On Mon, 2006-01-02 at 17:52 -0500, Stefan Teleman wrote:
> On 1/2/06, Erast Benson <[EMAIL PROTECTED]> wrote:
> > > Latest issue of Linux Magazine discusses gcc-4.x vs gcc-3.x. the
> > > results of the comparison are not spectacular, or much different than
> > > gcc 3.x.
> >
> > That was C benchmarks, AFAIR
> 
> and what makes you think C++ is going to be any better ?
> traditionally, C++ is always worse than C.

comparison, comparison wise ... :-)
gcc-3.4 and gcc-4.x C delta != C++ delta

> > > 1. Try compling QT with GCC (either version above 3.x.x) and then try
> > > compiling and linking KDE against that build of QT, with SunStudio,
> > > and let us know how the linking goes. :-)
> >
> > GNU ld had a bug related to wrappers which abuse weak symbols like
> > pthreads. Bug is fixed by Nexenta team and we now building/linking KDE
> > without any linking errors. btw, could you post linking error you've got
> > to verify that this is the same problem?
> 
> this has nothing to do with weak symbols in pthreads. this has
> everything to do with C++ mangling first and foremost, C++ virtual
> table implementation secondarily and C++ template instantiation
> lastly. these are inherently incompatible between SunStudio and GCC,
> and the C++ Standard has not standardized the C++ ABI. i am *very*
> skeptical that you have compiled QT with GCC and KDE with SunStudio
> and it actually linked.

of cause not...
still, when you have a chance, could you please pass linking error here
to discuss/verify?

> > > 2. Try building libmad, libxine or liba52dec (which are written in C)
> > > with SunStudio, and let us know the results.
> >
> > what should I expect?
> 
> expect it to
> 1. not compile
> 2. the code which compiles, won't work, because the code depends on
> some packing and alignment peculiarities of GCC.

sure, some software uses gcc-izms, so that's a known fact. How is this
related to gcc vs. studio performance problem?

_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to