Stefan Teleman <Stefan.Teleman at Sun.COM> writes: > Ian Lance Taylor wrote: >> Stefan Teleman <Stefan.Teleman at sun.com> writes: >> >>> That is not the case for C++. GCC's C++ ABI is not stable, and it >>> changes in an incompatible way, even between Minor Releases. >> >> That has not been the case for a while now. The C++ ABI generated by >> GCC has been stable since the GCC 3.4 release four years ago. > > False. > > The GCC C++ ABI has changed in an incompatible way several times > within the the 3.x Minor versions, and has changed in an incompatible > way yet again, for the GCC 4.x Major Release. > > A simple Google search for 'GCC C++ ABI' would yield relevant results: > > http://books.google.com/books?id=wQ6r3UTivJgC&pg=PA47&lpg=PA47&dq=GCC+C%2B%2B+ABI&source=web&ots=EJUnUr4CDt&sig=NibMWd8j1y9cs6_zSdGY_A0OXqA&hl=en&sa=X&oi=book_result&resnum=8&ct=result > > http://lists.alioth.debian.org/pipermail/glibc-bsd-devel/2005-July/000446.html > > These are just two examples. There doesn't even seem to be a consensus > as to which GCC versions introduce ABI incompatible changes.
I'm a GCC maintainer, I'm not some random guy off the street. The book is incorrect. The notes on the glibc-bsd-devel mailing list appears to be consistent with what I said: the C++ ABI has not changed since version 3.4. Ian