[boost] Re: placed_ptr<> -> shifted_ptr<> + operator new (size_t, gc)

From: Philippe A. Bouchard (philippeb_at_[hidden])
Date: 2003-01-15 23:11:41

"David B. Held" <dheld_at_[hidden]> wrote in message
news:b04n9f$ik2$1_at_main.gmane.org...

> [...] I don't see any documentation, except for what few
> comments exist in the source.  It isn't clear where you expect people to
> use your pointer, or if there are usages that should be discouraged.  You
> don't mention any gotchas that might occur.  These are things that should
> appear in the docs (which you will certainly need if you intend to submit
it
> for review).  Basically, it's hard to comment on your design without
> spending a lot of time reviewing the code.  Docs would help. ;)

You're perfectly right.  I guess I was more focusing on the design before
documenting it.

> [...] I don't see the source for your test.  Why don't you include that,
so
> people can compile and run it under different compilers and platforms?

I'll clean the folder up but it was basically copied from placed_ptr/.

> [...] The comparison with shared_ptr<> isn't fair.  Your ptr has no
locking
> for thread safety.  Try defining BOOST_DISABLE_THREADS and
> re-running the tests, or you risk just measuring the speed of thread sync.

I'll work on this but the non-locking version should be okay.

Philippe A. Bouchard