Kurt Pfeifle k1pfeifle at gmx.net
Wed Jan 4 19:58:20 GMT 2006

On Wednesday 04 January 2006 18:33, Alexander Neundorf wrote:
> On Wednesday 04 January 2006 19:21, you wrote:
> > On Wednesday 04 January 2006 19:12, Alexander Neundorf wrote:
> > > -  KURLRequesterDlg dlg( QString(), this, 0, true);
> > > +   KURLRequesterDlg dlg( QString::null, this, 0, true);
> >
> > It's starting to look like this is a gcc-3.2.x bug more than a "normal" C++
> > parsing problem, despite my explanation of that possible parsing problem.
> > Your gcc surely hits this much more than gcc-3.4 or gcc-4.
> >
> > No idea if gcc-2.95 and gcc-3.3 have the problem too; maybe someone should
> > try before you convert all the code back, indeed. If only 3.2.x has the 
> > problem, then we could simply forbid that compiler...
> Which would suck...
> I don't really intend to update my distribution in the next time. And I also 
> don't plan to update my compiler, or is gcc 3.3.x or 3.4.x ABI compatible to 
> gcc 3.2.3 ?

From painful experiences with klik packages, I can say that it is not (well, it
may be the case that 3.3.x and 3.2.3 are compatible). It is even official, there 
are 3 versions for GCC/C++ ABI on Linux:

 * one for GCC 2.95 ABI
 * one for GCC 3.2/3.3 ABI
 * one for GCC 3.4/4.0 ABI

In theory, the latter two should be identical (they both try to implement the
Itanium C++ ABI specification (go to http://www.codesourcery.com/cxx-abi/ for
a full documentation).

In practice, the 3.2/3.3  and  3.4/4.0 are so much different just through the
introduced "bugfixes" between the two 2 groups that mixing binaries/libraries 
created by these the two compiler generations quite frequently leads to some
"unexplainable" crashes. (I'm told, that in theory it should be possible to 
"indirectly" link C++ binaries from both ABIs, via a C libarary, but in practice 
this is prevented from an apparent GCC bug).

> This is a Slackware 9.1, I don't know which other distributions also ship with 
> gcc 3.2.x.
> Alex


More information about the kde-core-devel mailing list