Re: [review][JSON] Some thoughts on recent reviews

From: Hans Dembinski (hans.dembinski_at_[hidden])
Date: 2020-09-23 09:04:27

> On 22. Sep 2020, at 16:23, Peter Dimov via Boost <boost_at_[hidden]> wrote:
> 
> But in the meantime, there are people who have actual need for a nlohmann/json (because of speed) or RapidJSON (because of interface) replacement, and we don't have it. It doesn't make much sense to me to wait until 2032 to have that in Boost.

My rough count of accept votes indicates that Boost.JSON is going to be accepted, so you get what you want, but I feel we gave up on trying to achieve the best possible technical solution for this problem out of a wrong sense of urgency (also considering the emails by Bjørn and Vinícius I does not seem like we need to wait for 2032 for a different approach).

This is C++, we strifes for "zero overhead" and "maximum efficiency for all reasonable use cases", and to achieve that requires a careful interface design. I only worry about interfaces, because implementations can be improved at any time. However if an interface is designed to requires more work than absolutely necessary, then this cannot be fixed afterwards.

You said yourself that Boost.JSON is not as efficient as it could be during the conversion of "my data type" to JSON, because the existing data has to be copied into the json::value first. I am a young member of the Boost family, but my feeling is that this would have been a reason to reject the design in the past. Designing abstractions that enable users to get maximum performance if they want it is a core value of C++.

As my previous examples of pybind11 and cereal have shown, the lasting legacies of Boost are excellent interfaces. Making good interface is very difficult and that's where the review process really shines.  We have not achieved that here, since valid concerns are pushed aside by the argument: we have to offer a solution right now.

Best regards,
Hans