeeds, the stream is disassociated from the file and any buffer set by the setbuf or setvbuf function is disassociated from the stream (and deallocated if it was automatically allocated).

So, it seems that the handling of a null FILE * would be undefined. I guess the null pointer passed in to fclose wont necessarily cause a segfault, but it's certainly pointless to fclose a null file pointer (since it means the file was never opened in the first place), and since it may seg fault in some implementations, it's best to protect against it.

I wasn't disagreeing with it being a bad idea to add the necessary checks - just the part where you said it would segfault, since the standard wouldn't define something like that.

Quzah.