Re: [JSON][Review] Another Boost.JSON review

From: Vinnie Falco (vinnie.falco_at_[hidden])
Date: 2020-09-22 04:19:08

On Mon, Sep 21, 2020 at 8:47 PM Hadriel Kaplan via Boost
<boost_at_[hidden]> wrote:
> I vote to ACCEPT Boost.JSON into boost.

Thank you for your time and thoughtful review.

> Having yet another custom/purpose-built string implementation
> was surprising. It’s not an issue, but just surprising that yet another
> string type is being created.

I feel you here. What we want is a string type that has a very small
sizeof(), and the same sizeof() across implementations of the standard
library. And we want it to use the memory_resource / storage_ptr
system that Boost.JSON offers. This of course rules out std::string.
It is also nice to have a string type whose API is the same no matter
what version of C++ is used.

> It was also a bit surprising there wasn’t a function to pretty-print the
> serialized output (e..g, via an option arg for `serialize()`).

The current serializer prioritizes speed over anything else, but as
the library matures we will of course see other options.

> Boost.JSON has better performance and memory usage than some of the ones I know of

And this is an area of ongoing research and improvement - it will only
get better!

> unusual is frequently bad for interoperability.

Million dollar quote right here lol

>  Gcc doesn’t have pmr until version 9+, which is a shame
> ...
> I think being usable as standalone would increase the uptake
> for Boost.JSON… but requiring gcc 9+ negates that somewhat

Julien submitted a patch which is in develop and will get merged to master:

<https://github.com/CPPAlliance/json/commit/78c2bd301278bd9590b627451abcffe08b9ae67a>

This makes Boost.JSON use std::experimental::pmr::memory_resource on gcc 8.3.

Regards