> c++.c:35: error: aggregate ?s1_h h? has incomplete type and cannot be
ok, so it appears to be named structures inside structures, and then
using the inner structure as a separate variable. From your test it
looks like nested structures themselves are OK, but C++ doesn't allow
you to 'see' the inner structure except as part of the structure it
was defined in.
If volker applied the thumb screws hard enough and we had a real use
for C++ in mind then I'd probably relent on that one, especially if I
was sure that it was the last one we'd meet in the C++ saga. I do
wonder if it really is the last one though ... I get this nagging
feeling that there would be more C++ surprises to come!
I think the place where it will be most annoying is