Re[4]: [Gc] GC_get_stack_base() fallure on Mac OS X 10.6.7

Ivan Maidanski ivmai at mail.ru
Sun May 22 06:20:38 PDT 2011


Hi,

I don't understand. If the client (guile) creates several threads then it should define GC_THREADS on gc.h inclusion (so, if libgc is built as single-threaded, you will get a linkage error (GC_pthread_create or GC_register_my_thread is undefined)). Otherwise (if the client does not manipulate threads), single-threaded libgc should fit the client.

Where is the problem? And, why is it Darwin-specific?

Regards.

Sun, 22 May 2011 15:12:25 +0200 Hans Aberg <haberg-1 at telia.com>:
> On 22 May 2011, at 15:00, Ivan Maidanski wrote:
> 
> > In fact, single-threaded Darwin target supports GC_get_stack_base too (I
> added it several months ago).
> > 
> > Sat, 21 May 2011 19:17:03 +0200 Hans Aberg <haberg-1 at telia.com>:
> >> On 21 May 2011, at 16:49, Ivan Maidanski wrote:
> >> 
> >>> Get the recent GC snapshot from CVS. (the collector should be built with
> >> multi-threaded support otherwise GC_get_stack_base is unimplemented).
> >> 
> >> It worked, using ./configure --enable-threads=posix. It did not work on the
> >> latest package.
> >> 
> >> I think that under Mac OS X, threads should be included by default.
> > 
> > May be but I'm not sure whether it is worth to change the default behavior
> of ./configure.
> 
> Not having it as default just means that people will overlook it, like when
> used together with guile.
> 
> Hans



More information about the Gc mailing list