> Richard Oudkerk added the comment:
>
> I can't remember why I did not use fstat() -- probably it did not occur to me.

I probably have Alzeihmer, I was sure I heard a reasonable case for
dup() vs fstat().
The only thing I can think of is that fstat() can incur I/O, whereas
dup() shouldn't.

In any case, I too prefer fstat(), since it doesn't require creating a
new FD (which could fail with EMFILE/ENFILE), and is more logical.

The only comment I have about this patch is I think that the helper
function to detect the number of open FD might have its place in
test.support.