[llvm-bugs] [Bug 42100] New: Implementing COM interfaces in MinGW broken by "Add Attribute NoThrow as an Exception Specifier Type"

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=42100

            Bug ID: 42100
           Summary: Implementing COM interfaces in MinGW broken by "Add
                    Attribute NoThrow as an Exception Specifier Type"
           Product: clang
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: -New Bugs
          Assignee: unassignedclangbugs at nondot.org
          Reporter: martin at martin.st
                CC: erich.keane at intel.com, htmldeveloper at gmail.com,
                    llvm-bugs at lists.llvm.org, neeilans at live.com,
                    richard-llvm at metafoo.co.uk, rnk at google.com

The recent change "Add Attribute NoThrow as an Exception Specifier Type"
(https://reviews.llvm.org/D62435, SVN r362119) broke code implementing COM
interfaces for the MinGW target.

Such code often ends up looking like this:

    class Base { 
    public:
        virtual __declspec(nothrow) void foo() = 0;
    };
    class Sub : public Base {
    public:
        void foo() { }
    };

Before this change, this never even warned (and doesn't warn with GCC either,
nor with MSVC). After this change, this is suddenly a hard error without a way
to disable the error.

One could of course fix the code to add the same attribute at the
implementation of the interface as well, but this seems to be the common
pattern of how one implements a COM interface, so I would expect there to be a
lot of code out there looking exactly like this.

By adding -fms-extensions, it is turned into a warning instead of an error
though, thanks to the following bit in lib/Sema/SemaExceptionSpec.cpp in
Sema::CheckOverridingFunctionExceptionSpec:

  unsigned DiagID = diag::err_override_exception_spec;
  if (getLangOpts().MicrosoftExt)
    DiagID = diag::ext_override_exception_spec;


Now for the MinGW case, the ideal would be to only degrade this to a warning
for cases with `__declspec(nothrow)`, not for the proper C++11 noexcept
specifier.


Do you (Reid and Erich) have any opinions on this? Is it better to change "Add
Attribute NoThrow as an Exception Specifier Type" to not do it for the MinGW
case at all (which would match GCC), or to degrade all exception spec
mismatches to warnings for this target?

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20190601/4a5bdd0c/attachment.html>