[PATCH] D12747: Implement [depr.c.headers]

[Richard Smith via cfe-commits]

On Tue, Oct 6, 2015 at 4:16 PM, Sean Silva <chisophugis at gmail.com> wrote:

> On Tue, Oct 6, 2015 at 4:13 PM, Richard Smith <richard at metafoo.co.uk>
> wrote:
>
>> On Tue, Oct 6, 2015 at 4:11 PM, Sean Silva <chisophugis at gmail.com> wrote:
>>
>>> +extern "C++" {
>>> +#include <__nullptr>
>>> +using std::nullptr_t;
>>> +}
>>>
>>> Does this even compile with modules?
>>>
>>
>> Yes. You're allowed to import a C++ module within an 'extern "C++"' block.
>>
>
> Why do we even need `extern "C++"` if we are already under `#ifdef
> __cplusplus`?
>

People like to include C++'s <foo.h> headers inside 'extern "C"' blocks. We
shouldn't break that if we don't have to.


> -- Sean Silva
>
>
>>
>>
>>> On Tue, Oct 6, 2015 at 3:57 PM, Richard Smith via cfe-commits <
>>> cfe-commits at lists.llvm.org> wrote:
>>>
>>>> <stddef.h>. This one is tricky:
>>>>
>>>> 1) There's an (undocumented) interface between the C standard library
>>>> and this header, where the macros __need_ptrdiff_t, __need_size_t,
>>>> __need_wchar_t, __need_NULL, __need_wint_t request just a piece of this
>>>> header rather than the whole thing. If we see any of those, just go
>>>> straight to the underlying header.
>>>>
>>>> 2) We probably don't want <stddef.h> to include <cstddef> (for
>>>> consistency with other headers), but <stddef.h> must provide a ::nullptr_t
>>>> (which we don't want <cstddef> to provide). So neither header includes the
>>>> other. Instead, both include <__nullptr> for std::nullptr_t, and we
>>>> duplicate the definition of max_align_t between them, in the case where the
>>>> compiler's <stddef.h> doesn't provide it.
>>>>
>>>> If you prefer, I could make <stddef.h> include <cstddef> to avoid the
>>>> duplication of the max_align_t logic.
>>>>
>>>> This patch depends on patch 02.
>>>>
>>>> On Tue, Oct 6, 2015 at 3:42 PM, Richard Smith <richard at metafoo.co.uk>
>>>> wrote:
>>>>
>>>>> <setjmp.h>, an easy one. We guarantee a setjmp macro exists even if
>>>>> this header is somehow included from C (the C standard allows that, so it's
>>>>> not worth checking for __cplusplus).
>>>>>
>>>>> On Tue, Oct 6, 2015 at 3:36 PM, Richard Smith <richard at metafoo.co.uk>
>>>>> wrote:
>>>>>
>>>>>> Split <math.h> out of <cmath>. This is a big change, but the same
>>>>>> pattern as the prior ones.
>>>>>>
>>>>>> On Tue, Oct 6, 2015 at 3:28 PM, Richard Smith <richard at metafoo.co.uk>
>>>>>> wrote:
>>>>>>
>>>>>>> Likewise for <errno.h>, <float.h>, <inttypes.h>
>>>>>>>
>>>>>>> On Tue, Oct 6, 2015 at 3:20 PM, Richard Smith <richard at metafoo.co.uk
>>>>>>> > wrote:
>>>>>>>
>>>>>>>> Split <ctype.h> header out of <cctype>
>>>>>>>>
>>>>>>>> On Tue, Oct 6, 2015 at 3:07 PM, Richard Smith <
>>>>>>>> richard at metafoo.co.uk> wrote:
>>>>>>>>
>>>>>>>>> Next: factoring the definition of std::nullptr_t out into a
>>>>>>>>> separate file, so that <cstddef> and <stddef.h> can both use it, without
>>>>>>>>> <stddef.h> including <cstddef> and without <cstddef> providing a
>>>>>>>>> ::nullptr_t like <stddef.h> does.
>>>>>>>>>
>>>>>>>>> On Tue, Oct 6, 2015 at 3:02 PM, Eric Fiselier <eric at efcs.ca>
>>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>>> LGTM.
>>>>>>>>>>
>>>>>>>>>> On Tue, Oct 6, 2015 at 3:58 PM, Richard Smith <
>>>>>>>>>> richard at metafoo.co.uk> wrote:
>>>>>>>>>> > On Mon, Oct 5, 2015 at 7:10 PM, Eric Fiselier <eric at efcs.ca>
>>>>>>>>>> wrote:
>>>>>>>>>> >>
>>>>>>>>>> >> EricWF added a comment.
>>>>>>>>>> >>
>>>>>>>>>> >> I think thing change will help us close a number out
>>>>>>>>>> outstanding bugs. I
>>>>>>>>>> >> don't have any fundamental objections to this approach.
>>>>>>>>>> However the size of
>>>>>>>>>> >> this patch scares me.  I understand the changes are mostly
>>>>>>>>>> mechanical but
>>>>>>>>>> >> their size can hide things. For example has anybody noticed
>>>>>>>>>> that your
>>>>>>>>>> >> internal changes to `<deque>` are in this patch? It would be
>>>>>>>>>> nice to break
>>>>>>>>>> >> this down into more digestible pieces that could be quickly
>>>>>>>>>> spot checked.
>>>>>>>>>> >
>>>>>>>>>> >
>>>>>>>>>> > OK. First such patch is attached. It just removes the
>>>>>>>>>> macro-capturing
>>>>>>>>>> > wrapper functions.
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> cfe-commits mailing list
>>>> cfe-commits at lists.llvm.org
>>>> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20151006/fe07b610/attachment.html>