Jeffrey Hellrung escribió:
Ion Gaztañaga wrote:
...new versions of libraries Boost.Move and Boost.Container.
Few comments on proposed Boost.Move:
Thanks for the comments!
- Make all metafunctions true MPL metafunctions (as mentioned in previous postings).
Ok, but I would like to avoid any header dependency with MPL so that library is minimal.
- BOOST_COPYABLE_AND_MOVABLE definition scopes relative to the global scope, while other macros seems to scope relative to the local scope (e.g., the difference between ::boost::rv<TYPE> and boost::rv<TYPE>). I'm a stickler for consistency, and would probably consider ::boost::rv<TYPE> et al to be safer.
Yes, thanks.
- What is the rationale for placing BOOST_COPYABLE_AND_MOVABLE and BOOST_MOVABLE_BUT_NOT_COPYABLE in the private section of a class definition? IIRC, BOOST_ENABLE_MOVE_EMULATION was suppose to go in the public section, and I'm generally wondering what the motivations are for specifying public or private.
I just wanted to maintain public part as clean as possible, but there is no strong argument for that.
- Even though this can be gleaned from the code, I'd like to see a "How it works" section, e.g., how exactly does the library coerce rvalues into binding to the BOOST_RV_REF overload of operator= rather than the BOOST_COPY_ASSIGN_REF overload? Perhaps more generally, I think some rationale for why this particular emulation strategy was chosen over others would be nice (e.g., the previous proposed Boost.Move used a by-value operator= to assign from rvalues efficiently; why the change? How does this compare to Adobe's move emulation?). In most other respects, I think the documentation is very good and complete.
Ok, I'll add this.
- I would've renamed BOOST_COPY_REF_N_TEMPL_ARGS to BOOST_COPY_ASSIGN_REF_N_TEMPL_ARGS just to be consistent with BOOST_RV_REF and BOOST_RV_REF_N_TEMPL_ARGS.
Yes, thanks.
- Can you comment at all on the stability of the names of the "class declaration" macros (e.g., BOOST_MOVABLE_BUT_NOT_COPYABLE)? I recall a previous post that suggested alternative names. I have no opinion either way, I was just curious on the status of that suggestion.
I don't have any name preference, this can be discussed in the review so that we find a suitable name.
- In the clone_ptr example, would it be more correct to check for self-assignment in the move assignment operator (as is done in the copy assignment operator) rather than not?
Ok.
- Overall, great job. If there's anything I can do to help the transition, let me know.
- Jeff
Thanks! Ion