C++ structure sizes with long long ints under Darwin,

[Benjamin Kosnik]

>It includes a testcase that can be compiled.  The size of a C++ 
>structure
>with long long ints seems to be 4 bytes too big when compared to the 
>same
>structure when a double replacing the long long int and compared to 
>what is
>expected.
>
>It also does not match the results of gcc 2.95 under Apple or Apples' 
>gcc 3.1 with
>and without -malign-natural.

I would expect somebody from Apple's gcc team (Geoff K, Matt A, Mike S)
would care about this, as Apple seems to be thinking of using gcc-3.3
for their next release. You'll probably have better luck contacting them
directly, since the public posting seems to have been ignored.

-benjamin