Yes, this change was done intentionally to be compatible with their
behavior:

Microsoft (R) C/C++ Optimizing Compiler Version 19.00.22013.1 for x64
Copyright (C) Microsoft Corporation.  All rights reserved.

anon.c
anon.c(2): error C2027: use of undefined type 'undefined'
anon.c(2): note: see declaration of 'undefined'



On Thu, Sep 18, 2014 at 9:38 AM, jahanian <[email protected]> wrote:

>
> On Sep 18, 2014, at 9:33 AM, jahanian <[email protected]> wrote:
>
>
> On Sep 17, 2014, at 5:42 PM, David Majnemer <[email protected]>
> wrote:
>
> Author: majnemer
> Date: Wed Sep 17 19:42:05 2014
> New Revision: 218006
>
> URL: http://llvm.org/viewvc/llvm-project?rev=218006&view=rev
> Log:
> Sema: Diagnose undefined structs used as Microsoft anonymous structs
>
> Previously, we would not mark structs containing anonymous structs as
> invalid.  Later, horrific things would occur when trying to determine
> the size of the parent record.
>
> Instead, require the struct to be a complete type when used as an
> anonymous struct.  Mark both the anonymous field for the struct and the
> parent context as invalid (this is similar to what we do when a struct
> contains a field with an incomplete type.)
>
>
>
> Thank you David for working on this. Will this test case continue to work
> without error. This was the behavior before:
>
>
> struct anon_fault {
> struct undefined;
> };
>
> struct undefined {
>   int ii;
> };
>
> int main()
> {
>   return sizeof(struct anon_fault);
> }
>
>
> We get the error with your patch:
>
> *t.c:2:2: **error: **field has incomplete type 'struct undefined'*
> *t.c:2:9: note: *forward declaration of 'struct undefined'
>         struct undefined;
> *               ^*
> *I am not familiar with details of microsoft’s anonymous structs. Is this
> the expected behavior?*
>
> *- Thanks, Fariborz*
>
>
>
>
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to