Add build step for ICPC #3229

nlohmann · 2021-12-29T21:55:39Z

This PR adds a CI build step for the Intel C++ Compiler

coveralls · 2021-12-30T11:16:36Z

Coverage remained the same at 100.0% when pulling f9dfd63 on icpc into b21c345 on develop.

� Conflicts: � .github/workflows/ubuntu.yml � cmake/ci.cmake

nlohmann · 2021-12-30T13:44:39Z

ICPC fails to build the test suite.

https://github.com/nlohmann/json/runs/4665922491?check_suite_focus=true failed with

/__w/json/json/test/src/unit-regression2.cpp(810): error: more than one instance of constructor "std::filesystem::__cxx11::path::path" matches the argument list:
            function "std::filesystem::__cxx11::path::path(std::filesystem::__cxx11::path &&) noexcept"
            function "std::filesystem::__cxx11::path::path(std::filesystem::__cxx11::path::string_type &&, std::filesystem::__cxx11::path::format)"
            argument types are: (json)
          CHECK_THROWS_WITH_AS(nlohmann::detail::std_fs::path(json(1)), "[json.exception.type_error.302] type must be string, but is number", json::type_error);
          ^
/usr/include/c++/9/bits/fs_path.h(473): note: this candidate was rejected because mismatch in count of arguments
      path(basic_string_view<value_type> __str, _Type __type)
      ^
/usr/include/c++/9/bits/fs_path.h(213): note: this candidate was rejected because mismatch in count of arguments
        path(_InputIterator __first, _InputIterator __last, const locale& __loc,
        ^
/usr/include/c++/9/bits/fs_path.h(205): note: this candidate was rejected because mismatch in count of arguments
        path(_Source const& __source, const locale& __loc, format = auto_format)
        ^
/usr/include/c++/9/bits/fs_path.h(198): note: this candidate was rejected because mismatch in count of arguments
        path(_InputIterator __first, _InputIterator __last, format = auto_format)
        ^
/usr/include/c++/9/bits/fs_path.h(191): note: this candidate was rejected because at least one template argument could not be deduced
        path(_Source const& __source, format = auto_format)
        ^

Switching off std::filesystem tests with -DCMAKE_CXX_FLAGS="-DJSON_HAS_FILESYSTEM=0 -DJSON_HAS_EXPERIMENTAL_FILESYSTEM=0" fails (https://github.com/nlohmann/json/runs/4666752017?check_suite_focus=true) with

In file included from /__w/json/json/test/src/unit-algorithms.cpp(32):
/__w/json/json/single_include/nlohmann/json.hpp(3821): error: this operator is not allowed in a constant expression
  #elif JSON_HAS_FILESYSTEM
        ^

In file included from /__w/json/json/test/src/unit-algorithms.cpp(32):
/__w/json/json/single_include/nlohmann/json.hpp(4254): error: this operator is not allowed in a constant expression
  #if JSON_HAS_FILESYSTEM || JSON_HAS_EXPERIMENTAL_FILESYSTEM
      ^

In file included from /__w/json/json/test/src/unit-algorithms.cpp(32):
/__w/json/json/single_include/nlohmann/json.hpp(4507): error: this operator is not allowed in a constant expression
  #elif JSON_HAS_FILESYSTEM
        ^

In file included from /__w/json/json/test/src/unit-algorithms.cpp(32):
/__w/json/json/single_include/nlohmann/json.hpp(4887): error: this operator is not allowed in a constant expression
  #if JSON_HAS_FILESYSTEM || JSON_HAS_EXPERIMENTAL_FILESYSTEM
      ^

falbrechtskirchinger · 2022-03-07T21:43:20Z

Have a look at #3377. I had to disable the CHECK_THROWS_WITH_AS in #3379.

nlohmann · 2022-03-08T09:07:00Z

Have a look at #3377. I had to disable the CHECK_THROWS_WITH_AS in #3379.

Thanks for pointing this out. I am now confused as well. If I would just comment out the test, I'm afraid that I would ignore something not working...

falbrechtskirchinger · 2022-04-06T09:18:04Z

Adding $<$<CXX_COMPILER_ID:Intel>:-diag-disable=2196> or $<$<CXX_COMPILER_ID:Intel>:-wd2196> (deprecated) should get rid of

warning #2196: routine is both "inline" and "noinline"

if that is still an issue with current doctest.

falbrechtskirchinger · 2022-05-01T08:36:57Z

.github/workflows/ubuntu.yml

+      - name: cmake
+        run: cmake -S . -B build -DJSON_CI=On
+      - name: build
+        run: >


