[boost] Re: [review] Formal review of the Smart Container library

From: Thorsten Ottosen (nesotto_at_[hidden])
Date: 2004-09-27 14:26:34

"Pavel Vozenilek" <pavel_vozenilek_at_[hidden]> wrote in message
news:cj9ndu$8s3$1_at_sea.gmane.org...
|
| "Thorsten Ottosen" wrote:
|
| > | It may be also useful to provide few more
| > | clone managers in library:
| > |
| > | - possibly the one that is able to detect dynamic type
| > |    at runtime and properly contruct clone, w/o need
| > |    for special clone support in class,
| >
| > hm... how would that work?
| >
| It is possible. Exception handlers store relation between types
| and such handlers could be generated by metaprogramming.
| The polymorphic_map uses this trick.

Ok, I can evaluate it and make it part of a post-review once somebody
submit the code.

| > | - one using non-default allocator and then
| > |    placement new
| >
| > can't you do that from within
| >
| > T* allocate_clone( const T* ptr )
| > { return my_clone_func( ptr ); }
| >
| > ?
| >
| Yes but the clong would be separated into few pieces
| instead one.

ok, code please :-)

| > | - one using serialization/deserialization to create
| > |   clone (the serialization process may for example
| > |   fluch internal caches or re-adjust some internal
| > |   structures)
| >
| > hm...the thing is that the clone manager controls all cloning, also when
| > clones are inserted coming from other containers...seems wierd to me.
| >
| It is weird but should work.
| I use this trick somewhere, so I put it as option.

Ok, now I get it. It's like in Java where you can make a fake clone by
serializing the class.

br

Thorsten