Seamonkey core dumps at start in nsACString_internal::~nsACString_internal

[Marius Strobl]

On Fri, Jan 18, 2008 at 06:28:06PM -0500, Joe Marcus Clarke wrote:
> 
> On Thu, 2008-01-17 at 22:36 +0100, Marius Strobl wrote:
> > On Tue, Jan 01, 2008 at 02:45:56PM -0800, Stephen Hurd wrote:
> > > The current Seamonkey port is core dumping for me in 
> > > nsACString_internal::~nsACString_internal
> > > 
> > > My options were:
> > > # This file is auto-generated by 'make config'.
> > > # No user-servicable parts inside!
> > > # Options for seamonkey-1.1.7
> > > _OPTIONS_READ=seamonkey-1.1.7
> > > WITH_MAILNEWS=true
> > > WITH_COMPOSER=true
> > > WITHOUT_LDAP=true
> > > WITH_CHATZILLA=true
> > > WITH_JAVASCRIPT_DEBUGGER=true
> > > WITHOUT_SMB=true
> > > WITHOUT_DEBUG=true
> > > WITHOUT_LOGGING=true
> > > WITH_OPTIMIZED_CFLAGS=true
> > > 
> > > Anyone have a set of options that does work?  I expect that 
> > > WITHOUT_DEBUG and WITH_OPTIMIZED_CFLAGS at the very least should be frobbed.
> > 
> > This seems to be due to a miscompilation or mislinkage problem
> > causing two nsObsoleteAString::sCanonicalVTable and two
> > nsObsoleteACString::sCanonicalVTable symbols shadowing each-other
> > respectively. To me this smells a a least bit like "you get what
> > you asked for" though; I'm surprised their string code actually
> > works somewhere. See the attachement for a patch that works around
> > the problem by omitting the code in question.
> > Dear gnome@ maintainers, could you please commit the attached
> > patch and bump the PORTREVISIONs of the affected ports (the
> > problem also affects at least www/firefox) or approve committing
> > such a change?
> 
> Approved!

Thanks!

> Does this also apply to other Gecko ports, or just seamonkey?

It's known to affect at least firefox and seamonkey, though the
other xpcom consumers are also most likely affected.

Marius