On Fri, Oct 24, 2008 at 04:01:51PM -0400, Stefan Teleman wrote: > One thing i am not going to do is make a far reaching architectural > decision about GCC's future C++ ABI compatibility, based on mailing list > statements, in absolute disregard for significant and documented C++ ABI > compatibility breakage, within GCC Major Version 4, and in direct > contradiction of your claimed compatibility statements.
Hmmm, no, you can say that a particular version's ABI will be Committed in Solaris, meaning that "g++" (as opposed the full versioned path to the g++ executable, such as /usr/gnu/gcc/<version>/bin/g++), will not be changed to a new version that's backwards incompatible, not on any update/patch releases. And you can say that "g++" is Volatile, that only the full versioned paths to g++ are Committed. That's the choice. Leaving potentially-unstable interfaces out of /bin is an option (and what you're proposing), but not a very good one (at least one ARC member didn't like it). > Solaris' tolerance for ABI breakage is zero. We don't do "sorry we broke > the ABI" statements. The SDF has a process by which incompatible changes can be made and by which expectations are documented about when such changes may be made. See the ARC's best practices (interface stability taxonomy, release taxonomy, EOF/EOL processes). Solaris has certainly had backwards- incompatible changes during its history, all in accordance with such processes (including appeals). Nico --
