From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 18952 invoked by alias); 14 Sep 2004 02:20:05 -0000 Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org Received: (qmail 18881 invoked by uid 48); 14 Sep 2004 02:20:04 -0000 Date: Tue, 14 Sep 2004 02:20:00 -0000 Message-ID: <20040914022004.18880.qmail@sourceware.org> From: "austern at apple dot com" To: gcc-bugs@gcc.gnu.org In-Reply-To: <20040913213750.17470.austern@apple.com> References: <20040913213750.17470.austern@apple.com> Reply-To: gcc-bugzilla@gcc.gnu.org Subject: [Bug c++/17470] Visibility attribute ignored for explicit template instantiation X-Bugzilla-Reason: CC X-SW-Source: 2004-09/txt/msg01480.txt.bz2 List-Id: ------- Additional Comments From austern at apple dot com 2004-09-14 02:20 ------- Tastes differ, I suppose! My own feeling is that it's unnatural, when declaring an unbounded set of functions, to have to say that either all of them or none of them get exported. I find it more likely that a dynamic library author would decide that a selected few specializations are the library's interface. However, I don't insist on that point. What I do insist on: it's wrong for the compiler to silently ignore the visibility attribute on explicit instantiations. Either it should be honored, or else it should be documented and diagnosed as an error. Silently ignoring it is a nasty trap for users. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17470