This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: new concept checks and the 3.0 ABI
On Fri, Apr 06, 2001 at 09:07:07AM -0700, Joe Buck wrote:
> > | What I think Joe is asking about is the libstdc++ ABI; in other words,
> > | should people be able to expect that STL-using code compiled with 3.0
> > | will link with code compiled with 3.1?
> ...
> Without such a requirement, claiming that the C++ ABI is stable is a joke,
> and we should just drop the claim.
I spoke on the phone with Mark last year about this. The fact is, the
library is so far from being ready for an ABI freeze that any thought of
such occurring for 3.0 has never been more than an opium dream.
But that doesn't make the 3.0 compiler ABI a joke. It gives tool makers,
including those working on Gdb as well as Purify, a stable target. That's
no small thing. They don't much care what's in the library, but they
absolutely need to know what the compiler produces to code against it.
Furthermore, the upgrade from the many-years-unmaintained lib"std"c++
will expose the new library to orders of magnitude more users, and evoke
many times more contributory effort, than ever before. This is the only
thing that can make it possible to produce and stabilize a library
defensible as the basis for a library ABI.
Nathan Myers
ncm at cantrip dot org