The libgc that is distributed with Sun Studio is a commercial product that was produced by Geodesic (which has gone out of business). We have the sources but no authorization to change or distribute them.
I suspect that an implementation of a "Boehm-Demers-Weiser" GC being maintained by Hans Boehm will not have stability or quality problems. The compilers/tools group would have no objection to this version superceding the one from Geodesic. Hugh McIntyre wrote: > Irene Huang wrote: >> >>> I am setting the timeout for this case to be Feb 13th. >>> Please review the proposal at >>> http://sac.eng/Archives/CaseLog/arc/LSARC/2008/068/proposal.txt > > I have no particular objection to this case, because there a bunch of > FOSS that wants to use it. But, assuming Indiana plans to support > installing the Studio compilers in future, would it be a good idea to > add a short note to the end of any supplied man page to note: > > 1) The Studio compilers also supply a conflicting libgc.so under > /opt/SUNWspro/lib/libgc.so. See > http://developers.sun.com/solaris/articles/libgc.html. > > 2) The Studio version does not need people to edit their programs to > replace malloc() with GC_MALLOC() and the like. But equally, someone > compiling with the Studio compilers will probably pick up the Studio > version first because of the default link path, even if they are > expecting the version from /usr/lib in this case instead. > > Maybe there needs to be some provision here to avoid conflicts, or at > least to tell people how to avoid getting such conflicts by mistake? > > Hugh.
