Re: [boost] Review Request: impl_ptr (pimpl)

Subject: Re: [boost] Review Request: impl_ptr (pimpl)
From: Vladimir Batov (Vladimir.Batov_at_[hidden])
Date: 2017-09-09 07:52:15

On 2017-09-09 17:12, Richard Hodges via Boost wrote:
> Hi Vladimir, great to see that good has come of the review process.

Well, to clarify -- there was no Boost review... There was some initial 
(and very useful) input to my submission request but Boost-wise the 
submission did not get anywhere... as the review mgr could not be 
found... so far anyway. Still, something good indeed has come of as 
after Giel's effort impl_ptr is a serious smart pointer for the purpose.

> If you'll allow my first commit to add a CMakeLists.txt file I'll 
> happily
> just start using it and contributing where possible.

All contributions are most and truly welcome. For starters I'd suggest 
you fork the project first into your GitHub userspace. Apply your 
changes. Then, submit a pull request, Giel or I will be happy to merge 
it in. I might not be able to do that for a few more days (I am away). 
Looking forward to your input.

> On 8 September 2017 at 22:53, Vladimir Batov wrote:
>> On 2017-09-09 01:34, Richard Hodges via Boost wrote:
>> 
>>> How is the review process moving on? I have a clear need for this in 
>>> my
>>> code right now.
>>> 
>>> Would love to see it get protection from obscurity by being accepted 
>>> into
>>> boost sometime soon.
>> 
>> Thank you for your interest and encouragement. Much appreciated.
>> 
>> RE: review
>> 
>> As for the actual review, then (unfortunately) there is no one. From 
>> my
>> understanding the Boost review process has changed and now a 
>> submission is
>> only scheduled for a review IFF it gets a review manager. It is not a
>> position people queue for. :-) So, no one has come forward for 
>> impl_ptr to
>> be a review manager... as I can see for other submissions also... 
>> Before
>> such a manager-less submission would be put in the queue and stay on 
>> the
>> radar... Now such a submission generates initial interest on the list, 
>> then
>> drifts out of the scope and is left behind/forgotten. It's 
>> unfortunate.
>> 
>> RE: obscurity
>> 
>> You might consider going to https://github.com/yet-another-user/pimpl 
>> and
>> adding a star to the project. It raises its visibility in a GitHub 
>> search
>> with everything following.
>> 
>> RE: accepted into boost
>> 
>> Initially I personally had my doubts if it was not too simple, obvious 
>> and
>> basic. Now Giel van Schijndel joined in and made immense
>> contributions/improvements to all policies. Namely, 'unique' and 
>> 'copied'
>> policies are std::unique_ptr-based and pimpl-objects are of the 
>> 'void*'
>> size... no memory overhead!.. Hugely useful IMO. Then, for high 
>> performance
>> two in-place (no dyn. mem. allocation) policies are really well-done 
>> with
>> one such policy not having any mem. overhead at all. So, IMO the 
>> submission
>> has certainly something to offer functionality-wise and 
>> deployment-wise
>> beyond manual pimpl-idiom implementation and would be a useful 
>> addition to
>> the existing set of smart pointers...