I'd suggest using the literal style block scalar | over the flow style one > for multiline commands:

run: | . /opt/intel/oneapi/setvars.sh cmake --build build --target ci_icpc

falbrechtskirchinger · 2022-05-01T08:39:26Z

tests/CMakeLists.txt

@@ -49,6 +49,9 @@ target_compile_options(test_main PUBLIC
    # https://github.com/nlohmann/json/issues/1114
    $<$<CXX_COMPILER_ID:MSVC>:/bigobj> $<$<BOOL:${MINGW}>:-Wa,-mbig-obj>

+    # https://github.com/nlohmann/json/pull/3229
+    $<$<CXX_COMPILER_ID:Intel>:-diag-disable=2196>


This may need a version check. It's -wd2196 for older versions but I'm not familiar with ICPC, so I don't know since when the new flag is available. 🤷‍♂️

Also, 1786 to suppress deprecation warnings.

I couldn't find a version number in the documentation.

falbrechtskirchinger · 2022-05-01T09:01:31Z

cmake/ci.cmake

+    COMMAND ${CMAKE_COMMAND}
+        -DCMAKE_BUILD_TYPE=Debug -GNinja
+        -DCMAKE_C_COMPILER=icc -DCMAKE_CXX_COMPILER=icpc
+        -DCMAKE_CXX_FLAGS="-DJSON_HAS_FILESYSTEM=0 -DJSON_HAS_EXPERIMENTAL_FILESYSTEM=0"


Appears as -DJSON_HAS_FILESYSTEM=0\ -DJSON_HAS_EXPERIMENTAL_FILESYSTEM=0 on the commandline and results in:

/__w/json/json/single_include/nlohmann/json.hpp(4116): error: this operator is not allowed in a constant expression #elif JSON_HAS_FILESYSTEM

~~Remove? Don't know why <filesystem> detection was an issue on ICPC but might have been fixed by now?~~
Edit: Remove. Works (almost; see below) without.

falbrechtskirchinger · 2022-05-01T09:22:43Z

Intel needs to be excluded here as well. :-(

#if JSON_HAS_FILESYSTEM || JSON_HAS_EXPERIMENTAL_FILESYSTEM
    // JSON_HAS_CPP_17 (do not remove; see note at top of file)
    SECTION("issue #3070 - Version 3.10.3 breaks backward-compatibility with 3.10.2 ")
    {
        nlohmann::detail::std_fs::path text_path("/tmp/text.txt");
        json j(text_path);

        const auto j_path = j.get<nlohmann::detail::std_fs::path>();
        CHECK(j_path == text_path);

#if !defined(_MSC_VER) && !(defined(__GNUC__) && __GNUC__ == 8 && __GNUC_MINOR__ < 4)
        // works everywhere but on MSVC and GCC <8.4
        CHECK_THROWS_WITH_AS(nlohmann::detail::std_fs::path(json(1)), "[json.exception.type_error.302] type must be string, but is number", json::type_error);
#endif
    }
#endif

Maybe this condition should just be changed to:

#if defined(__clang__) || ((defined(__GNUC__) && !defined(__INTEL_COMPILER)) && (__GNUC__ > 8 || (__GNUC__ == 8 && __GNUC_MINOR__ >= 4)))
        // only known to work on Clang and GCC >=8.4
        CHECK_THROWS_WITH_AS(nlohmann::detail::std_fs::path(json(1)), "[json.exception.type_error.302] type must be string, but is number", json::type_error);
#endif

Edit: We don't have JSON_HEDLEY_GNUC_VERSION_CHECK in the test suite. Fixed.
Edit 2: Condition was still not correct. Hopefully now.

falbrechtskirchinger · 2022-05-01T11:17:00Z

Another issue:

    SECTION("issue #1647 - compile error when deserializing enum if both non-default from_json and non-member operator== exists for other type")
    {
#if !(defined(__INTEL_COMPILER) && __cplusplus >= 202000)
        {
            json j;
            NonDefaultFromJsonStruct x(j);
            NonDefaultFromJsonStruct y;
            CHECK(x == y);
        }
#endif

        auto val = nlohmann::json("one").get<for_1647>();
        CHECK(val == for_1647::one);
        json j = val;
    }

Test fails to compile on ICPC without the preprocessor guard when targeting C++20.

/__w/json/json/tests/src/unit-regression2.cpp(517): error: too many initializer values
              NonDefaultFromJsonStruct x(j);

These lines compile fine:

    auto x2 = j.get<NonDefaultFromJsonStruct>();
    auto x3 = j.operator NonDefaultFromJsonStruct();

Edit: First try #if condition didn't work.

falbrechtskirchinger · 2022-05-01T11:58:54Z

@nlohmann I'm still waiting for CI, so I don't know if anything else is needed. In any event, there are now a quite a few changes. You can either reference my branch here or I can open a PR, whichever you prefer.

nlohmann · 2022-05-01T11:59:57Z

@nlohmann I'm still waiting for CI, so I don't know if anything else is needed. In any event, there are now a quite a few changes. You can either reference my branch here or I can open a PR, whichever you prefer.

