issue 19949 - C++ Mangling: add support for abi-tags from the Itanium ABI

[SSoulaimane]

Special UDA for mangling extern(C++) symbols.
Required by dlang/dmd#9995 to fix issue 19949.

[dlang-bot]

Thanks for your pull request, @SSoulaimane!

Bugzilla references

Auto-close	Bugzilla	Severity	Description
✗	19949	normal	C++ Mangling: no support for abi-tags from the Itanium ABI

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub run digger -- build "master + druntime#2639"

[wilzbach]

Name needs to be in D Style

[jacob-carlborg]

Put all the documentation in a version (CoreDdoc) block along with a copy of the struct declaration. I think it's enough with the constructor (without a body) in the copy. Otherwise the documentation will only be available if generated on a Posix system. That's currently a problem with the above selector attribute. Have a look here, it's empty: https://dlang.org/phobos/core_attribute.html.

[SSoulaimane]

How about something like this:

version (Posix)
    version = UdaGNUAbiTag;

version (D_ObjectiveC)
    version = UdaSelector;

version (CoreDdoc)
{
    version = UdaGNUAbiTag;
    version = UdaSelector;
}

[jacob-carlborg]

How about something like this:

Hmm, that might work. I was actually only referring to the documentation you added but if you want to fix the selector UDA as well, that would be great.

[Geod24]

Tags on an inline namespace are not represented in the mangled name of the namespace, but they are subject to the above tag propagation

https://itanium-cxx-abi.github.io/cxx-abi/abi.html#mangle.abi-tag

The attribute can also be applied to an inline namespace, but does not affect the mangled name of the namespace; in this case it is only used for -Wabi-tag warnings and automatic tagging of functions and variables. Tagging inline namespaces is generally preferable to tagging individual declarations, but the latter is sometimes necessary, such as when only certain members of a class need to be tagged.

https://gcc.gnu.org/onlinedocs/gcc/C_002b_002b-Attributes.html

So it should be inherited by the member. But that's a C++ side thing, we should represent this with @gnuAbiTag("XXX") { /* decls */ } not by trying to mimic the C++ code.
I would just remove this sentence.

[SSoulaimane]

I don't understand what you mean.

That sentence is about the current DMD behavior of inheriting UDAs from a namespace.
for example:

@("abc")
extern (C++, "namespace")
{
    void fun();
}

fun has the attribute "abc".

while it is not the same in the following case:

@("abc")
extern (C++)
class C
{
    void fun();
}

fun doesn't have the attribute "abc".

[SSoulaimane]

This special UDA is not supposed to be inherited from a namespace.

[Geod24]

Yes and that makes sense. But what your sentence reads to me is that in this case:

@gnuAbiTag("Test")
extern (C++, "Test") {
    void fun();
}

fun would not have the funAbiTag("Test") attribute.

[SSoulaimane]

Yes that's what I meant, It is handled specially by the compiler that's why it's documented.

Doc version added

[Geod24]

There were several issues raised with this.

EDIT: Ah, it's actually the Druntime part, the issues were raised on the DMD PR.
Nevertheless, I am finalizing a different version of the DMD PR at the moment, as I explicitly mentioned to you.

[thewilsonator]

Superseded by #2990