Yes, that is solid reasoning.  I do wish the man pages for fclose(),
   flush(), and close() were not so confusing.  fflush accepts NULL,
   if the fp is invalid it shoudl return EOF with errno set to EBADF,
   etc.. Nothing about dumping core!

So if the pointer is 0xdeadbeef, it should nicely return EBADF rather
than core dumping, too?

Just how aggressive do you think the C library should be about
validating what you tell it?


------------------------------------------------------------------------------