[pimpl] Proposal. Determining interest.

I wrote a Pimple generalization thin interface layer on top of boost::shared_ptr or boost::impl_ptr (uploaded in the Boost Vault as Pimpl-01.zip). Based on previous discussions I extended my original suggestion to support pointer and value semantics. The code is very simple but offers the following advantages: 1. First is obviously implementation hiding. The reason Pimpl come to being in the first place. 2. Pimpl-based classes only define pure application interface without any usual scaffolding. 3. Memory management is fully encapsulated (delete and new) to minimize abuse/mis-use. Currently I've been using and testing it only in pointer-semantics mode. If it is determined that there is sufficient interest in the suggestion, we'll need to bring in Peter Dimov's impl_ptr and extend it a bit (that's my current impression.Peter might prove me wrong) Thanks, Vladimir.
participants (1)
-
Vladimir.Batov@wrsa.com.au