libstdc++ ABI break

[Mark Mitchell]

Benjamin Kosnik wrote:
>> At a guess, you weren't intending to change the ABI there?
> 
> No, I did not. Gcc-4.2.0 is ok, so it looks like (as of right now)
> gcc-4.2.1 is the only breakage.  
> 
> UGH. 
> 
> Mark, we should probably try to fix this for 4.2.2.

This is not an easy call.  We've generally avoided any ABI change on
release branches, even "good" changes, to make sure that people could
upgrade smoothly along a release branch.  Here, we can't win; we've
already broken things between 4.2.0 and 4.2.1, and now we're talking
about breaking again between 4.2.1 and 4.2.2.

Other than this, is the 4.2.x libstdc++ actually ABI-compatible with
4.1.x?  Is there any evidence that people are actually mixing and
matching in that way?  I'm not asking if the libstdc++ ABI tests pass;
I'm asking if we actually know that people are actually mixing things.
Like, are people building Qt applications with 4.2 and linking them to
4.1 libstdc++?

The reason I ask is that if there's nobody actually mixing stuff, then
we might as well stay with the new ABI and avoid yet more breakage.  As
far as I know, there have been up until now zero reports of this change
being an actual problem.

I was planning on doing the 4.2.2 release Very Soon -- as in, I was
within minutes of starting the checklist when I got your email.  I guess
I'll hold off for a day or two in the hopes we can get this sorted.

Thanks,

-- 
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713