Mitchell N Charity <[EMAIL PROTECTED]> wrote: > Perhaps it is time to get "multiple gc regimes can coexist" working?
Sounds good, but AFAIK doesn't work - or isn't practical. I can only imagine to have some #defines in place, to switch/test different schemes, as currently LEA allocator. > In addition to complexity cost, downsides include an extra test/indirection > so multiple PMC layouts can coexist. That's the problem. When you touch each PMC during a DOD run, you defeat e.g. my current approach with separated DOD flags, where PMC memory is not touched. > Mitchell leo
