Re: C++ Memory Management Innovation: GC Allocator

"xu****wei" <xu****weizh@[EMAIL PROTECTED]
> wrote in message
news:eb0c46b9-e605-451a-9479-192224c11cb2@[EMAIL PROTECTED]
> "Chris Thomasson" <cris...@[EMAIL PROTECTED]
> wrote:
>> Here is a very crude dynamic region allocator I did for C++:
>>
>>
http://groups.google.com/group/comp.lang.c++/browse_frm/thread/68bdc7...
>
> I have a suggestion about your region allocator: Use allocator
> instances instead of global new/delete procedures.

Yeah. I just did that for simplicity within the tiny test code. However it
would be simple to do something like:

typedef region_allocator<65536> thread_allocator;

void* thread_entry(void* state) {
  thread_allocator this_malloc(1, 3);
  pthread_setspecific(..., &this_malloc);
      [...];
  return 0;
}

Now any thread-local abstraction will only use this single allocator
instance (e.g. scope_alloc, whatever) via. a call to
'pthread_getspecific()'.

void* foo_alloc(std::size_t sz) {
  return ((thread_allocator*)pthread_getspecific(...))->allocate(sz);
}

void foo_free(void* p) {
  ((thread_allocator*)pthread_getspecific(...))->deallocate(p);
}


> I don't like to override the DEFAULT operator new/delete. This makes
> your code doesn't work well together 3rdparty code (if it also
> override them).

Agreed.

[...]


-- 
      [ See http://www.gotw.ca/resources/clcm.htm
for info about ]
      [ comp.lang.c++.moderated.    First time posters: Do this! ]