[cfe-dev] Extra #defines for Windows SDK 6.0a/VS2008

[Francois Pichet]

Based on my investigation so far there are probably dozen of small
patches needed to support all MS extensions but the major problem is
dealing with template. The Microsoft compiler seems to postpone all
template type checking and name resolution to instantiation time.
CLang is far more strict. For proper MS compatibility related to
template, CLang will probably need something similar to GCC's
-fpermissive mode.


On Fri, Aug 6, 2010 at 12:01 PM, Jesse Towner <townerj at gmail.com> wrote:
> On Fri, 2010-08-06 at 17:05 +0200, Per Lindén wrote:
>> The sad thing is that socket and thread support needed by libc++
>> indirectly cause inclusion of <windows.h>, at least as far as I can see
>> right now. Might be worked around by creating custom declarations of the
>> stuff needed but it usually spans 50 files... Also, all C99 standard lib
>> stuff must be replicated somehow. Et cetera... (does anyone feel
>> responsible for http://wiki.llvm.org/WinC99CLib:Windows_C99_clib ?)
>
> What part of the C++0x Standard Library or libc++ uses sockets? The
> Windows API calls for the threads library, and perhaps
> std::chrono::monotonic_clock and a few other classes, that you might
> want included in header files so that it can be inlined to reduce call
> overhead can probably be isolated to a hundred lines of code or less.
> And then for the rest for the stuff that is best placed in source files,
> including Windows.h isn't an issue.
>
> Many of the C99 functions that are a part of C++0x will probably end up
> in the next version of MSVC++'s CRT as it begins to implement more of C
> ++0x (cstdint is already in MSVC++2010), but it's probably still going
> to be missing a lot of C99 functionality that will be needed by Clang in
> C99 mode (such as no stdint.h or tgmath.h for example, just cstdint and
> ctgmath). Writing the C99 math functions isn't too difficult, a lot of
> it is pretty easy, and the transcendental functions just use various
> numerical methods or identities with other existing functions. But
> getting it to be fully standards compliant would require some rigor and
> thorough testing. Plus providing optimized versions for say x86/x86-64
> SSE/SSE2+ targets would be more involved. Isn't this stuff already apart
> of the MinGW/BSD implementations that the WinC99 library was going to be
> using?
>
>> There are of course other problems than MS extensions in the SDK
>> headers, such as template abuse, redefinition of "__null" etc. The
>> extension stuff implemented in clang so far works fairly well, with the
>> exception of some "throw(...)" exception specifications I am
>> investigating. In short <windows.h> really sucks, but finding
>> non-trivial Windows code that does not pull it in is hard. The question
>> is how to solve it without becoming incompatible, violating copyrights
>> etc. I volunteer to test and implement different solutions whenever I
>> get the time.
>>
>> Thank you all for engaging in such a hopeless platform.
>
> Yeah, another thing is implementing all of the COM/COM+ extensions too.
> Does Clang implement any of this yet?
>
> - Jesse Towner
>
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
>