[cfe-commits] [PATCH] #import directive in MS compatibility mode

Aaron Ballman aaron at aaronballman.com
Sat Mar 17 17:11:40 PDT 2012

On Sat, Mar 17, 2012 at 7:03 PM, Douglas Gregor <dgregor at apple.com> wrote:
> On Mar 17, 2012, at 5:00 PM, Aaron Ballman wrote:
>> On Thu, Mar 15, 2012 at 7:29 AM, Aaron Ballman <aaron at aaronballman.com> wrote:
>>> On Wed, Mar 14, 2012 at 11:25 PM, J B <jayblanchard.cpp at gmail.com> wrote:
>>>> The reason I recommend an error is because the code will error out later
>>>> without the .tlh (where it is requested via attributes or absence of).  The
>>>> errors later will be a direct result of the failure of #import to function
>>>> properly.
>>> That's a valid reason to error in my book.  I'll go down that route.
>>> Thanks for the insights!
>> This is the revised patch based on feedback from everyone.  It causes
>> a (shorter) error to happen when using #import in MS mode, and does
>> not perform the actual include.  It also eats the optional attributes
>> so that parsing can continue.
>> Again, this will fix PR 10727
> +def err_pp_import_directive_ms : Error<
> +  "#import does not convert a type library to C++ classes in Microsoft Mode">,
> +  InGroup<Microsoft>;
> This should really just say that this Microsoft feature is not implemented.
> -  if (!LangOpts.ObjC1)  // #import is standard for ObjC.
> +  if (LangOpts.MicrosoftMode)
> +    return HandleMicrosoftImportDirective(ImportTok);
> +  else if (!LangOpts.ObjC1)  // #import is standard for ObjC.
>     Diag(ImportTok, diag::ext_pp_import_directive);
> I think we should do this the other way: if LangOpts.ObjC1, we treat it as an Objective-C #import. Otherwise, we either error (if in Microsoft mode) or give an extension warning (not in Microsoft mode).

Good points -- rectified in this patch.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: MicrosoftImport.patch
Type: application/octet-stream
Size: 4117 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20120317/d7c9f74e/attachment.obj>

More information about the cfe-commits mailing list