[Bug c/58531] [4.7/4.8 Regression] Strange array element ordering with O1 flag

[guillaume at eulerian dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58531

--- Comment #4 from Guillaume <guillaume at eulerian dot com> ---
Ok i understand.

I took a sad brain shortcut assuming a single block declaration was generating
an (only possible) increasing address for each compound.
This was working on all previous gcc version i used (as far as i remember).
That's why it is worrying me.
I will correct my mistake and go punish myself ;)

Thanks a lot.