[ADD] Add test-results for apple-clang-14

[filipsajdak]

As suggested (here #133 (comment)) I have created apple-clang-14 in test-results that compiles regression tests with:

Apple clang version 14.0.0 (clang-1400.0.29.202)
Target: arm64-apple-darwin22.1.0
Thread model: posix
InstalledDir: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin

The main issue is that it does not support <ranges> - that means that there are no correct results for:

• mixed-captures-in-expressions-and-postconditions.cpp2
• mixed-function-expression-and-std-ranges-for-each-with-capture.cpp2
• mixed-function-expression-and-std-ranges-for-each.cpp2
• mixed-function-expression-with-pointer-capture.cpp2
• mixed-function-expression-with-repeated-capture.cpp2

I have added failed .output files for them (with error):

mixed-function-expression-and-std-ranges-for-each-with-capture.cpp2:14:18: error: no member named 'for_each' in namespace 'std::ranges'
    std::ranges::for_each
    ~~~~~~~~~~~~~^
mixed-function-expression-and-std-ranges-for-each-with-capture.cpp2:18:18: error: no member named 'for_each' in namespace 'std::ranges'
    std::ranges::for_each( view, std::move(callback));
    ~~~~~~~~~~~~~^
2 errors generated.

[hsutter]

Thanks!

No <ranges>, then? Interesting. Looking at the cppreference C++20 conformance table, I see Apple Clang is also missing <=> (spaceship), jthread, and of course <format> (the last one doesn't surprise me but the others do).

Of these, the one that concerns me most is <=>, because I was planning to soon take a hard dependency on it in emitted Cpp1 code as I add support for user-defined types... and not having that is going to be interesting because I don't want to have to emit all that pre-C++20 comparison boilerplate just to the emitted Cpp1 source works on Apple Clang. :-( The three-way comparison <=> is the first Cpp2-derived feature to be standardized in ISO C++, and I was expecting to be able to rely on its being there to generate code. Well, I know actual Clang's support for <=> is "partial" but it seemed to work as expected the last time I happened to try it in passing... maybe Apple Clang will add it in the next release.

[filipsajdak]

You already use operator<=> in cppfront in common.h file:

struct source_position
{
    lineno_t    lineno;     // one-based offset into program source
    colno_t     colno;      // one-based offset into line

    source_position(lineno_t l = 1, colno_t  c = 1 )
        : lineno{ l }, colno{ c }
    {
    }

    auto operator<=>(source_position const&) const = default;

    auto to_string() const -> std::string
    {
        return "(" + std::to_string(lineno) + "," + std::to_string(colno) + ")";
    }
};

It compiles and works... does it means it is working?

[hsutter]

Right, that's in cppfront's own source. So if Apple Clang doesn't support <=> then presumably it won't be able to compile cppfront itself. If you try to compile cppfront.cpp I would expect it would break if it doesn't understand that common.h spaceship operator.

But someone using an already-compiled cppfront executable to compile some Cpp2 code to Cpp1 and using Apple Clang as the second (Cpp1) compiler will be fine, because executing cppfront doesn't currently generate <=> in the Cpp1 output (including that there's no <=> in cpp2util.h that the Cpp1 output will include).

[filipsajdak]

Taking a look at https://en.cppreference.com/w/cpp/compiler_support and I see that P1185R2 is supported by apple clang 12 (I am currently using 14).

[hsutter]

Strange that it would have P1185 supported but P0515 not. Oh well, maybe the table is not consistent or up to date -- if you try to compile cppfront.cpp with Apple Clang that should tell you pretty quickly whether it supports <=>. :)

[filipsajdak]

This is my main compiler, and I have been compiling cppfront with Apple clang since my first contribution to cppfront.

That is why I was surprised to see that it is not supported.

[filipsajdak]

That means apple clang has support for operator <=> and you don't have to change your plans.

[hsutter]

Thanks. I'll ping Sergey and let him know he might want to update cppreference.