Re: opposition to LSB 2.0 rc1

[Gabriel Dos Reis <gdr at integrable-solutions dot net>]

Benjamin Kosnik <bkoz@redhat.com> writes:

| >There are enough other good arguments that there's no need to make dubious
| >claims of "increased conformance to C99" to justify the benefits of 3.4.  
| >You will observe that gcc-3.3/c99status.html and gcc-3.4/c99status.html
| >are identical apart from the version number: there are no new C99 features
| >in 3.4.  Various technical conformance issues are improved, but some
| >regressions in c-decl.c were postponed to 3.5 so there isn't monotonic
| >improvement either.
| 
| Okay, thanks for the clarification.
| 
| >Even 3.4 doesn't implement the standard C++ ABI fully correctly
| ><http://gcc.gnu.org/ml/gcc/2004-07/msg00138.html>.  (This is apart from
| >the "cannot be mangled due to a defect in the C++ ABI" cases, but those
| >don't cause binary compatibility problems between releases, just failures
| >to compile in the problem cases.)
| 
| Of course. That's not my point. There will always be bugs in
| implementations, that is also not my point. 
| 
| I think it undisputed that 3.4 does a better job of implementing the C++
| ABI than 3.2.x or 3.3.x compilers. 

Yes.  For example, name lookup  is improved (even if not 100% standard).
More C++ constructs are correctly parsed and analyzed.

-- Gaby