The biggest problem currently seems to be string_to_cstring. Some time
ago, I made a change, that the returned cstr has the
BUFFER_immobile_FLAG set, which should be the Right Thing(tm), *but* -
though immobile strings don't move - the memory_pool, where these
strings live, get freed at the end of compact_pool.
So the best proabably would be, to allocate system memory for such strings.
leo
