------- Additional Comments From rassahah at neofonie dot de 2005-08-15 10:41 ------- (In reply to comment #3) > Hmm on powerpc-darwin we get: > a = 1, b = 3 > > > Which is still wrong. You mean before or after the application of the suggested patch? Perhaps there may be other alignment constraints on powerpc-darwin than on i686-linux. The alignment with the ROUND-macro as i see it would not work with an alignment constraint any other than `alignment == multiple of data size'. BTW: The bug is at least in gcc-2.7.2.3. I am wondering how it got along undetected, since the program with the encoding "{ii}" is an example taken from `Object Oriented Programming & the Objective-C Programming Language'. But however as i checked the various libs that are open source and compilable under linux (libFoundation etc.), nobody uses it. They all handle their serialization of compound object by themselves...
-- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23108