Gavin, Thank you for the input. On 2017-08-21 18:08, Gavin Lambert via Boost wrote:
On 15/08/2017 12:51, Vladimir Batov wrote: ... On a quick pass through the docs, http://yet-another-user.github.io/pimpl/doc/html/impl_ptr/behind_the_interfa... suggests that Book::operator== be implemented using (*this)->title etc. ... I would recommend not overloading operator-> and operator* for impl access and instead requiring that this be more explicit, such as impl().title (or as the page also suggests impl.title, but after a glance at the implementation I didn't see how that would work).
Yes, I hear you. I've been using the syntax for quite some time and I am used to it. Certainly, there is nothing wrong in additionally providing impl() or something of that sort.
From http://yet-another-user.github.io/pimpl/doc/html/impl_ptr/construction_and_i..., perhaps consider implicit construction from nullptr_t as another method to create an empty instance; that might be more natural for people.
Good point. I'll add it.
Also, the generated reference docs seem a bit useless -- http://yet-another-user.github.io/pimpl/doc/html/reference.html shows only two types, including _internal_impl_ptr, which is not itself documented and just 404s.
Indeed. Just added it today and still fighting Doxygen... it refuses to document the "detail" namespace... no matter what I try.