Please try using this CVS snapshot:

  http://snaps.php.net/php5.1-latest.tar.gz
 
For Windows:
 
  http://snaps.php.net/win32/php5.1-win32-latest.zip

is there a cvs for 4.4.1

No, try the 5.1 snapshot.

nope - 5.1.2.2 cvs - same story - feof() still ignores a zero file

what i don't understand is that the win/linux lower filesystem layer maintains a filepointer of which it should be easy to detect/report a EOF condition. The PHP handle as a file descriptor resource should among things maintain this.

But i don't maintain php and thus don't know the reasons.

Wez, sounds like something isn't working quite well in the streams world?

I get the same result with this C code:

#include <stdlib.h>
#include <stdio.h>

int main() {
    FILE *fd;
    fd = fopen("/tmp/xxx", "r");
    while (!feof(fd)) {
        char buf[100];
        fgets(buf, 100, fd); // remove that and you'll get endless cycle
        printf("read\n");
    }
    fclose(fd);
    return 0;
}
(it reads once and stops just after that)
Doesn't look like a bug to me.
PHP behaves in the same way as the underlying OS.

Taken from the PHP manual
--------------------------------
feof -- Tests for end-of-file on a file pointer

Description

bool feof ( resource handle )

Returns TRUE if the file pointer is at EOF or an error occurs (including socket timeout); otherwise returns FALSE. 
--------------------------------

Read a file with only 4 chars "0123". The chars correspond to what ftell() reports (0 at open) and 4 after reading 4 with fgets(5).

File-pointer is then 4 - aka the next read position which is beyond last char and thus EOF. This must be proof enough of inconsistency. Either the manual or the feof() need to be corrected - i prefer feof().

Still php is great - and i help to make it better !