Please open a PR. You're already much deeper in this topic.

falbrechtskirchinger · 2022-05-01T12:02:05Z

Okay. Of course CI just failed so there's at least one more issue to solve...

nlohmann added 3 commits December 29, 2021 21:29

⬆️ Doctest 2.4.7

f65ade0

👷 add CI step for ICPC

ab557ae

👷 add CI step for ICPC

15eaae2

nlohmann marked this pull request as draft December 29, 2021 21:55

nlohmann added 6 commits December 30, 2021 09:19

👷 add CI step for ICPC

7b60a2a

⬇️ downgrade to Doctest 2.4.6

604e449

👷 add CI step for ICPC

e40843f

👷 add CI step for ICPC

3822537

👷 add CI step for ICPC

42acea3

👷 add CI step for ICPC

0152bac

nlohmann added 2 commits December 30, 2021 13:41

Merge branch 'develop' of https://github.com/nlohmann/json into icpc

8cbe3ed

� Conflicts: � .github/workflows/ubuntu.yml � cmake/ci.cmake

👷 add CI step for ICPC

23a7a2d

nlohmann added the state: help needed the issue needs help to proceed label Dec 30, 2021

nlohmann mentioned this pull request Dec 30, 2021

Help needed #3112

Open

Merge branch 'develop' of https://github.com/nlohmann/json into icpc

3a4da7c

falbrechtskirchinger mentioned this pull request Apr 5, 2022

Restore disabled check for #3070 (except on MSVC) #3421

Merged

nlohmann added 3 commits April 29, 2022 21:46

Merge branch 'develop' of https://github.com/nlohmann/json into icpc

3e767c7

Merge branch 'develop' of https://github.com/nlohmann/json into icpc

e1190d6

🔇 suppress warning #2196: routine is both "inline" and "noinline"

f9dfd63

falbrechtskirchinger mentioned this pull request May 1, 2022

Fix C++20/gcc-12 issues (Part 2) #3446

Merged

9 tasks

falbrechtskirchinger reviewed May 1, 2022

View reviewed changes

falbrechtskirchinger mentioned this pull request May 1, 2022

Add build step for ICPC (with fixes) #3465

Merged

nlohmann closed this May 1, 2022

nlohmann deleted the icpc branch June 18, 2022 19:08

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add build step for ICPC #3229

Add build step for ICPC #3229

nlohmann commented Dec 29, 2021

coveralls commented Dec 30, 2021 •

edited

Loading

nlohmann commented Dec 30, 2021

falbrechtskirchinger commented Mar 7, 2022

nlohmann commented Mar 8, 2022

falbrechtskirchinger commented Apr 6, 2022

falbrechtskirchinger May 1, 2022

falbrechtskirchinger May 1, 2022 •

edited

Loading

falbrechtskirchinger May 1, 2022

falbrechtskirchinger May 1, 2022 •

edited

Loading

falbrechtskirchinger commented May 1, 2022 •

edited

Loading

falbrechtskirchinger commented May 1, 2022 •

edited

Loading

falbrechtskirchinger commented May 1, 2022

nlohmann commented May 1, 2022

falbrechtskirchinger commented May 1, 2022

Add build step for ICPC #3229

Add build step for ICPC #3229

Conversation

nlohmann commented Dec 29, 2021

coveralls commented Dec 30, 2021 • edited Loading

nlohmann commented Dec 30, 2021

falbrechtskirchinger commented Mar 7, 2022

nlohmann commented Mar 8, 2022

falbrechtskirchinger commented Apr 6, 2022

falbrechtskirchinger May 1, 2022

Choose a reason for hiding this comment

falbrechtskirchinger May 1, 2022 • edited Loading

Choose a reason for hiding this comment

falbrechtskirchinger May 1, 2022

Choose a reason for hiding this comment

falbrechtskirchinger May 1, 2022 • edited Loading

Choose a reason for hiding this comment

falbrechtskirchinger commented May 1, 2022 • edited Loading

falbrechtskirchinger commented May 1, 2022 • edited Loading

falbrechtskirchinger commented May 1, 2022

nlohmann commented May 1, 2022

falbrechtskirchinger commented May 1, 2022

coveralls commented Dec 30, 2021 •

edited

Loading

falbrechtskirchinger May 1, 2022 •

edited

Loading

falbrechtskirchinger May 1, 2022 •

edited

Loading

falbrechtskirchinger commented May 1, 2022 •

edited

Loading

falbrechtskirchinger commented May 1, 2022 •

edited

Loading