On Wed, 12 Jan 2005, Marilyn Davis wrote:

Hi Marilyn,

Oh! Can you explain why you're mixing the low-level 'os.write()' and 'os.close()' stuff with the high-level file methods?

The 'os' functions work at a different level of abstraction than the file object methods, so there's no guarantee that:

os.close(fp)

will do the proper flushing of the file object's internal character buffers.

Try this instead:

### fobj = open('/tmp/xxx','w') fobj.write('ooo\n') fobj.write('x\n') fobj.close() ###

The documentation on os.write() says:

"""Note: This function is intended for low-level I/O and must be applied to a file descriptor as returned by open() or pipe(). To write a ``file object'' returned by the built-in function open() or by popen() or fdopen(), or sys.stdout or sys.stderr, use its write() method."""

(http://www.python.org/doc/lib/os-fd-ops.html#l2h-1555)

I think the documentation is trying to say: "don't mix high-level and low-level IO".

For most purposes, we can usually avoid using the low-level IO functions os.open() and os.write(). If we're using the low-level file functions because of pipes, then we can actually turn pipes into file-like objects by using os.fdopen(). os.fdopen() is a bridge that transforms file descriptors into file-like objects. See:

http://www.python.org/doc/lib/os-newstreams.html

for more information on os.fdopen().

I hope this helps!