Boost logo

Boost :

From: Julien Blanc (julien.blanc_at_[hidden])
Date: 2019-11-25 05:19:32


Le dimanche 24 novembre 2019 à 07:39 -0800, Vinnie Falco a écrit :

> * QtJsonValue is part of a huge application framework. This alone is
> enough to make it unsuitable as a vocabulary type: No one is writing
> libraries that depend on Qt (imagine a Boost library being proposed
> that required Qt).

While the last part of the sentence is true in the sense that it is
really unlikely to happen (and thus we need a boost::jsonvalue), the
first is obviously completely wrong. See
https://github.com/fargies/qjsonrpc as a counter example.

> Counterexample:
>
> <
> https://github.com/ripple/rippled/blob/232975bfdbde12a65499130d78f938f261c989ba/src/ripple/rpc/handlers/AccountInfo.cpp#L101
> >

Seems i did not express myself correctly. What i wanted to say is that
i don't want to use a json type in the core part (the “business logic”)
of a program. Obviously i'll use one in any serialization adapter / rpc
part (which is the case in your example) if needed.

Although i still hope that a solution giving the same level of
abstraction than .net core, where you expose standard types, and the
framework takes care of all the serialization / deserialization part,
is doable in pure c++ (made some experiments there back in time, was
fun but very hard to maintain and very limited in functionalities).

Regards,

Julien


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk