From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 17690 invoked by alias); 28 Feb 2008 01:01:25 -0000 Received: (qmail 17584 invoked by uid 22791); 28 Feb 2008 01:01:24 -0000 X-Spam-Check-By: sourceware.org Received: from mail.codesourcery.com (HELO mail.codesourcery.com) (65.74.133.4) by sourceware.org (qpsmtpd/0.31) with ESMTP; Thu, 28 Feb 2008 01:01:07 +0000 Received: (qmail 12373 invoked from network); 28 Feb 2008 01:01:05 -0000 Received: from unknown (HELO digraph.polyomino.org.uk) (joseph@127.0.0.2) by mail.codesourcery.com with ESMTPA; 28 Feb 2008 01:01:05 -0000 Received: from jsm28 (helo=localhost) by digraph.polyomino.org.uk with local-esmtp (Exim 4.68) (envelope-from ) id 1JUX8y-0006WA-9h; Thu, 28 Feb 2008 01:01:04 +0000 Date: Thu, 28 Feb 2008 07:12:00 -0000 From: "Joseph S. Myers" To: Mark Mitchell cc: Danny Smith , gcc-patches@gcc.gnu.org Subject: Re: Reject dllimport on scalar typedefs (bug 33963) In-Reply-To: <47C592C9.2000906@codesourcery.com> Message-ID: References: <47C48734.1030806@codesourcery.com> <9c03c2dd0802262240s5629426eo4cda4d4a2e3c77bb@mail.gmail.com> <47C592C9.2000906@codesourcery.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org X-SW-Source: 2008-02/txt/msg01392.txt.bz2 On Wed, 27 Feb 2008, Mark Mitchell wrote: > Joseph S. Myers wrote: > > > > Why do we allow dllimport/dllexport for typedefs of structs and > > > unions? That usage is not documented in MSDN > > > http://msdn2.microsoft.com/en-us/library/3y1sfaz2%28VS.71%29.aspx > > > > I presume TYPE_DECL is arising for some C++ cases other than explicit > > typedefs, relating to visibility for classes (otherwise Mark wouldn't have > > needed to make the TYPE_DECL change in the first place). > > See, for example, the g++.dg/ext/visibility/visibility-9.C, which was one of > the tests checked in with the patch that allowed these attributes on > TYPE_DECLs. The issue is probably not the dllimport per se, but, rather > __attribute__((visibility(...))) -- and variants of that spelling, like > __declspec(notshared). So is this patch OK or not? As far as I can tell it fixes the testsuite regression - but there may be other cases that have also regressed, not covered by the testsuite, that would best have separate bugs filed and marked as regressions. -- Joseph S. Myers joseph@codesourcery.com