Support for C++20 <ranges> library

[evilenzo]

I met a problem that I need to iterate through reversed json array. What I'd like to use in C++20 is

for (auto i: my_json["some_array"] | std::views::reverse) {
    // do something
} 

Is it such a problem to implement this kind of feature?

[nlohmann]

In order to asses your issue, we need the following information:

• What is the issue you have?

• Please describe the steps to reproduce the issue. Can you provide a small but working code example?

• What is the expected behavior?

• And what is the actual behavior instead?

• Which compiler and operating system are you using? Is it a supported compiler?

• Did you use a released version of the library or the version from the develop branch?

• If you experience a compilation error: can you compile and run the unit tests?

[gregmarr]

Is it such a problem to implement this kind of feature?

No, this works just fine, with std::views::reverse instead of std::views::reversed, at least with GCC trunk.

Compiler Explorer

[falbrechtskirchinger]

To my knowledge no release version works because of the iterator issue I fixed last month. I had trouble using std::ranges algorithms before.

Edit: Just checked with 3.10.5 and develop. The former fails to compile with an undecipherable error message.

[gregmarr]

Ah, I was using trunk.

[evilenzo]

Of course, I was talking about std::views::reverse, my mistake

[falbrechtskirchinger]

Try using an unreleased version after commit 700b95f. That should work.

Edit: @nlohmann You might want to mention that specifically in the 3.10.6 release notes because almost no one will read "Make iterator operator++/--(int) equality-preserving" and think that that means support for std::ranges. Also I should contribute some unit tests to be sure most things really work.

[nlohmann]

There is also #3130 - it would be great to have this fixed together.

[falbrechtskirchinger]

There is also #3130 - it would be great to have this fixed together.

I imagine it is but I haven't checked yet. Might do so later.

[gregmarr]

I imagine it is but I haven't checked yet. Might do so later.

Unfortunately, still broken.

source>:13:24: error: no match for 'operator|' (operand types are 'nlohmann::basic_json<>::iteration_proxy<nlohmann::detail::iter_impl<nlohmann::basic_json<> > >' {aka 'nlohmann::detail::iteration_proxy<nlohmann::detail::iter_impl<nlohmann::basic_json<> > >'} and 'std::ranges::views::__adaptor::_Partial<std::ranges::views::_Transform, main()::<lambda(auto:23&&)> >')
13 | auto t = j.items() | std::ranges::views::transform([](auto&& _v) { return _v.key(); })
| ~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| | |
| | std::ranges::views::__adaptor::_Partial<std::ranges::views::_Transform, main()::<lambda(auto:23&&)> >
| nlohmann::basic_json<>::iteration_proxy<nlohmann::detail::iter_impl<nlohmann::basic_json<> > > {aka nlohmann::detail::iteration_proxy<nlohmann::detail::iter_impl<nlohmann::basic_json<> > >}

[falbrechtskirchinger]

Yeah, I forgot that there are other outstanding issues in the iterator implementation. Too bad.

[falbrechtskirchinger]

I imagine it is but I haven't checked yet. Might do so later.

Unfortunately, still broken.

source>:13:24: error: no match for 'operator|' (operand types are 'nlohmann::basic_json<>::iteration_proxy<nlohmann::detail::iter_impl<nlohmann::basic_json<> > >' {aka 'nlohmann::detail::iteration_proxy<nlohmann::detail::iter_impl<nlohmann::basic_json<> > >'} and 'std::ranges::views::__adaptor::_Partial<std::ranges::views::_Transform, main()::<lambda(auto:23&&)> >')
13 | auto t = j.items() | std::ranges::views::transform([](auto&& _v) { return _v.key(); })
| ~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| | |
| | std::ranges::views::__adaptor::_Partial<std::ranges::views::_Transform, main()::<lambda(auto:23&&)> >
| nlohmann::basic_json<>::iteration_proxy<nlohmann::detail::iter_impl<nlohmann::basic_json<> > > {aka nlohmann::detail::iteration_proxy<nlohmann::detail::iter_impl<nlohmann::basic_json<> > >}

Good news! Thanks to @gregmarr's prior work here and a few changes and additions of my own, I'm able to successfully compile and run this line of code now. PR soon.

I only wish I had seen @davidstone's comment in that thread before now. It took me forever to figure out why std::indirectly_readable isn't satisfied.