Re: Memory corruption errors!?

JR> I did not realize that --enable-newgc is completely independent JR> from --enable-kkcc. NEWGC is not independent from KKCC. KKCC is a prerequisite of NEWGC. If you configure with --enable-newgc, configure turns on KKCC. We have a separate configure option for it because KKCC can also be used with the old allocator/garbage collector. JR> If there is an intention of keeping the old code around, perhaps JR> better descriptions should be provided in configure? (as in JR> --with-gc=vanilla for folks who want to go back to a working JR> default) The options --enable-kkcc and --enable-newgc are the only options that mess with gc-related code. NEWGC is still defaulted to off, so people have to explicitly add this option to the configure string. Thus, I assume they know what they are doing. And that KKCC is turned on by default is not a secret, see the output of configure: Using the new GC mark algorithms (KKCC). WARNING: --------------------------------------------------------- WARNING: The new algorithms are experimental. They are enabled by WARNING: default for this release. Use `--disable-kkcc' to WARNING: turn it off. WARNING: --------------------------------------------------------- You can turn off KKCC, then you will not find any gc-related output in your Installation file. This means that you use the old allocator and the old garbage collector. Although it might not be a "working" one either, my last patch might have introduced some problems with the old collector as well. I am still trying to figure this out, so gdb and/or valgrind output of your crashes can be helpful. -- Marcus