/Zc:__cplusplus leads to C2416

[crusader-mike]

OS: Windows 7
Compiler: MSVC 2017 (15.9.11)

If project uses /Zc:__cplusplus json lib v3.7.0 fails to compile with this:

C2416: attribute 'deprecated' cannot be applied in this context

Looks like MSVC doesn't like this attribute applied to operator>> definition.

[jaredgrubb]

I don't see why there is any bug here. Can you explain why the deprecated attribute is not legal in that context?

[crusader-mike]

I assume it complains because [[deprecated]] attribute was applied to definition. According to C++17 draft I have (10.6.4.p2):

The attribute may be applied to the declaration of a class, a typedef-name, a variable, a non-static data member, a function, a namespace, an enumeration, an enumerator, or a template specialization.

Looks like MS read it as "can't happen on a definition". I have no opinion about correctness of this position.

[jaredgrubb]

Yeh, that's a nuance that I can't make a call on myself. Dumb fix would be just to disable C2416 for this file :)

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[crusader-mike]

Bad bot!

[t-b]

Not sure how the project rules here are with respect to selectively disabling warnings but why not propose a PR which uses pragma push/pop logic for MSVC to disable that warning for this piece of code? See https://stackoverflow.com/a/18463996 as a starting point.

[crusader-mike]

or I could simply patch this lib in my vcpkg instance and avoid dealing with PR/etc ;)

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[mrstux]

I'm seeing this too.

I tried disabling the warning via the push/pop mechanism, but it seems the compiler front-end doesn't understand that its a compiler warning, and the website doesn't document it either.

In the end I added a check for MSVC to the deprecated language checks. Oh well.

A bit hacky, but unblocking.

/*
TODO: remove _MSC_VER when this is fixed.
https:/nlohmann/json/issues/1695
*/
#if defined(__cplusplus) && (__cplusplus >= 201402L) && !defined(_MSC_VER)
    #define JSON_HEDLEY_DEPRECATED(since) JSON_HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[deprecated("Since " #since)]])
    #define JSON_HEDLEY_DEPRECATED_FOR(since, replacement) JSON_HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[deprecated("Since " #since "; use " #replacement)]])
#elif \

And regarding closing this issue because its "stale", I guess its a novel approach to solving issues, just close them after 31 days.

[nlohmann]

This should be fixed on develop branch as it contains an updated version of Hedley. Can anyone try this?

[t-b]

I've added a PR which compiles with /Zc:__cplusplus. Let's see if that works now.

[t-b]

Fails in CI as well, see https://ci.appveyor.com/project/nlohmann/json/builds/28319126/job/qp8uvagt7i56g5bp.

@mrstux The best thing here is to fix the issue in https:/nemequ/hedley, then incorporat the changes into the linked PR.

[t-b]

It's not about the definition. MSVC chokes on the friend. Using "__declspec(deprecated)" does not trigger the warning. see https://godbolt.org/z/lhPjC-.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[t-b]

@nlohmann Please reopen.

[LanderlYoung]

Is there any progress here?

As far as I know, the easy way is to disable the attribute for windows by:

#if defined(__cplusplus) && (__cplusplus >= 201402L) && !defined(_MSC_VER)

BTW: MSVC sucks on this and the __cplusplus Marco bug.

Also, once MSVC fixed it, the last condition can be checking _MSC_VER range.

[t-b]

@LanderlYoung The upstream bug in hedley was fixed.

@nlohmann How do I upgrade to a newer hedley version?

[dota17]

@t-b
I had compared the current version and the newest version of hedley.
There are not a lot of differences or special changes. So it maybe ok to update the source codes.

[nlohmann]

I upgraded Hedley to version 13 in https:/nlohmann/json/tree/feature/update_hedley - can someone check if this fixes this issue? @t-b ?

[t-b]

@nlohmann Yeep, I've rebased #1958.

[kenr]

Is the 3.7.4/3.4.0 release with this fix planned anytime soon? I use bitsery with std::variant and that requires /Zc:__cplusplus for MSVC to set the correct __cplusplus value for C++17.

[nlohmann]

The next release will be version 3.8.0. I'm not sure how much more time I will need until it is ready.

[kenr]

@nlohmann Is there anything I can help out with towards the 3.8.0 release?

[nlohmann]

Not really - but if you are using MSVC and want to help, you may want to have a look at #2117.