In article <[EMAIL PROTECTED]>, Mikael Djurfeldt <[EMAIL PROTECTED]> wrote: >> I believe we would never have sufficient manpower to compete with it >> (and it seems that most other language runtime implementors arrived to >> the same conclusion). > >Yet, as long as the current GC is more efficient (as measured by >performance tests), there is no reason to switch, right?
There is a reason. It is actually quite tricky to get the machinery of SMOB mark functions to work exactly right in all cases. I have many late-night debugging sessions that can attest to this. BGC will make it easier to write GUILE extensions, because memory managements will be less of a headache: - no need for mark() functions - no need for scm_remember_upto_here() calls - no more tricky interactions between half-finished SMOBs and GC mark functions - no need for scm_gc_protect_object() _______________________________________________ Guile-devel mailing list Guile-devel@gnu.org http://lists.gnu.org/mailman/listinfo/guile-devel
