High-level file routines such as those already described are usually
convenient and easy to use. However, low-level file routines such as
the ones in this section have some advantages. For example, they do not
treat all file input/output as text streams, as the high-level routines
do; for that reason, working with binary files may be easier using
low-level routines. For another thing, low-level routines do not buffer
their input and output, so you will never need to remember to flush your
streams with fflush or similar functions, as you sometimes must
with high-level routines.
Unfortunately, because low-level routines work at a lower level of
abstraction, they can be tricky, even dangerous to use -- that is to
say, if used incorrectly, they may corrupt your data or cause your
program to terminate unexpectedly; never fear, they will not explode
your monitor or cause your computer to become sapient and attempt world
domination.
As mentioned, low-level file routines do not use text streams; instead,
the connection they open to your file is an integer called a file
descriptor. You pass the file descriptor that designates your file to
most low-level file routines, just as you pass the stream that
designates your file to most high-level file routines. For example,
while the low-level open function takes a filename string to open a file, the matched
close function takes the file descriptor returned by open:
my_file_descriptor = open ("foo_file", O_RDONLY);
close_err = close (my_file_descriptor);
