Re: c/10826: [3.3/3.4 regression] Anonymous union regression

Re: c/10826: [3.3/3.4 regression] Anonymous union regression

[Janis Johnson]

The test case starts getting an error with this patch:

> 2002-06-16  Richard Henderson  <rth@redhat.com>
>
>       * c-common.c (flag_ms_extensions): Move from c++ front end.
>       * c-common.h (flag_ms_extensions): Declare.
>       * c-decl.c (c_decode_option): Add -fms-extensions.
>       (grokfield): Don't accept anonymous structures in ISO C mode;
>       accept only unnamed anonymous structures in GNU C mode; accept
>       Plan 9 extensions in MS mode.
>       * c-parse.in (SAVE_EXT_FLAGS, RESTORE_EXT_FLAGS): Rename from
>       SAVE/RESTORE_WARN_FLAGS; add flag_iso frobbing; update all
callers.
>       (extension): Clear flag_iso.
>       * doc/invoke.texi (C Dialect Options): Add -fms-extensions.
>
--- gcc/gcc/cp/ChangeLog ---
> 2002-06-16  Richard Henderson  <rth@redhat.com>
>
>       * cp-tree.h, decl2.c (flag_ms_extensions): Move to c-common.

It doesn't seem to be a regression after all, but a fix.

Janis

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10826

Re: c/10826: [3.3/3.4 regression] Anonymous union regression

[Richard Henderson]

On Tue, May 20, 2003 at 06:35:22PM -0700, Janis Johnson wrote:
> It doesn't seem to be a regression after all, but a fix.

Indeed it is.  The sample code for this pr is not valid C++,
and the GCC C extension mirrors the C++ feature.


r~

Re: c/10826: [3.3/3.4 regression] Anonymous union regression

[bangerth]

Synopsis: [3.3/3.4 regression] Anonymous union regression

State-Changed-From-To: analyzed->closed
State-Changed-By: bangerth
State-Changed-When: Wed May 21 13:56:20 2003
State-Changed-Why:
    Not a bug, but previously accepted erroneously. I now also
    see what rth meant: if you write "union type {...}" you
    really _declare_ a nested type. That explains why the
    compiler says that there is no such member. Only if you
    omit the type name, is this the declaration of an anonymous
    union, which not only declares a type, but also a "member
    variable" (in C++ speak).
    
    W.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10826

Re: c/10826: [3.3/3.4 regression] Anonymous union regression

[Bart Oldeman]

On 21 May 2003 bangerth@dealii.org wrote:

> Synopsis: [3.3/3.4 regression] Anonymous union regression
> 
> State-Changed-From-To: analyzed->closed
> State-Changed-By: bangerth
> State-Changed-When: Wed May 21 13:56:20 2003
> State-Changed-Why:
>     Not a bug, but previously accepted erroneously. I now also
>     see what rth meant: if you write "union type {...}" you
>     really _declare_ a nested type. That explains why the
>     compiler says that there is no such member. Only if you
>     omit the type name, is this the declaration of an anonymous
>     union, which not only declares a type, but also a "member
>     variable" (in C++ speak).
>     
>     W.
> 
> http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10826

Ok, in that case -fms-extension does the job. Are "after the
fact" changes to gcc/NEWS accepted?

Bart

Re: c/10826: [3.3/3.4 regression] Anonymous union regression

[bangerth]

Old Synopsis: Anonymous union regression between gcc 3.2 and 3.3
New Synopsis: [3.3/3.4 regression] Anonymous union regression

State-Changed-From-To: open->analyzed
State-Changed-By: bangerth
State-Changed-When: Fri May 16 20:39:51 2003
State-Changed-Why:
    Confirmed. It worked up to 3.2, but no longer in 3.3 and 3.4.
    It works again if "union type {..." is replaced with just
    "union { ...".

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10826